Neural Network-based Load-Frequency Control in Power Grids

Any change in load or trip of generator causes power mismatch in the system which is initially compensated by the kinetic energy stored in the form of inertia and then by the governor control actions. In order to keep the frequency at nominal value and also to keep the tieline flow at the scheduled value, Load Frequency Control is employed. As the controller needs to be robust, we proposed a Load Frequency Controller based on LSTM Neural Network. To validate the performance of the proposed controller, it is compared with traditional integral controller with various disturbance like increment of load, decrement of load, and removal of generating units. The results show that the proposed LSTM controller is able to capture the details of the dynamics of the traditional integral controller and can be used in place of the traditional controller in single area as well as two-area power systems.