2018/04/25 by Amirhassan Fallah Dizche, Aranya Chakrabortty, Dizche, Amirhassan Fallah +3
Engineering · #Power System Optimization and Stability #Frequency Control in Power Systems #Microgrid Control and Optimization
paper · pdf · doi:10.48550/arxiv.1804.09827
In this paper we present an online wide-area oscillation damping control\n(WAC) design for uncertain models of power systems using ideas from\nreinforcement learning. We assume that the exact small-signal model of the\npower system at the onset of a contingency is not known to the operator and use\nthe nominal model and online measurements of the generator states and control\ninputs to rapidly converge to a state-feedback controller that minimizes a\ngiven quadratic energy cost. However, unlike conventional linear quadratic\nregulators (LQR), we intend our controller to be sparse, so its implementation\nreduces the communication costs. We, therefore, employ the gradient support\npursuit (GraSP) optimization algorithm to impose sparsity constraints on the\ncontrol gain matrix during learning. The sparse controller is thereafter\nimplemented using distributed communication. Using the IEEE 39-bus power system\nmodel with 1149 unknown parameters, it is demonstrated that the proposed\nlearning method provides reliable LQR performance while the controller matched\nto the nominal model becomes unstable for severely uncertain systems.\n