Load Frequency Control of Re Integrated Smart Deregulated Power System Using Reinforcement Learning Based Controller

When there is a lag between generation and load in a power system, frequency deviation happens. The frequency deviation in the deregulated power system is caused by consumers switching to various DISCOs, which causes the load on DISCOs to fluctuate the frequency which causes the undesirable operation in the the power system. The EV aggregators are introduced in each control area to supply power to the DISCOs in the event of a contract violation. This paper presents a reinforcement learning controller for load frequency control of a Smart Deregulated Power System (SDPS) that consists of two control areas, each of which contains thermal, solar PV plants, and hydro, wind plants respectively. The superiority of reinforcement learning controller over model predictive and robust controllers is that the neural network is trained from the control and system parameters of the open access power system under various operating scenarios. The Reinforcement learning controller is called Actor-Critic agent based Deep Deterministic Policy Gradient (DDPG) controller tested on two area model of deregulated power system under different possible contract scenarios and under various operating conditions. The Actor-Cretic reinforcement learning approach for LFC compared to FOPI and PI controllers under different possible contract scenarios in smart deregulated environment.