Dynamic quadrature booster control using reinforcement learning

The paper is concerned with the application of a reinforcement learning technique for the learning control of dynamic quadrature boosters to enhance the stability of electric power systems. Learning automata are used to search for optimal controller parameters according to a given performance index. The learning is carried out in a stochastic environment. Simulation results show that this control strategy can be used as an online control strategy for the dynamic quadrature booster installed on a tie-line linking two areas of a power system.