AGC in two-area deregulated power system using reinforced learning neural network controller

摘要

In the present work, the effect of bilateral contract have been analyzed on the dynamics of conventional two-area Automatic Generation Control (AGC) system. Then a multilayer perceptron neural network (MLPNN) controller for each area in a two area deregulated power system with reinforced learning is considered for the system. The weights of the MLPNN are dynamically adjusted online using backpropagation method and its performances are compared with the integral controllers whose integral gain and speed regulation parameter are simultaneously optimized using simulated annealing algorithm (SA) for various loading conditions, contract participation among generating units and contract violation by the distribution companies. Investigation reveals that MLPNN controller gives better performances compared to integral controllers obtained using SA.