Towards intelligent and integrated architecture for hydrogen fuel cell system: challenges and approaches

Xiaoyu Guo; Zhen Dong; Jiabin Shen; Yiqiao Xu; Qiaohui He; Xiaowei Zhao; Zhengtao Ding

doi:10.1360/nso/20220038

All issues

Volume 2 / No 4 (2023)

Natl Sci Open, 2 4 (2023) 20220038

Full HTML

Open Access

Review

Issue		Natl Sci Open Volume 2, Number 4, 2023


Article Number		20220038
Number of page(s)		26
Section		Information Sciences
DOI		https://doi.org/10.1360/nso/20220038
Published online		06 February 2023

National Science Open 2: 20220038, 2023

REVIEW

Towards intelligent and integrated architecture for hydrogen fuel cell system: challenges and approaches

Xiaoyu Guo¹, Zhen Dong³, Jiabin Shen², Yiqiao Xu¹, Qiaohui He³, Xiaowei Zhao³ and Zhengtao Ding¹^*

¹ Department of Electrical and Electronic Engineering, University of Manchester, Manchester M13 9PL, UK
² General Motors Canada Company, Oshawa L1J 0C5, Canada
³ School of Engineering, University of Warwick, Coventry CV4 7AL, UK

^* Corresponding author (email: zhengtao.ding@manchester.ac.uk)

Received: 6 April 2022
Revised: 26 September 2022
Accepted: 29 September 2022

Abstract

The hydrogen fuel cell is rapidly attracting research interest for its potential in power generation and electrified transportation. The fuel cell stack (FCS) is a complex system comprising multiple coupled subsystems, and in order to maximize the utilization of an FCS, the system-level design and control can be optimized through modeling, data-based analytics and monitoring. To this end, a systematic overview of the system architecture and control of hydrogen fuel cells is provided in this review, with focus on integration and intelligence. Firstly, the fuel cell subsystems, namely the cathode, anode and cooling loops are reviewed, where their respective control methods and impact on FCS performance are discussed. DC/DC converters are another core component of FCS, and we present an overview of fuel cell DC/DC converter topologies and integrated control of DC/DC converter and air compressor. Finally, the system-level integration of fuel cells in power systems is surveyed. In the conclusions, we discuss the challenges and perspectives concerning the integrated architecture and intelligent control for FCS, including cohesive dynamic models, data-based approaches, and integrated hardware architecture.

Key words: polymer electrolyte membrane fuel cell / system integration / control strategy / DC/DC converter

© The Author(s) 2023. Published by China Science Publishing & Media Ltd. and EDP Sciences.

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Introduction

Targeting a carbon neutral future, which the United Nations sets to achieve by 2050, renewable energy is taking up large market shares in stationary power generation and electrified transportation sectors [1]. It is a consensus in society to endeavor towards achieving the carbon neutral goal with a renewable energy-based closed loop, where hydrogen can play a significant role as a media to connect energy generation and utilization as shown in Figure 1. In the proposed renewable energy ecosystem comprising fuel cells and large-scale wind and solar power generators, the electricity generated can be used for feeding to the grid, while the redundant power can be adopted to an electrolyzer for generating hydrogen to compensate for the unpredictability of wind and solar energy. The hydrogen from renewable energy sources, referred to as the green hydrogen, is supplied to fuel cells for stationary and transportation applications.

Figure 1

Hydrogen-based energy ecosystem.

The significance of hydrogen has attracted much attention globally and efforts were paid to promote its adoptions. The hydrogen exhibition center in Shanghai has presented a hydrogen electricity-thermal co-generation system with efficiency up to 80%–90%. Hydrogen fuel cells were also applied in heavy-duty mobility applications, such as urban buses and logistic vehicles. With high energy content by mass nearly 3× more than Li-ion batteries, hydrogen fuel cell compensates for the disadvantages of Li-ion batteries of long charging time and low operating range. China is targeted to grow the number of fuel cell vehicles to 30 million and hydrogen fuel cell stations to 12 thousand by 2050. In terms of cost, the United States Department of Energy is targeting reducing the price of fuel cell systems below $30/kw and hydrogen cost at the pump below $4/kg [2]. As a vital part of the global initiative towards hydrogen societies lead by the United Nations Industrial Development Organization, fuel cell technology has received great research interest.

A fuel cell receives hydrogen in the anode side and oxygen in the cathode side, where the hydrogen and oxygen react by exchanging the protons through the fuel cell membrane and letting the electrons go through the outside circuits. The core parts of the fuel cell, namely the catalyst, membrane, gas diffusion layer (GDL) and the fuel cell assembly have received great attention in both academia and industries.

Besides the above-mentioned fuel cell components, a functioning FCS is a complex system with multiple subsystems to provide desirable conditions for fuel cells to work optimally. Moreover, the output of an FCS is soft and unregulated due to the various voltage losses and slow internal dynamics. Therefore, to deliver the regulated power to a load, a DC/DC converter is indispensable to condition the outputs of an FCS. Finally, when the FCS is connected to an energy system, the control strategy of an FCS may require extra design efforts to optimize its operating characteristics. This study provides an overview on fuel cells at the module level, power conditioning level, and energy system level. In particular, this study endeavors to reveal the efforts of achieving more integrated and intelligent fuel cells systems at these levels through modeling, data-based analytics, control strategies, monitoring, and fault detection.

At the fuel cell module level, there are three main loops, namely the anode loop, cathode loop, and cooling loop. The anode loop controls the hydrogen supply, where the main duties are regulating the pressure at the anode side with a pressure valve and preventing nitrogen and water accumulation using an appropriate purging strategy. The cathode loop involves an air compressor to deliver the needed oxygen. Due to the relatively slow dynamics of air compressors, the insufficient oxygen supply in the cathode could happen when the fuel cell is subjected to fast-changing load demands. Therefore, this study provides a detailed review on previous studies on cathode loop modeling and control strategies of oxygen supply at the cathode side. The cooling loop of an FCS is in charge of the temperature regulation through the cooling pump, fans, and valves. Generally, a higher temperature helps to improve the efficiency of the stack. However, when the temperature excesses the maximal level, the drying condition would occur on the membrane and may lead to severe performance degradation. Therefore, a discussion on the optimal temperature setpoint and corresponding control approaches is given in this study. In Figure 2, the schematic of the fuel cell system is given, which shows the logical relationship of the subsystems. As is demonstrated in the schematic, heat, water, and air are transmitted across the various loops of the fuel cell stack. From a system performance perspective, the state of the subsystems has a joint influence on the state of the stack, including but not limited to the pressure difference between the cathode and anode subsystem, the overall management of the water generated from the stack and the water from the cooling subsystem, and the heat transfer across the subsystems. Therefore, the aforementioned subsystems require an integrated management approach in both the hardware and software level to function in a cooperative manner.

Figure 2

Schematic of the fuel cell stack.

This study also presents an overview of the fuel cell interconnecting DC/DC converters, where different topologies were explored to fit the low-voltage high-current characteristics of an FCS and minimize the switching ripples to extend the lifetime of an FCS. Moreover, there is great interest across both academia and industries on incorporating electrochemical impedance spectroscopy (EIS) function into the DC/DC converter, which enhances the intelligence to the FCS systems with advanced diagnostic capabilities. Dedicated reviews and discussions are presented in this study on this topic. Finally, the integrated control of the DC/DC converter and air compressor of an FCS is discussed to improve the fuel cell performances from a system perspective.

In the final part of this study, the FCS utilization in power systems is presented, where the control task is partitioned into two layers based on the time scale, namely slow timescale scheduling and fast timescale energy management. The scheduling is performed by minimizing the operational cost over a certain period of time, while the energy management minimizes the hydrogen fuel consumption with the assistance of energy storage. Accordingly, reviews and discussions are presented for each layer.

This study gives an overview on fuel cells from lower to higher levels and special considerations are given to the advancements and benefits brought by intelligence and integrated control approaches in FCS. In the following, working principles of the proton exchange membrane (PEM) fuel cell technique are reviewed. Then, the anode, cathode, and cooling loops of an FCS are presented. In the next section, a review on the fuel cell DC/DC converter is given, followed by an overview on the integration of hydrogen power generator into power systems. Finally, conclusions are given and future perspectives are discussed.

Fuel cells at module level

Working principle and modeling of fuel cells

The core component of a PEMFC is the membrane electrode assembly (MEA), which is sandwiched by the anode and cathode as shown in Figure 3. The hydrogen is supplied to the anode and the oxygen is supplied to the cathode, while the hydrogen and oxygen react through the membrane. A stack is made up of multiple cells connected in series. The reaction of hydrogen and oxygen in an FCS generates electric power and heat when the hydrogen and oxygen supply are ample [3, 4].

Figure 3

Fuel cell sandwiched structure.

In the electrochemical reactions, the hydrogen oxidation takes place at the anode side and is given by $H_{2} \to 2 H^{+} + 2 e^{-}$ . The successful reaction at the anode side depends on the P_t catalyst, where it is noted that a more effective and cheap substitution is an active research field in fuel cell chemistry to lower the overall cost of manufacturing. The newly formed protons H⁺ permeate through the PEM to the cathode side, while the electrons travel to the cathode side through external circuits, generating the electrical output of a fuel cell. The oxygen reduction reaction takes place at the cathode side and is given by $\frac{1}{2} O_{2} + 2 H^{+} + 2 e^{-} \to H_{2} O$ . Combining the reactions at both sides, the overall electrochemical formula of an FCS can be summarized as $H_{2} + \frac{1}{2} O_{2} \to H_{2} O + Electricity + Heat$ .

