Decentralized stochastic control of multi-machine power systems

Abstract

A decentralized feedback control scheme is proposed for optimization of large-scale systems. First, local controllers are used to optimize each subsystem, ignoring the interconnections. Next, an additional compensating controller was applied to minimize the effect of interactions and improve the performance of the overall system. At the cost of the suboptimal performance, this optimization strategy ensures stability of the systems under structural perturbations. To account for the modeling uncertainties, both a local Kalman filter and recursive least square algorithm are used to estimate all local states and interactions for each subsystem. The controller uses these estimates, optimizes a given performance index and then regulates the system. A sample three-bus system is given to illustrate the proposed methodologies.