Power-grid network regulation. (A, B) From t=0 s to t=1.5 s, all 341 generators in the network were operating at the equilibrium with Δδ_i=δ_i-δs and Δw being zeros. At t=1.5 s, exploration noise was added to the multi-machine power network, and the data over the time interval [10, 15] s are collected for policy and value learning. The learned output-feedback policy was installed on each generator governor over [15, 20] s for reaching the equilibrium of the power-grid network. (C) The convergence of the policy and value learning is shown for each local generator.