Theoretically, the maximal achievable efficiency of an FCS is 83% at the stack temperature of 298 K calculated from Gibbs free energy equation. However, in practice, an FCS suffers from three main voltage losses, named ohmic η_ohm, activation η_act, and concentration η_trans losses. Firstly, the ohmic loss is caused by the electrical resistance of the cell components, including the electrolyte, the catalyst layer, the gas diffusion layer, bipolar plates, interface contacts, and terminal connections. The activation loss arises from driving the electrochemical reaction on the catalytic surface and is dominant in lower current density regions. Finally, the concentration loss, also referred to as mass transport loss, comes from the changes in reactant concentrations at electrode surfaces, is nonlinear with the fuel cell outputs, and is more prevalent in high current densities. Thus, the output voltage of a fuel cell E_cell is formulated as E_OCV - η_ohm - η_act - η_trans, where E_OCV denotes the open cell voltage of the FCS. The FCS output voltage-current curve, referred to as the polarization curve, is commonly used to reveal the output characteristics, where a demonstration is shown in Figure 4.

Figure 4

Fuel cell polarization curve.

Modeling of fuel cells

In the research of PEM fuel cells, there is great interest in establishing a reliable and accurate fuel cell model to facilitate the understanding of fuel cell characteristics and implementation of control strategies. When designing and integrating fuel cell-based systems, such as a hybrid fuel cell powertrain, the fuel cell output characteristics are needed for interactions with a DC/DC converter and energy storage. Therefore, model identification techniques were applied to reveal the fuel cell output characteristics, where load stepping was used in ref. [5] and the sinusoidal signal of EIS was adopted in ref. [6]. Parameter estimation is a popular technique adopted to determine an accurate parametric model based on known model dynamics [7, 8]. For instance, the following parameter optimization algorithm can be introduced [9]: $\min_{(ξ, R_{c}, b)} (y = \sum_{j = 1}^{J} {(V_{sm} - V_{s})}^{2}),$ where y is the objective function, V_sm is the experimental voltage, V_s is the voltage calculated by the existing parameter model, J is the number of the experimental data points, and ξ, R_c, b represent parametric coefficients relating to the activation voltage, the equivalent resistance of the membrane, and the mass transport voltage loss coefficient, respectively.

In the aforementioned literature, the model identification approach is able to reveal the static and dynamic performance of an FCS. On the other hand, a degradation model of FCS was proposed in ref. [10] to include degradations in an EIS-identified impedance-based model. It was pointed out in ref. [11] that the model of an FCS is subject to change during its runtime, therefore, a neural network (NN) based online identification approach was developed to be used in conjunction with predictive controllers.

In general, there are two types of fuel cell models, namely physical and empirical models. The physical model is constructed by using the laws of chemical and thermodynamical reactions and requires deep knowledge of the fuel cells. A widely adopted physical PEMFC model based on electrochemical principles was introduced in ref. [12]. A detailed model of PEM fuel cells was outlined in ref. [13] with explicit equations. The biggest drawback of a physical model is its complexity, which requires large computational resources limiting its applications in real-time. To address this, many efforts were paid to suitable models to be implemented with hardware-in-the-loop (HIL). A multiphysics cell-level 1-dimensional dynamic model was proposed in ref. [14], where a ‘top-down’ modeling approach is adopted, covering electrical, fluidic, and thermal domains. A demonstration of this model adopted in fuel cell emulation was presented in ref. [15] using a buck converter to emulate the electrical output of an FCS. To further reduce the computational burden, a simplified model was proposed in ref. [16] by combining the states of middle homogenous cells. Furthermore, a real-time applicable model was proposed in ref. [17], which uses a single-cell model considering electrical and thermal layers in addition to a stack cell interface model. From the fuel cell system perspective, control-oriented FCS models were proposed in refs. [18-20].

Another difficulty of adopting a physical model in practice is that the parameters required may not be readily available from the datasheets. Uncertainties in the parametric model could greatly impact its accuracy. In comparison, an empirical model is relatively simple and requires less information about the stack, making it more suitable for implementation. Aiming to find a suitable model for power electronic interaction, simplified empirical models were developed in ref. [21], where the models consider stack level only and curve fittings were applied to attain unmeasurable parameters. Moreover, NNs were also used for modeling FCS in ref. [22], where minimal knowledge and information of the fuel cell are needed, and accurate behaviors can be provided.

Cathode loop

The cathode loop is composed of the air compressor, intake manifold, humidifier, cathode flow field, and outlet manifold. The most significant part of the cathode loop is the air compressor, which blows the air into the intake manifold. Most air compressors are made up of a reciprocating piston, and a rotating blade or a rotating screw. Of all types of air compressors, the centrifugal ones are widely adopted. The manifolds, both intake and outlet, connect different cathode loop components. The humidifier is usually located between the intake manifold and the cathode input flow field, and its function is to increase the humidity of the intake air to hydrate the membrane for better reactions. The cathode flow field is placed where the oxygen reduction takes place in an FCS. In the flow field, the flow and pressure of oxygen, nitrogen, and water vapor vary according to the progress of the reaction.

Principle of cathode air-flow control

Fuel cell air-flow control at the cathode side is one of the most significant and challenging aspects in fuel cell management. When the current demand alters drastically, it is necessary to ensure that abundant oxygen is supplied to the cathode by adjusting the air compressor appropriately. Otherwise, oxygen starvation would occur, leading to drying conditions and hot spots on the membrane. In the end, severe oxygen insufficiency would cause irreversible damage to the proton exchange membrane. Oxygen excess ratio (OER) is defined as the ratio between the oxygen supplied and the oxygen consumed, which is commonly adopted to evaluate the sufficiency of oxygen in the cathode. On the other hand, the air compressor consumes the largest amount of power among fuel cell auxiliaries, especially under high load conditions, where it can consume up to 30% of the total fuel cell output power [23]. Therefore, sufficient air should be supplied to avoid oxygen starvation and the air compressor should be controlled to reduce its power consumption at the same time. The main challenges associated with the cathode air supply are the dynamic limitations of the high-speed compressors and the immeasurability of the OER.

In the existing literature on air compressor control, one line of work aims to find the optimal reference for OER, while the other involves estimation of the OER and control towards its reference. To prevent oxygen starvation, a relatively high value, e.g., 2–2.5, is set as the fixed reference in refs. [24-27]. However, a fixed OER will not guarantee the optimal efficiency of an FCS under varying loads. Therefore, an optimized OER trajectory was proposed, where the trajectory was derived from experiments and curve fitting in refs. [28, 29]. The optimal OER from experiments may not always hold true when parameters vary, or when the air compressor and other components of the fuel cell degrade. To address this, the extremum seeking technique was adopted in ref. [30] for finding the optimal OER online.

Since the OER is not directly measurable, its estimation is conducted based on measurable variables. Observers were designed in refs. [26, 27] for estimating the OER in the closed-loop control implementation. A simplified model with the experimental curve fitting for calculating OER was proposed in ref. [31], which depends only on the air compressor mass flow rate and fuel cell current. In order to link the cathode loop control problem to existing control methods, a formulation of the control problem can be given as [32] ${\begin{array}{l} \dot{x} = f_{x} (x, u, w), \\ y = f_{y} (x, u, w), \\ z = f_{z} (x, u, w), \end{array}$ where $x = {[ω_{m}, p_{ca, in}, m_{O_{2}}, m_{N_{2}}, m_{vap, ca}, m_{H_{2}}, m_{vap, an} p_{ca, out}, m_{liq, ca}]}^{T}$ denotes the states of the cathode subsystem detailed in ref. [32], u denotes the control inputs, with u_m the compressor motor voltage, u_A the back-pressure valve opening, and u_pv the proportional pressure regulator input. The measurement y includes the compressor inlet air flow rate W_cp, the cathode inlet and outlet pressures p_{ca, in}, p_{ca, out}, the anode inlet pressure p_{an, in}, and the stack voltage V_stack. The control objective z=[SR_c, p_{ca, in}, Δ p, P_net]^T includes the air stoichiometric ratio, cathode inlet pressure, the cathode/anode pressure difference, and the net power of FCS.

Air-flow control methodologies

(a) Linear feedback control. Linear control techniques were applied for the fuel cell air flow control. A linear observability technique was employed to include the FCS output voltage as a feedback signal in addition to the current feedforward signal, resulting in an improvement on transient oxygen regulation [24]. A diagonal linear control method was developed, and the controllability conditions for achieving desired fuel cell performances were presented in ref. [33]. The model of an FCS can vary under different load conditions and through its usage. Therefore, linear controllers with fixed parameters cannot guarantee the stability and optimal performance.

(b) Adaptive control. Adaptive control continuously estimates and updates control variables to fit the plant variations and unknown parameters over time [34-36]. A robust adaptive control method is proposed in ref. [34] with a closed-loop least squares parameter identification algorithm and closed-loop pole configuration to estimate and prevent oxygen starvation. A Lyapunov-based robust and adaptive high-order sliding mode controller is introduced in ref. [35] and an adaptive neural network (NN) control is applied in ref. [36].

(c) Model predictive control. Model predictive control (MPC) is an online optimization-based control strategy, which finds the optimal control variables by solving the optimization problem regarding a given cost function online. The MPC scheme is also applied to fuel cell air compressor control, where a constrained explicit MPC is proposed for a reduced computational effort in ref. [37]. In this study, both the prevention of air starvation and improvement of overall fuel cell efficiency are considered to form the cost function. To further reduce the computational burden, a linear MPC is proposed in ref. [38].

(d) Fuzzy logic control. The above-mentioned control approaches are primarily model-based, and their performances rely on the accuracy of the model to a great extent. Fuzzy logic-based control approaches can address a complex nonlinear system by developing appropriate fuzzy rules and the complete knowledge of the plant is not required, which makes it attractive in fuel cell control. The work of [39] applies the fuzzy method to the fuel cell air flow control loops, while in ref. [40], fuzzy logics were adopted in both finding the air compressor setpoint and regulating the air flow to the setpoints. A fuzzy-based output current feedforward is used with PI feedback to regulate the OER in ref. [41]. NN-based control approaches were also applied to the model identification and control enhancement of an FCS system. In ref. [42], an NN is adopted to identify and model the nonlinear fuel cell dynamic system, while in ref. [43] the neural optimal control (NOC) was proposed to both generate a direct model of the fuel cell plant and a control voltage output. A summary of control strategies for fuel cell air flow control is given in Table 1.

