A Comparative Study on AGC of Power Systems Using Reinforcement Learning and Genetic Algorithm

Abstract

In this paper the automatic generation control (AGC) of interconnected power systems is considered. The main objective of Automatic Generation Control (AGC) is to regulate the output of the power system within an area in response to changes in system frequency and tie-line flow. AGC helps to maintain the scheduled system frequency and tie-line power flow with the other areas within the limits. AGCs are mostly composed of an integral controller. The integrator gain is set as a compromise between fast transient recovery and low overshoot in the dynamic response of the overall system. This type of controller is slow in action and does not consider non-linearities in the generator unit. Also it is not robust. So in order to avoid these drawbacks two artificial intelligence techniques are used to tune the integral gains of conventional controller. The reinforcement learning and Genetic algorithm are used for the parameter tuning of AGC. The performance of RL & GA based controller is found to be better than conventional controller, has less complication in controlling power system.