A Novel Neuro-Optimal Approach for Decentralized LFC Design in Multi-Area Power System

This paper introduces a robust decentralized load-frequency controller (LFC) for a power system, which is configured based on the structured singular values (SSVs). A new linear quadratic regulator (LQR) solution is applied for a two-area power system, using an artificial neural network (ANN). Subject to a restriction which is based on SSVs, each LFC can be designed independently. The proposed neuro-optimal controller combines the advantages of ANN and the good performance of the SSV-LQR controller. To overcome the problem of SSV-LQR, which needs the measurement of all system states; the suggested controller has only one input from system states in each area and regenerates the other states just using lags of the frequency variations. The results show that the proposed controller has improved dynamic behavior of the system, compared to the both SSV-LQR controller and a genetic algorithm (GA) optimization applied to LFC.