In addition to the control approaches, optimization techniques have been applied to derive an optimal operating point for the cathode loop with respect to energy efficiency [44, 45]. For instance, such an objective function can be constructed for the maximization of energy efficiency by deriving the optimal compressor speed N and current density J [44]. $f_{obj} ([N, J]) = {(P_{net} ([N, J]) - P_{net}^{*})}^{2} β (1 - δ) - η_{el} ([N, J]) δ,$ where P_net([N, J]), $P_{net}^{*}$ , and η_el([N, J]) represent the net electric power, required net power, and the efficiency of the fuel cell system for the parameter set [N, J]. β and δ are the optimization weights. With the objective function, the optimization problem can be formulated as $\begin{array}{l} {[N, J] (P_{net}^{*}) = ArgMin (f_{obj} ([N, J])) |}_{for a given P_{net}^{*}} \\ subject to {\begin{array}{l} J - J_{\min} \geq 0, \\ J_{\max} - J_{\min} \geq 0, \\ N - N_{\min} (J) \geq 0, \\ N_{\max} (J) - N \geq 0. \end{array} \end{array}$ The constraints N_min(J) and N_max (J) can be approximated from the experimental compressor map. This optimization problem forms the basis for optimal net power air flow control. Following this optimization procedure, a reference compressor speed N^* and reference current density J^* can be given for each reference net power $P_{net}^{*}$ , and reference tracking control can be implemented accordingly. A schematic of the various control strategies for the fuel cell is given in Figure 5.

Figure 5

Schematic of control strategies for the fuel cell. (A) Linear feedback control; (B) model predictive control; (C) fuzzy logic control; (D) neural-based control.

On the other hand, water management is also critical for the operation of the cathode loop, and humidifiers are installed in the cathode loop to adjust the humidification of an FCS. Analytical physical-based models were proposed in refs. [46, 47] for estimating the water content. Gain scheduling control was used for fuel cell water content control in ref. [48], while a decoupled control was proposed in ref. [49] to control OER with an air compressor and regulate humidification of inlet air with supply manifold pressure.

Table 1

Summary of control strategies for air flow control

Anode loop

Anode loop configurations

The anode loop, also known as the hydrogen supply system, is in charge of delivering needed hydrogen to the anode, where its structure can be classified into flow-through, dead-end, and circulation configurations as shown in Figure 6. The flow-through type is normally only adopted in research fields due to the poor fuel economy caused by constantly purging the unreacted hydrogen as exhausts.

Figure 6

Different configurations of anode loop. (A) Flow-through configuration; (B) dead-end configuration; (C) circulation configuration.

In a dead-end configuration, the fuel economy is improved by closing the back purge valve. However, the nitrogen and water vapor would accumulate in the anode channel. The accumulation of nitrogen and water at the anode side would decrease the hydrogen concentration and even block the reactions. It was pointed out in refs. [50, 4] that a low hydrogen concentration rate would lead to carbon corrosion and even irreversible degradation of the catalyst at the cathode side. Therefore, it is necessary to maintain a proper hydrogen concentration and a purging strategy should be developed with consideration towards both the fuel economy and hydrogen concentration.

On the other hand, in the circulation configuration, unreacted hydrogen is circulated back to the reactor inlet using a paralleled hydrogen recirculating loop. This configuration significantly elevates the hydrogen utilization making it the most popular configuration in the industry. However, the installation of extra components in the circulation configuration would increase the system complexity and cost, and the reliability and limited-service time of the circulating pump under humidified conditions need to be considered. Therefore, improvements on the circulation components are needed to further promote its adoption. There are different ways of realizing the circulation loop, including a single hydrogen circulating pump (HCP) or ejector, or multiple ejectors in parallel, or the HCPs and ejectors in parallel. Among them, an HCP is the most popular choice. However, in practice, the HCP occupies a considerable amount of space and weight, as well as consumes a large amount of power, which is not conducive to the improvement of system power density. In contrast, the operation of the ejector does not require power, which is beneficial to the system efficiency [51]. However, its performance at low power is not guaranteed. Therefore, multiple ejectors in parallel or combining the ejector and HCP have been considered, where high-pressure and low-pressure dual ejectors were proposed in ref. [52] and a combination of an HCP and an ejector was established in ref. [53]. In industrial applications, the ejectors are integrated with proportional valves, air-water separators, sensors, and standalone controllers for performance enhancement and condition monitoring. Such integrated subsystem configurations facilitate the development of an intelligent and self-monitoring PEMFC system.

Anode loop control

From the previous section, it can be concluded that the configuration of the anode loop varies widely for each setup, and it is difficult to obtain a uniform dynamic model for the anode loop. Nevertheless, a number of literature have been presented on the control of specific anode loop configurations. Controlling of anode hydrogen pressure was discussed in ref. [32], where the control target is to regulate the pressure difference between the anode and cathode. With the proposed multivariable nonlinear control method in ref. [32], improvements on the transient performance were achieved compared with PI controllers. MPC method to control the purging to maintain a desirable cathode and anode pressure difference was presented in ref. [54].

The accumulation of nitrogen in the anode channel is one of the causes for the performance degradation of an FCS. A simplified analytical model of nitrogen diffusion was established in ref. [3] and the results showed that the accumulation of nitrogen would seriously dilute the hydrogen on the anode side. The concentration of nitrogen can be significantly reduced by purging at a small flow rate. A trans-membrane transportation model of nitrogen with neural networks was proposed and revealed in ref. [53]. Ref. [55] showed that the accumulation of nitrogen mole fraction in the anode side is approximately linear and can be used as a significant indicator for purge schedule. Therefore, the goal of the purging strategy is to maintain a proper hydrogen concentration, which is interchangeable to controlling the nitrogen concentration, as indicated in ref. [56] that when the hydrogen concentration is lower than 75%, the stack voltage will drop rapidly. To this end, ref. [57] proposed an anode nitrogen concentration observer to monitor the nitrogen concentration and developed a corresponding purging strategy to keep the anode hydrogen concentration above 75%. While ensuring the voltage stability of the stack, the hydrogen utilization rate of the system could be more than 99% in this work. Moreover, MPC was also applied for purge control to regulate both hydrogen concentration and pressure difference of cathode and anode in ref. [54].

Cooling loop

When an FCS generates electricity, it also produces heat as a by-product of the operation. In order to ensure efficient electrochemical reactions and extend the lifetime of an FCS, the stack temperature should be kept within an appropriate range. The change of temperature would cause the change of electrochemical reaction rate, as well as the evaporation and condensation of water in the reaction gas. Appropriately increasing the reaction temperature improves FCS system performance as the electrochemical reaction rate could be hastened. However, an over-increased temperature reduces the performance of the stack, and even affects the lifetime of a stack, by causing membrane drying, shrinkage, and even rupture. To this end, it is necessary to develop an efficient heat management subsystem to maintain a uniform temperature distribution at an optimal level throughout the reactor [58, 59].

Thermal management subsystem

The fuel cell thermal management subsystem includes a cooling water pump, radiator plus its fan, electronic thermostat, and positive temperature coefficient (PTC) heaters. The cooling water pump drives the coolant flow, while the radiator extracts heat from the coolant and the radiator fan provides airflow for accelerated cooling. On the other hand, PTC heaters provide heat energy to heat the stack when the cooling system temperature is too low. The heat management subsystem structures of a single and dual loop are shown in Figure 7, where the main difference is whether the fan and heater share the same loop.

Figure 7

Cooling loop.

Lumped control-oriented models were proposed in refs. [20, 59], where the fuel cell was considered as a whole and the dynamics of the entire cooling loop was considered. For the temperature control of the stack, the stack is often regarded as a whole, that is, the temperature throughout the stack is assumed to be uniform. This assumption greatly reduces the calculation difficulty, but the accuracy of the calculation results will be affected. Therefore, the cathode flow channel was divided into thermal control volumes in ref. [47] ensuring the accuracy of the results, while the computational burden was limited to a reasonable level. More complex and sophisticated modeling of fuel cell temperature is given in previous sections on fuel cell modeling. With the increasing implementation of large-scale fuel cell systems such as fuel cell electric vehicles and residential combined heat and power systems, the heat transfer within the PEMFCs has also attracted research attention [60, 61]. In ref. [61], a 3D numerical thermal model was presented to analyze the heat transfer and adopted for predicting the temperature distribution in air-cooled PEMFCs.

Reference operating temperature

A key aspect of thermal management is selecting the optimal reference operating temperature. The reliability and the durability of the membrane, and the bounded stack temperature during transients in an FCS are ought to be taken into consideration. For thermal control FCS, the inlet coolant temperature is often regarded as the temperature inside the fuel cell. However, as the inlet coolant temperature is raised closer to the fuel cell temperature, a higher coolant flow is required to remove the heat and to maintain the same fuel cell operating temperature. Therefore, a higher pump power consumption is required. The trade-off between the temperature distribution effect and power consumption of the thermal management system should be investigated to obtain the optimized reference values of inlet coolant temperature and thus, generate the corresponding coolant flow rate. A series of numerical experiments have been conducted to generate an optimal coolant temperature path and its corresponding flow rate in ref. [62]. Conventionally, the stack is regulated to a constant temperature, according to the datasheet, under various load conditions. In refs. [63, 64], optimal temperature setpoints were derived from testing on open-cathode fuel cells, where the fan controls both temperature and air delivery.

Cooling loop control

The temperature change inside an FCS is a relatively slow process, and the true temperature inside of a fuel cell is not directly measurable. Control-oriented fuel cell thermal modeling was investigated in refs. [20, 59], as well as the temperature regulation under dynamic fuel cell load conditions. A thermoelectric generator was adopted in ref. [65] for regenerating electricity from the wasted fuel cell heat, which improved the overall system efficiency. In the most commonly adopted thermal management strategies, the radiator fan speed is applied to regulate the coolant inlet temperature, while the water pump speed is used to regulate the coolant inlet and outlet temperature difference. This control method can control the stack temperature within a reasonable range, however, due to the strong coupling between the water pump and radiator fan, the stack temperature will fluctuate greatly especially under dynamic load conditions. Therefore, a simplified open cathode FCS thermal model was proposed and the MPC control strategy was presented in ref. [66]. The simulation results show that the control scheme has good response speed and control accuracy for the fuel cell temperature. In order to study the nonlinearities in the thermal management system, ref. [67] proposed a model reference adaptive controller which shows superior performances compared with a linear feedback controller. Another work on applying adaptive control to fuel cell thermal management was proposed in ref. [68], which guarantees the stability of system. A model-based fault tolerant temperature control strategy was proposed in ref. [69] and the results show that the proposed approach can control the stack outlet temperature towards an error limit of ±0.5℃C, as well as providing the diagnostic capability to the components in the cooling loop. To facilitate the understanding of the control logic for the cooling loop, the temperature dynamics of the FCS can be given as [70] $m_{st} C_{P_{st}} \frac{d T_{st}}{d t} = \frac{C_{P_{gas}}}{X n_{e} F} I (T_{in} - T_{st}) - \frac{Δ H_{rxn}}{n_{e} F} I - U A \sum_{i = 1}^{n} (T_{st} - T_{w, i}) - n_{cell} I V_{cell},$ where m_st is the mass of the stack, $C_{P_{st}}$ and $C_{P_{gas}}$ are the heat capacity of the stack and inlet gas, T_st, T_in and T_w,i are the temperature of the stack, the inlet, and the cooling water, I, V_cell, n are the current, voltage and number of fuel cell stack. Xn_e, F, Δ H_rxn, n_e and UA are constants that are detailed in ref. [70]. By partitioning the inlet water into n zones, the temperature of the inlet water can be updated by $\frac{d T_{i}}{d t} = \frac{{\dot{m}}_{cw}}{ρ V_{i}} (T_{i - 1} - T_{i}) + \frac{UA (T_{st} - T_{i})}{ρ V_{i} C_{P_{cw}}},$ where ρ, V_i, and $C_{P_{cw}}$ are the density, volume, and heat capacity of the cooling water. ${\dot{m}}_{cw}$ is the mass flow rate of the cooling water and the control input of the cooling loop.

DC/DC converter for fuel cells

Fuel cell DC/DC converter

Since an FCS comprises serially connecting low voltage cells and the output voltage varies in a wide range depending on the load, a DC/DC converter is necessary for boosting up and fixing the voltage when supplying to a load. The available selection for the DC/DC converter topology is an isolated or a non-isolated one, where the isolated topology provides the galvanic isolation, higher efficiency by achieving soft switching, and easy to reach higher conversion ratio, which is necessary for fuel cell applications as it is normally a low-voltage high-current device. However, more devices and complex designs are required in an isolated converter, leading to a higher cost and lower reliability design.

Therefore, non-isolated converters were investigated in fuel cell applications. In practice, an interleaved boost converter is a common choice for connecting the fuel cell to the load. However, a conventional boost converter is not suitable for high conversion ratio applications due to the increased losses. Therefore, extensive research was conducted searching for optimal topologies of non-isolated high conversion ratio converters. Reviews were conducted on this topic [71] and in ref. [72] for the topologies with the assistance of coupled inductors. Targeting to the fuel cell powertrain applications, a series of studies are done in refs. [73-78], focusing on deriving a simple and efficient topology to fit the wide input range requirement of an FCS using non-isolated topologies.

Apart from the above considerations, minimizing the switching ripple of the fuel cell interfacing converter is also an important aspect. Durability testing was conducted in refs. [79, 80] and revealed that ripples at the fuel cell output tend to accelerate the degradation of a stack. Minimizing the switching ripple is normally achieved with adopting larger components or adding additional components by interleaving. Alternatively, selecting a proper topology can also assist this goal. Therefore, an isolated current-fed dual active bridge converter was proposed in ref. [81], where the primary side duty cycle is fixed to achieve zero switching ripples and the phase shift is adopted for regulating outputs. A patent was submitted by NOVUM engineer ING GmbH, a company dedicated to electrochemical impedance spectroscopy (EIS) technique using converters, to reduce the ripple of a DC/DC converter with a new typology [82]. On the other hand, the advancement of wide bandgap devices, i.e., SiC mosfets, is adopted by many fuel cell DC/DC manufactures for higher efficiency and low switching ripples by achieving higher switching frequencies.

EIS function-based diagnosis

Besides the above-mentioned conventional functions of a DC/DC converter, there is a great interest in embedding fuel cell EIS-based diagnostics to this converter. EIS is a promising technique for fuel cell testing and diagnosis, which was normally used in laboratories on single cells or short stacks. In recent years, with the increasing adoption of fuel cells in the transportation sector, EIS-based diagnosis is highly desirable for automotive high-voltage high-power stacks. Facing the reliability and durability challenges [83, 84], EIS has been proved to be effective for the fault diagnostics [85-87] and durability analysis [10, 88]. The complete process of EIS is presenting ac perturbations at various frequencies to the FCS, calculating the impedance with measured responses, and mapping the impedance spectrum to different conditions of a fuel cell. A schematic of the EIS approach is provided in Figure 8. The principle of EIS diagnosis is based on a simplified FCS model with an open-circuit voltage V_oc and impedance Z_fc, where Z_fc is a complex impedance with both real and imaginary parts Z_fc=RE_fc+jI M_fc. The relation between the output current and voltage is $V_{fc} = \underset{V_{fc}^{dc}}{\underset{︸}{(V_{oc} - Z_{fc} I_{fc}^{dc})}} - \underset{V_{fc}^{ac}}{\underset{︸}{Z_{fc} i_{fc}^{ac}}},$ where the first part is the dc component $V_{f c}^{d c}$ and the second part is the ac component $v_{f c}^{a c}$ . An ac perturbation term is then generated to extract the impedance.

Figure 8

Schematic of the EIS approach.

EIS enabled by the DC/DC converter provides a promising solution of bringing advanced diagnostics to an FCS. However, generating ac perturbations with DC/DC converter is only the first step. The voltage response measurement needs to extract the small ac signals from the high dc level and two solutions were proposed in refs. [89, 90]. Impedance calculation algorithms were proposed in ref. [91]. Finally, mapping the obtained impedance spectrum requires detailed testing and validations on the targeted FCS. The final diagnostic results could be used as an altering system. Moreover, given the identification of fault types, fuel cell embedded controls can be used to mitigate the fault severity [92].

Integrated control of DC/DC converter and air compressor

It was revealed that the FCS reliability and durability are greatly challenged when adopted in the automotive field [83, 84]. This arises from the fast dynamics in the vehicular driving cycle imposed on the fuel cell output directly. From the system perspective, the air compressor of an FCS might not be able to catch up with fast output dynamics leading to air starvation, inappropriate water managements, and hot spots, which could lead to pin holes [93]. Therefore, the hybridization with a battery and proper energy management is required in a fuel cell-powered vehicle to let fuel cell output relatively constant power, which is the strategy adopted in the commercialized Toyota Mira vehicles [94].

On the other hand, to tackle this issue, coordination between the DC/DC converter and air compressor is of great interest. It was pointed out in refs. [21, 95] that the slew rate of the output current should be low, and an integrated control of DC/DC converter and air compressor can avoid oxygen starvation effectively. An important factor on evaluating oxygen supply is the oxygen excess ratio, which is not a direct measurable value and was estimated and controlled in refs. [31, 28, 29]. The oxygen excess ratio needs to be kept high for preventing oxygen starvation and achieving high efficiency [28, 29], which was set as the goal of the coordination control. In the coordination control applications, hybridization with an energy storage is necessary as the slew rate of the fuel cell output is constrained and the energy storage needs to supply the load to avoid a large voltage dip. From hardware integration aspect, DC/DC converter and air compressor driver circuits are combined into one power control unit by Shenzhen Foripower Electric, facilitating the implementation of coordination control. The trend of integration and hybridization will help the fuel cell reduce power fluctuations and significantly enhance its durability.

Integration of fuel cells in power systems

In this section, the coordination and optimization strategies related to fuel cell generation systems (FCGS) are systematically reviewed with a focus on their stationary applications. It is identified from the available literature that the operation of an FC generation system can be framed into a three-layer hierarchy comprising the scheduling layer, the energy management layer, and the control layer. The control layers of FCGS are depicted in Figure 9. The schedule layer calls for the FC generation system to be jointly dispatched with other distributed energy sources (DERs) ahead of time based on forecasting data, the energy management layer optimizes the stack current and manages the power splitting between cells, and the control layer is responsible for regulating each subsystem so as to deliver the power committed.

Figure 9

Different control layers of FCGS.

Conventionally, control of power is achieved by classical feedback control that maintains a fixed ratio of stack current to input hydrogen flow. Here, a key parameter affecting the operational reliability refers to cell utilization, i.e., the fraction of the total hydrogen fed into a fuel cell that reacts electrochemically [96]. For stack protection, it must be kept within a reasonable range when adjusting the input hydrogen flow. Another challenge lies in the energy efficiency of air flow control, especially under heavy load conditions, as experimental studies have revealed that the air-feed system may consume a large portion of the fuel cell power [95]. Readers can refer to Section Fuel cells at module level for the detailed survey on control strategies on the module level. We would like to point out that incorporating the aforementioned technical aspects, such as the cell utilization and the system net power, into upper-layer decision-making is necessary in a practical sense.

In the rest of this section, we reclassify the technical issues that FCGSs may encounter into two categories: slow timescale scheduling and fast timescale energy management. To date, real world applications have shown that the scale-up and implementation of FC generation systems require converters to work in conjunction with battery energy storage systems (BESSs) or ultracapacitors, otherwise, it is difficult to achieve satisfactory dynamic characteristics, peak power capacity, and cold start capability [97]. The long-term scheduling is subject of increased complexity in a power system environment, where the power-to-hydrogen and hydrogen-to-power may be dynamically coupled [98]. In this circumstance, the hydrogen needs to be purchased from the hydrogen market or produced by the electrolyzer powered by the renewables. Therefore, the level of hydrogen (LOH) of both the compressed tank and the water-sealed tank within the electrolyzer system for temporary storage must be strategically managed. An easy-to-implement strategy is activating the FCs during power deficiency and switching to the electrolyzer during power surplus. The inclusion of these ON/OFF states in the FC and electrolyzer introduces logical variables. As demonstrated by refs. [99, 100], ruled-based algorithms can be effective when there are not numerous DERs to coordinate.

Alternatively, the long-term scheduling can be structured as an optimization problem that seeks to minimize the total operating expenditure (OPEX) [101] over a certain period of time while meeting the power/energy demands [102]. For large-scale applications, model predictive control (MPC) is often used to take account of future system behavior, load forecasts, and various operational constraints. Other than this, dynamic programming is also a competitive option by virtue of the special consideration on a sequence of scheduling periods instead of just one [103]. Along these directions, the optimization problem is often reformulated as a mixed-integer quadratic programming (MIQP) problem or linearized further as a mixed-integer linear programming (MILP) problem. They have similar computational efficiency, but the MIQP shows a relatively lower degradation effect in the FC and electrolyzer, as claimed by ref. [104]. Several studies have adopted distributed multi-agent frameworks for improved MPC properties (e.g., [105, 106]) and for intra-hour schedule [107]. Utilizing a communication network, the optimization problem can be solved iteratively at each distributed unit using either consensus algorithm [108], alternating direction method of multipliers [109], or cooperative game theory [110], thereby enhancing the coordination between FCGSs with other DERs. Also, stochastic optimization [111-113] and robust optimization [114, 115] have been incorporated to accommodate the uncertainties. Likewise, ref. [116] included an adjustable uncertainty budget in the cost function to achieve a flexible control on solution robustness and avoid overly conservative decisions. Furthermore, reinforcement learning methods [117-119] have been proposed to relax the requirement of an explicit system model, which determines the optimal time-sequential actions by interacting with the unknown environment.

In view of the expenditure induced by the production, storage, and transportation of hydrogen, a high priority has been placed on the minimization of hydrogen consumption in order to improve the cost competitiveness of FCGSs, which falls into the main task of real-time energy management. In ref. [104], a maximum power point tracking (MPPT) strategy to drive the FCGSs to operate at the maximum power point (MPP) is derived. In order to obtain a higher power level, FCSs are often series-connected, parallel-connected, and series-parallel connected, forming a multi-stack FCGS. Harvesting at the MPP all the time is no longer suited due to the presence of multiple stacks. For the sake of improved fuel efficiency, coordinated power splitting algorithms have been proposed in refs. [120-122], which calculate the optimal allocation of power requested by the scheduling layer between FCSs. During the service of FCGSs, heterogeneity may occur amongst individual cells inside a stack. A typical phenomenon is that some cells exhibit a drop in voltage below the ideal value, making the FCGSs irreversibly degraded. This remains a major barrier to the commercial deployment of FCGSs. In order to mitigate the degradation effects associated with the start-up and shut-down cycles of FCGSs, an energy management strategy is devised in ref. [97] based on a cascade MPC controller. As discussed in ref. [123], comparing the actual voltage with the initial voltage quantifies the degradation degree and thus constitutes a performance loss rate. And based on this, ref. [124] developed a distributed consensus control strategy to handle the performance consistency and guarantee uniform life cycle of all stacks. Moreover, the lifespan of a stack can also be extended by applying energy management to cell voltage. To this end, equalizer systems are proposed in the literature (e.g., [125]) that send energy to the cells with lower voltage to compensate for voltage drops. Fulfilling the schedule contracted with the system operator hours or one day ahead of time, on the other hand, might be sub-optimal and even destructive in practice. Therefore, research efforts have been recently devoted to bridging the timescale gap of control hierarchy in energy systems via real-time optimization [126, 127]. These novel schemes can be generalized to hydrogen-based energy systems to enable cost-effective coordination between FCSs, as well as across FCGSs and other DERs, especially under uncertain operating conditions.

Conclusions and perspectives

Conclusions

This study provides an overview on hydrogen fuel cell technologies with an emphasis on the integrated architectures and intelligent controls at the fuel cell module level, as well as its interconnecting DC/DC converters and applications in energy systems. It is revealed in the literature that apart from the conventional modeling and control techniques of fuel cells, intelligent algorithms, i.e., parameter optimization, and NNs, could achieve similar performance with minimal fuel cell knowledge and robustness to parameter variations. At the fuel cell converter level, hardware and control integration of two major electrical components, which are the DC/DC converter and air compressor, facilitates the performance improvement on oxygen delivery of an FCS. On the other hand, the advanced diagnostics enabled by converter-based EIS can provide more information on the condition of fuel cells, increasing their reliability. Finally, from the system perspective, multiple time-scale layers reveal the need for fuel cell integration to energy systems to perform optimization and management at the system level.

Perspectives on intelligent and integrated fuel cell systems

Based on the surveyed literature, we can conclude that the next generation of intelligent and integrated FCS will be in demand in the near future. Future research focus includes the integrated system level design on both the software and hardware level, as well as utilizing intelligent algorithms to enhance the adaptability of FCSs. A development roadmap towards intelligent and integrated fuel cell systems is given in Figure 10.

Figure 10

Development roadmap for fuel cell systems.

In Sections Fuel cells at module level and DC/DC converter for fuel cells, the control, monitoring, and management of the cathode loop, anode loop, cooling loop, and DC/DC converters are surveyed, respectively. As the subsystems have a joint influence on the performance of the stack, the core objective for future fuel cell systems is developing a cohesive control and management system which includes all the aforementioned subsystems. In order to carry out control and management algorithms for the subsystems, a dynamic model for each subsystem has been developed, where extensive model identification and validation have been performed. However, establishing a comprehensive dynamic model for all subsystems of the fuel cell system is a challenging subject. On one hand, the hardware configuration for FCS varies widely depending on their power and utility. On the other hand, the dynamic model and control algorithm is usually bundled and sold with the component hardware, and the availability for commercial control toolboxes for FCS is limited. To serve the growing research community, providing a comparison table for the fuel cell toolboxes from different companies would be a priority for future research. As surveyed in Section Fuel cells at module level, detailed models for the entire fuel cell system have been developed, but they are generally computationally demanding and are not suitable for real-time implementation. Moreover, modeling uncertainties in the comprehensive model could propagate across other subsystems and amplify the error in control signals. One possible solution to this challenge is the careful selection of the control/management objective, which facilitates the cooperation between subsystems without using the computationally expensive detailed model of the fuel cell system. Another solution is to use data-based algorithms to reduce the reliance on an explicit fuel cell dynamic model. In some applications, artificial intelligence-based approaches have been found to have advantages over model-based numerical approaches, including less computational time, increased accuracy, and better convergence.

Contrary to the traditional model-based numerical approaches, artificial intelligence-based approaches automate the optimization and approximation process by utilizing data and building models relevant to the control problem without explicit model knowledge. For instance, with the availability of reference power and air flow data, fuel cells can be ‘taught’ to manage their air flow to track the optimal oxygen excess ratio. With data-based multi-objective optimization algorithms, more complicated optimization problems can also be solved, providing great potential for more intelligent fuel cell system management.

However, the implementation of artificial intelligence approaches also comes with challenges. The intricate nature of intelligent algorithms means that the availability and more importantly, integrity of data greatly affect their performance. In order to develop a reliable data-based model, extensive data harvesting is required, where a range of fuel cell working conditions are expected to be covered to ensure the data integrity. Considering that security is a key concern in hydrogen fuel cell systems, it is desirable for intelligent approaches to be coupled with secure FCS management strategies with consideration towards risk assessment.

Another major obstacle towards achieving an intelligent and integrated fuel cell system is the hardware architecture. In the current form, most fuel cell systems are equipped with separate controller units for the DC/DC converter, the air compressor, and the fuel cell stack. The main factor for this separation is due to providers for each component presenting their ‘component+control’ solution. In addition, the power and frequency mismatch in the fuel cell components presents challenges to developing an integrated fuel cell controller. In spite of the challenges, meaningful progress has been made on integrated controller hardware for fuel cells, including fuel cell controller unit air compressor controller on the same unit, as well as integrated DC/DC converter and air compressor controller unit as surveyed in Subsection Integrated control of DC/DC converter and air compressor. The next generation of fuel cell controller hardware should integrate the DC/DC converter control circuit, air compressor control circuit, and fuel cell control circuit under the same unit.

Conflict of interest

The authors declare no conflict of interest.

References

United Nations Industrial Development Organization (UNIDO). Towards hydrogen societies: Current advancements in hydrogen technology and pathways to deep decar-bonisation. 2019. https://www.unido.org/sites/default/files/files/2019-04/REPORT_Towards_Hydrogen_Societies.pdf. [Google Scholar]
Satyapal S. DOE hydrogen and fuel cell perspectives and overview of the international partnership for hydrogen and fuel cells in the economy (IPHE). 2020. https://www.energy.gov/sites/prod/files/2020/07/f77/hfto-satyapal-gabi-workshop-jul20.pdf. [Google Scholar]
Promislow K, St-Pierre J, Wetton B. A simple, analytic model of polymer electrolyte membrane fuel cell anode recirculation at operating power including nitrogen crossover. J Power Sources 2011; 196: 10050-10056.[Article] [CrossRef] [Google Scholar]
Matsuura T, Chen J, Siegel JB, et al. Degradation phenomena in PEM fuel cell with dead-ended anode. Int J Hydrogen Energy 2013; 38: 11346-11356.[Article] [CrossRef] [Google Scholar]
Raga C, Barrado A, Lazaro A, et al. Black-box model, identification technique and frequency analysis for PEM fuel cell with overshooted transient response. IEEE Trans Power Electron 2013; 29: 5334-5346.[Article] [Google Scholar]
Dhirde AM, Dale NV, Salehfar H, et al. Equivalent electric circuit modeling and performance analysis of a PEM fuel cell stack using impedance spectroscopy. IEEE Trans Energy Convers 2010; 25: 778-786.[Article] [Google Scholar]
Li Q, Chen W, Wang Y, et al. Parameter identification for PEM fuel-cell mechanism model based on effective informed adaptive particle swarm optimization. IEEE Trans Ind Electron 2010; 58: 2410-2419.[Article] [Google Scholar]
Askarzadeh A, Rezazadeh A. An innovative global harmony search algorithm for parameter identification of a PEM fuel cell model. IEEE Trans Ind Electron 2012; 59: 3473-3480.[Article] [Google Scholar]
Mo ZJ, Zhu XJ, Wei LY, et al. Parameter optimization for a PEMFC model with a hybrid genetic algorithm. Int J Energy Res 2006; 30: 585-597.[Article] [CrossRef] [Google Scholar]
Kim J, Lee J, Cho BH. Equivalent circuit modeling of PEM fuel cell degradation combined with a LFRC. IEEE Trans Ind Electron 2012; 60: 5086-5094.[Article] [Google Scholar]
da Costa LF, Watanabe EH, Rolim LGB. A control-oriented model of a PEM fuel cell stack based on NARX and NOE neural networks. IEEE Trans Ind Electron 2015; 62: 5155-5163.[Article] [Google Scholar]
Mann RF, Amphlett JC, Hooper MAI, et al. Development and application of a generalised steady-state electrochemical model for a PEM fuel cell. J Power Sources 2000; 86: 173-180.[Article] [CrossRef] [Google Scholar]
Spiegel C. PEM Fuel Cell Modeling and Simulation Using MATLAB. Pittsburgh: Academic Press, 2011 [Google Scholar]
Gao F, Blunier B, Miraoui A, et al. A multiphysic dynamic 1-D model of a proton-exchange-membrane fuel-cell stack for real-time simulation. IEEE Trans Ind Electron 2009; 57: 1853-1864.[Article] [Google Scholar]
Gao F, Blunier B, Miraoui A. PEM fuel cell stack modeling for real-time emulation in hardware-in-the-loop applications. IEEE Trans Energy Convers 2010; 26: 184-194.[Article] [Google Scholar]
Gao F, Blunier B, Chrenko D, et al. Multirate fuel cell emulation with spatial reduced real-time fuel cell modeling. IEEE Trans Ind Applicat 2012; 48: 1127-1135.[Article] [Google Scholar]
Jung J-H, Ahmed S, Enjeti P. PEM fuel cell stack model development for real-time simulation applications. IEEE Trans Ind Electron 2010; 58: 4217-4231.[Article] [Google Scholar]
Grasser F, Rufer AC. A fully analytical PEM fuel cell system model for control applications. In: Proceedings of Conference Record of the 2006 IEEE Industry Applications Conference, 2006. 2162–2168 [Google Scholar]
Kunusch C, Puleston PF, Mayosky MA, et al. Control-oriented modeling and experimental validation of a PEMFC generation system. IEEE Trans Energy Convers 2011; 26: 851-861.[Article] [Google Scholar]
Rojas JD, Kunusch C, Ocampo-Martinez C, et al. Control-oriented thermal modeling methodology for water-cooled PEM fuel-cell-based systems. IEEE Trans Ind Electron 2015; 62: 5146-5154.[Article] [Google Scholar]
Ramos-Paja CA, Giral R, Martinez-Salamero L, et al. A PEM fuel-cell model featuring oxygen-excess-ratio estimation and power-electronics interaction. IEEE Trans Ind Electron 2009; 57: 1914-1924.[Article] [Google Scholar]
JemeÏJemei S, Hissel D, PéraPera MC, et al. A new modeling approach of embedded fuel-cell power generators based on artificial neural network. IEEE Trans Ind Electron 2008; 55: 437-447.[Article] [Google Scholar]
Zhao D, Gao F, Bouquain D, et al. Sliding-mode control of an ultrahigh-speed centrifugal compressor for the air management of fuel-cell systems for automotive applications. IEEE Trans Veh Technol 2013; 63: 51-61.[Article] [Google Scholar]
Pukrushpan JT, Stefanopoulou AG, Peng H. Control of fuel cell breathing. IEEE Control Syst Mag 2004; 24: 30-46 [MathSciNet] [Google Scholar]
Pukrushpan JT, Peng H, Stefanopoulou AG. Simulation and analysis of transient fuel cell system performance based on a dynamic reactant flow model. In: Proceedings of ASME International Mechanical Engineering Congress and Exposition, 2002. 637–648 [Google Scholar]
Pilloni A, Pisano A, Usai E. Observer-based air excess ratio control of a PEM fuel cell system via high-order sliding mode. IEEE Trans Ind Electron 2015; 62: 5236-5246.[Article] [Google Scholar]
Deng H, Li Q, Chen W, et al. High-order sliding mode observer based OER control for PEM fuel cell air-feed system. IEEE Trans Energy Convers 2017; 33: 232-244.[Article] [Google Scholar]
Ramos-Paja CA, Bordons C, Romero A, et al. Minimum fuel consumption strategy for PEM fuel cells. IEEE Trans Ind Electron 2008; 56: 685-696.[Article] [Google Scholar]
Chen J, Liu Z, Wang F, et al. Optimal oxygen excess ratio control for PEM fuel cells. IEEE Trans Contr Syst Technol 2017; 26: 1711-1721.[Article] [Google Scholar]
Hayati MR, Khayatian A, Dehghani M. Simultaneous optimization of net power and enhancement of PEM fuel cell lifespan using extremum seeking and sliding mode control techniques. IEEE Trans Energy Convers 2016; 31: 688-696.[Article] [Google Scholar]
Restrepo C, Konjedic T, Guarnizo C, et al. Simplified mathematical model for calculating the oxygen excess ratio of a PEM fuel cell system in real-time applications. IEEE Trans Ind Electron 2013; 61: 2816-2825.[Article] [Google Scholar]
Bao C, Ouyang M, Yi B. Modeling and control of air stream and hydrogen flow with recirculation in a PEM fuel cell system-II. Linear and adaptive nonlinear control. Int J Hydrogen Energy 2006; 31: 1897-1913.[Article] [Google Scholar]
Serra M, Aguado J, Ansede X, et al. Controllability analysis of decentralised linear controllers for polymeric fuel cells. J Power Sources 2005; 151: 93-102.[Article] [CrossRef] [Google Scholar]
Zhang J, Liu G, Yu W, et al. Adaptive control of the airflow of a PEM fuel cell system. J Power Sources 2008; 179: 649-659.[Article] [CrossRef] [Google Scholar]
Laghrouche S, Harmouche M, Ahmed FS, et al. Control of PEMFC air-feed system using Lyapunov-based robust and adaptive higher order sliding mode control. IEEE Trans Contr Syst Technol 2014; 23: 1594-1601.[Article] [Google Scholar]
Wang Y, Wang Y, Xu J, et al. Observer-based discrete adaptive neural network control for automotive PEMFC air-feed subsystem. IEEE Trans Veh Technol 2021; 70: 3149-3163.[Article] [CrossRef] [Google Scholar]
Bordons C, Arce A, Del Real A J. Constrained predictive control strategies for PEM fuel cells. In: Proceedings of 2006 American Control Conference, 2006 [Google Scholar]
Arce A, Ramirez D R, Del Real A, et al. Constrained explicit predictive control strategies for PEM fuel cell systems. In: Proceedings of the 46th IEEE Conference on Decision and Control, 2007. 6088–6093 [Google Scholar]
Williams JG, Liu G, Chai S, et al. Intelligent control for improvements in PEM fuel cell flow performance. Int J Autom Comput 2008; 5: 145-151.[Article] [CrossRef] [Google Scholar]
Tekin M, Hissel D, Pera MC, et al. Energy-management strategy for embedded fuel-cell systems using fuzzy logic. IEEE Trans Ind Electron 2007; 54: 595-603.[Article] [Google Scholar]
Rgab O, Yu DL, Gomm JB. Polymer electrolyte membrane fuel cell control with feed-forward and feedback strategy. Int J Eng Sci Tech 2010; 2: 56-66.[Article] [Google Scholar]
Sun T, Cao G-Y, Zhu X-J. Nonlinear modeling of PEMFC based on neural networks identification. J Zheijang Univ-Sci A 2005; 6: 365-370.[Article] [Google Scholar]
Almeida PE, Simoes MG. Neural optimal control of PEM-fuel cells with parametric CMAC networks. In: Proceedings of the 38th IAS Annual Meeting on Conference Record of the Industry Applications Conference, 2003. 723–730 [Google Scholar]
Tirnovan R, Giurgea S. Efficiency improvement of a PEMFC power source by optimization of the air management. Int J Hydrogen Energy 2012; 37: 7745-7756.[Article] [CrossRef] [Google Scholar]
Bizon N. Improving the PEMFC energy efficiency by optimizing the fueling rates based on extremum seeking algorithm. Int J Hydrogen Energy 2014; 39: 10641-10654.[Article] [CrossRef] [Google Scholar]
Bhatia D, Sabharwal M, Duelk C. Analytical model of a membrane humidifier for polymer electrolyte membrane fuel cell systems. Int J Heat Mass Transfer 2013; 58: 702-717.[Article] [CrossRef] [Google Scholar]
Headley AJ, Chen D. Critical control volume sizing for improved transient thermal modeling of PEM fuel cells. Int J Hydrogen Energy 2015; 40: 7762-7768.[Article] [CrossRef] [Google Scholar]
Karnik AY, Sun J, Stefanopoulou AG, et al. Humidity and pressure regulation in a PEM fuel cell using a gain-scheduled static feedback controller. IEEE Trans Contr Syst Technol 2008; 17: 283-297.[Article] [Google Scholar]
Liu Z, Chen J, Chen S, et al. Modeling and control of cathode air humidity for PEM fuel cell systems. IFAC-PapersOnLine 2017; 50: 4751-4756.[Article] [CrossRef] [Google Scholar]
Chen J, Siegel JB, Matsuura T, et al. Carbon corrosion in PEM fuel cell dead-ended anode operations. J Electrochem Soc 2011; 158: B1164.[Article] [CrossRef] [Google Scholar]
Zhu Y, Li Y. New theoretical model for convergent nozzle ejector in the proton exchange membrane fuel cell system. J Power Sources 2009; 191: 510-519.[Article] [CrossRef] [Google Scholar]
He J, Choe SY, Hong CO. Analysis and control of a hybrid fuel delivery system for a polymer electrolyte membrane fuel cell. J Power Sources 2008; 185: 973-984.[Article] [CrossRef] [Google Scholar]
Berning T, Djilali N. A 3D, multiphase, multicomponent model of the cathode and anode of a PEM fuel cell. J Electrochem Soc 2003; 150: A1589.[Article] [CrossRef] [Google Scholar]
Ebadighajari A, Homayouni H, DeVaal J, et al. Model predictive control of polymer electrolyte membrane fuel cell with dead-end anode and periodic purging. In: Proceedings of 2016 IEEE Conference on Control Applications (CCA), 2016. 1500–1505 [Google Scholar]
Yang CW, Chen YS. A mathematical model to study the performance of a proton exchange membrane fuel cell in a dead-ended anode mode. Appl Energy 2014; 130: 113-121.[Article] [CrossRef] [Google Scholar]
Okedi TI, Meyer Q, Hunter HMA, et al. Development of a polymer electrolyte fuel cell dead-ended anode purge strategy for use with a nitrogen-containing hydrogen GAS supply. Int J Hydrogen Energy 2017; 42: 13850-13859.[Article] [CrossRef] [Google Scholar]
Liu Z, Chen J, Liu H, et al. Anode purge management for hydrogen utilization and stack durability improvement of PEM fuel cell systems. Appl Energy 2020; 275: 115110.[Article] [CrossRef] [Google Scholar]
Asghari S, Akhgar H, Imani BF. Design of thermal management subsystem for a 5kW polymer electrolyte membrane fuel cell system. J Power Sources 2011; 196: 3141-3148.[Article] [CrossRef] [Google Scholar]
Vasu G, Tangirala AK. Control-orientated thermal model for proton-exchange membrane fuel cell systems. J Power Sources 2008; 183: 98-108.[Article] [CrossRef] [Google Scholar]
Rao Z, Wang S. A review of power battery thermal energy management. Renew Sustain Energy Rev 2011; 15: 4554-4571.[Article] [CrossRef] [Google Scholar]
Shahsavari S, Desouza A, Bahrami M, et al. Thermal analysis of air-cooled PEM fuel cells. Int J Hydrogen Energy 2012; 37: 18261-18271.[Article] [CrossRef] [Google Scholar]
Yu S, Jung D. Thermal management strategy for a proton exchange membrane fuel cell system with a large active cell area. Renew Energy 2008; 33: 2540-2548.[Article] [CrossRef] [Google Scholar]
Kandidayeni M, Macias FA, Boulon L, et al. Efficiency enhancement of an open cathode fuel cell through a systemic management. IEEE Trans Veh Technol 2019; 68: 11462-11472.[Article] [CrossRef] [Google Scholar]
Kandidayeni M, Macias A, Boulon L, et al. Efficiency upgrade of hybrid fuel cell vehicles’ energy management strategies by online systemic management of fuel cell. IEEE Trans Ind Electron 2020; 68: 4941-4953.[Article] [Google Scholar]
Hasani M, Rahbar N. Application of thermoelectric cooler as a power generator in waste heat recovery from a PEM fuel cell—An experimental study. Int J Hydrogen Energy 2015; 40: 15040-15051.[Article] [CrossRef] [Google Scholar]
Zhang B, Lin F, Zhang C, et al. Design and implementation of model predictive control for an open-cathode fuel cell thermal management system. Renew Energy 2020; 154: 1014-1024.[Article] [CrossRef] [Google Scholar]
Han J, Yu S, Yi S. Advanced thermal management of automotive fuel cells using a model reference adaptive control algorithm. Int J Hydrogen Energy 2017; 42: 4328-4341.[Article] [CrossRef] [Google Scholar]
Huang W, Qahouq JA. An online battery impedance measurement method using DC-DC power converter control. IEEE Trans Ind Electron 2014; 61: 5987-5995.[Article] [Google Scholar]
Yan C, Chen J, Liu H, et al. Model-based fault tolerant control for the thermal management of PEMFC systems. IEEE Trans Ind Electron 2019; 67: 2875-2884.[Article] [Google Scholar]
O’Keefe D, El-Sharkh MY, Telotte JC, et al. Temperature dynamics and control of a water-cooled fuel cell stack. J Power Sources 2014; 256: 470-478.[Article] [CrossRef] [Google Scholar]
Koç Y, Birbir Y, Bodur H. Non-isolated high step-up DC/DC converters—An overview. Alexandria Eng J 2022; 61: 1091-1132.[Article] [CrossRef] [Google Scholar]
Liu H, Hu H, Wu H, et al. Overview of high-step-up coupled-inductor boost converters. IEEE J Emerg Sel Top Power Electron 2016; 4: 689-704.[Article] [CrossRef] [Google Scholar]
Zhang Y, Shi J, Zhou L, et al. Wide input-voltage range boost three-level DC-DC converter with quasi-Z source for fuel cell vehicles. IEEE Trans Power Electron 2016; 32: 6728-6738.[Article] [Google Scholar]
Wang P, Zhou L, Zhang Y, et al. Input-parallel output-series DC-DC boost converter with a wide input voltage range, for fuel cell vehicles. IEEE Trans Veh Technol 2017; 66: 7771-7781.[Article] [CrossRef] [Google Scholar]
Zhang Y, Fu C, Sumner M, et al. A wide input-voltage range quasi-z-source boost DC-DC converter with high-voltage gain for fuel cell vehicles. IEEE Trans Ind Electron 2017; 65: 5201-5212.[Article] [Google Scholar]
Zhang Y, Zhou L, Sumner M, et al. Single-switch, wide voltage-gain range, boost DC-DC converter for fuel cell vehicles. IEEE Trans Veh Technol 2017; 67: 134-145.[Article] [Google Scholar]
Zhang Y, Liu H, Li J, et al. DC-DC boost converter with a wide input range and high voltage gain for fuel cell vehicles. IEEE Trans Power Electron 2018; 34: 4100-4111.[Article] [Google Scholar]
Bi H, Wang P, Che Y. A capacitor clamped H-type boost DC-DC converter with wide voltage-gain range for fuel cell vehicles. IEEE Trans Veh Technol 2018; 68: 276-290.[Article] [Google Scholar]
Wahdame B, Girardot L, Hissel D, et al. Impact of power converter current ripple on the durability of a fuel cell stack. In: Proceedings of 2008 IEEE International Symposium on Industrial Electronics, 2008. 1495–1500 [Google Scholar]
Wahdame B, Candusso D, FranÇois X, et al. Analysis of a fuel cell durability test based on design of experiment approach. IEEE Trans On Energy Convers 2008; 23: 1093-1104.[Article] [Google Scholar]
Sha D, Xu Y, Zhang J, et al. Current-fed hybrid dual active bridge DC-DC converter for a fuel cell power conditioning system with reduced input current ripple. IEEE Trans Ind Electron 2017; 64: 6628-6638.[Article] [Google Scholar]
Birth S. DC/DC converter with reduced ripple. International Patent WO2017108033A1, Jun. 29, 2017 [Google Scholar]
Li X, Li J, Xu L, et al. Performance analysis of proton-exchange membrane fuel cell stacks used in Beijing urban-route buses trial project. Int J Hydrogen Energy 2010; 35: 3841-3847.[Article] [CrossRef] [Google Scholar]
Post M, Eudy L. BC Transit Fuel Cell Bus Project Evaluation Results: Second Report. Technical Report, National Renewable Energy Laboratory, Golden, CO (United States), 2014 [Google Scholar]
Fouquet N, Doulet C, Nouillant C, et al. Model based PEM fuel cell state-of-health monitoring via ac impedance measurements. J Power Sources 2006; 159: 905-913.[Article] [CrossRef] [Google Scholar]
Gebregergis A, Pillay P, Rengaswamy R. PEMFC fault diagnosis, modeling, and mitigation. IEEE Trans Ind Applicat 2010; 46: 295-303.[Article] [Google Scholar]
Zheng Z, Péra MC, Hissel D, et al. A double-fuzzy diagnostic methodology dedicated to online fault diagnosis of proton exchange membrane fuel cell stacks. J Power Sources 2014; 271: 570-581.[Article] [CrossRef] [Google Scholar]
Hissel D, Candusso D, Harel F. Fuzzy-clustering durability diagnosis of polymer electrolyte fuel cells dedicated to transportation applications. IEEE Trans Veh Technol 2007; 56: 2414-2420.[Article] [CrossRef] [Google Scholar]
Homayouni H, DeVaal J, Golnaraghi F, et al. Voltage reduction technique for use with electrochemical impedance spectroscopy in high-voltage fuel cell and battery systems. IEEE Trans Transp Electrific 2018; 4: 418-431.[Article] [CrossRef] [Google Scholar]
Islam SMR, Park SY. Precise online electrochemical impedance spectroscopy strategies for Li-ion batteries. IEEE Trans Ind Applicat 2020; 56: 1661-1669.[Article] [Google Scholar]
Ordonez M, Sonnaillon MO, Quaicoe JE, et al. An embedded frequency response analyzer for fuel cell monitoring and characterization. IEEE Trans Ind Electron 2009; 57: 1925-1934.[Article] [Google Scholar]
Yan C, Chen J, Liu H, et al. Health management for PEM fuel cells based on an active fault tolerant control strategy. IEEE Trans Sustain Energy 2020; 12: 1311-1320.[Article] [Google Scholar]
Schmittinger W, Vahidi A. A review of the main parameters influencing long-term performance and durability of PEM fuel cells. J Power Sources 2008; 180: 1-14.[Article] [CrossRef] [Google Scholar]
Sugiura T, Tanida A, Tamura K. Efficiency improvement of boost converter for fuel cell bus by silicon carbide diodes. SAE Int J Alt Power 2016; 5: 294-298.[Article] [CrossRef] [Google Scholar]
Vahidi A, Stefanopoulou A, Peng H. Current management in a hybrid fuel cell power system: a model-predictive control approach. IEEE Trans Contr Syst Technol 2006; 14: 1047-1057.[Article] [Google Scholar]
Sedghisigarchi K, Feliachi A. Impact of fuel cells on load-frequency control in power distribution systems. IEEE Trans Energy Convers 2006; 21: 250-256.[Article] [Google Scholar]
Garcia-Torres F, Valverde L, Bordons C. Optimal load sharing of hydrogen-based microgrids with hybrid storage using model-predictive control. IEEE Trans Ind Electron 2016; 63: 4919-4928.[Article] [Google Scholar]
Nojavan S, Akbari-Dibavar A, Farahmand-Zahed A, et al. Risk-constrained scheduling of a CHP-based microgrid including hydrogen energy storage using robust optimization approach. Int J Hydrogen Energy 2020; 45: 32269-32284.[Article] [CrossRef] [Google Scholar]
Zhang Y, Campana PE, Lundblad A, et al. Comparative study of hydrogen storage and battery storage in grid connected photovoltaic system: Storage sizing and rule-based operation. Appl Energy 2017; 201: 397-411.[Article] [CrossRef] [Google Scholar]
Wang Y, Sun Z, Chen Z. Development of energy management system based on a rule-based power distribution strategy for hybrid power sources. Energy 2019; 175: 1055-1066.[Article] [CrossRef] [Google Scholar]
Liu J, Xu Z, Wu J, et al. Optimal planning of distributed hydrogen-based multi-energy systems. Appl Energy 2021; 281: 116107.[Article] [CrossRef] [Google Scholar]
Mahmoodi M, Shamsi P, Fahimi B. Economic dispatch of a hybrid microgrid with distributed energy storage. IEEE Trans Smart Grid 2015; 6: 2607-2614.[Article] [CrossRef] [Google Scholar]
Wu H, Liu X, Ding M. Dynamic economic dispatch of a microgrid: Mathematical models and solution algorithm. Int J Electrical Power Energy Syst 2014; 63: 336-346.[Article] [CrossRef] [Google Scholar]
Garcia-Torres F, Bordons C. Optimal economical schedule of hydrogen-based microgrids with hybrid storage using model predictive control. IEEE Trans Ind Electron 2015; 62: 5195-5207.[Article] [Google Scholar]
Parisio A, Wiezorek C, Kyntaja T, et al. Cooperative MPC-based energy management for networked microgrids. IEEE Trans Smart Grid 2017; 8: 3066-3074.[Article] [CrossRef] [Google Scholar]
Garcia-Torres F, Bordons C, Ridao MA. Optimal economic schedule for a network of microgrids with hybrid energy storage system using distributed model predictive control. IEEE Trans Ind Electron 2019; 66: 1919-1929.[Article] [Google Scholar]
Velasquez MA, Barreiro-Gomez J, Quijano N, et al. Intra-hour microgrid economic dispatch based on model predictive control. IEEE Trans Smart Grid 2020; 11: 1968-1979.[Article] [Google Scholar]
Tang Z, Hill DJ, Liu T. A novel consensus-based economic dispatch for microgrids. IEEE Trans Smart Grid 2018; 9: 3920-3922.[Article] [CrossRef] [Google Scholar]
Ma WJ, Wang J, Gupta V, et al. Distributed energy management for networked microgrids using online ADMM with regret. IEEE Trans Smart Grid 2018; 9: 847-856.[Article] [CrossRef] [Google Scholar]
Du Y, Wang Z, Liu G, et al. A cooperative game approach for coordinating multi-microgrid operation within distribution systems. Appl Energy 2018; 222: 383-395.[Article] [CrossRef] [Google Scholar]
Kou P, Liang D, Gao L. Distributed EMPC of multiple microgrids for coordinated stochastic energy management. Appl Energy 2017; 185: 939-952.[Article] [CrossRef] [Google Scholar]
Romero-Quete D, Cañizares CA. An affine arithmetic-based energy management system for isolated microgrids. IEEE Trans Smart Grid 2019; 10: 2989-2998.[Article] [CrossRef] [Google Scholar]
Shuai H, Fang J, Ai X, et al. Stochastic optimization of economic dispatch for microgrid based on approximate dynamic programming. IEEE Trans Smart Grid 2019; 10: 2440-2452.[Article] [Google Scholar]
Xiang Y, Liu J, Liu Y. Robust energy management of microgrid with uncertain renewable generation and load. IEEE Trans Smart Grid 2016; 7: 1034-1043.[Article] [Google Scholar]
Gabrielli P, Fürer F, Mavromatidis G, et al. Robust and optimal design of multi-energy systems with seasonal storage through uncertainty analysis. Appl Energy 2019; 238: 1192-1210.[Article] [CrossRef] [Google Scholar]
Peng C, Xie P, Pan L, et al. Flexible robust optimization dispatch for hybrid wind/photovoltaic/hydro/thermal power system. IEEE Trans Smart Grid 2016; 7: 751-762.[Article] [Google Scholar]
Foruzan E, Soh LK, Asgarpoor S. Reinforcement learning approach for optimal distributed energy management in a microgrid. IEEE Trans Power Syst 2018; 33: 5749-5758.[Article] [Google Scholar]
Liu W, Zhuang P, Liang H, et al. Distributed economic dispatch in microgrids based on cooperative reinforcement learning. IEEE Trans Neural Netw Learn Syst 2018; 29: 2192-2203.[Article] [Google Scholar]
Du Y, Li F. Intelligent multi-microgrid energy management based on deep neural network and model-free reinforcement learning. IEEE Trans Smart Grid 2020; 11: 1066-1076.[Article] [Google Scholar]
Garcia JE, Herrera DF, Boulon L, et al. Power sharing for efficiency optimisation into a multi fuel cell system. In: Proceedings of IEEE 23rd International Symposium on Industrial Electronics, 2014. 218–223 [Google Scholar]
Wang T, Li Q, Yin L, et al. Hydrogen consumption minimization method based on the online identification for multi-stack PEMFCs system. Int J Hydrogen Energy 2019; 44: 5074-5081.[Article] [CrossRef] [Google Scholar]
Han Y, Li Q, Wang T, et al. Multisource coordination energy management strategy based on SOC consensus for a PEMFC-battery-supercapacitor hybrid tramway. IEEE Trans Veh Technol 2018; 67: 296-305.[Article] [CrossRef] [Google Scholar]
Wang T, Li Q, Yang H, et al. Adaptive current distribution method for parallel-connected PEMFC generation system considering performance consistency. Energy Convers Manage 2019; 196: 866-877.[Article] [CrossRef] [Google Scholar]
Meng X, Li Q, Huang T, et al. A distributed performance consensus control strategy of multistack PEMFC generation system for hydrogen EMU trains. IEEE Trans Ind Electron 2021; 68: 8207-8218.[Article] [Google Scholar]
Bahrami M, Martin JP, Maranzana G, et al. Design and modeling of an equalizer for fuel cell energy management systems. IEEE Trans Power Electron 2019; 34: 10925-10935.[Article] [Google Scholar]
Xu Y, Dong Z, Li Z, et al. Distributed optimization for integrated frequency regulation and economic dispatch in microgrids. IEEE Trans Smart Grid 2021; 12: 4595-4606.[Article] [Google Scholar]
Li Z, Cheng Z, Liang J, et al. Distributed event-triggered secondary control for economic dispatch and frequency restoration control of droop-controlled AC microgrids. IEEE Trans Sustain Energy 2020; 11: 1938-1950.[Article] [CrossRef] [Google Scholar]

All Tables

Table 1

Summary of control strategies for air flow control

In the text

All Figures

	Figure 1 Hydrogen-based energy ecosystem.
In the text

	Figure 2 Schematic of the fuel cell stack.
In the text

	Figure 3 Fuel cell sandwiched structure.
In the text

	Figure 4 Fuel cell polarization curve.
In the text

	Figure 5 Schematic of control strategies for the fuel cell. (A) Linear feedback control; (B) model predictive control; (C) fuzzy logic control; (D) neural-based control.
In the text

	Figure 6 Different configurations of anode loop. (A) Flow-through configuration; (B) dead-end configuration; (C) circulation configuration.
In the text

	Figure 7 Cooling loop.
In the text

	Figure 8 Schematic of the EIS approach.
In the text

	Figure 9 Different control layers of FCGS.
In the text

	Figure 10 Development roadmap for fuel cell systems.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.