A comparative study of adaptive load frequency controller designs in a power system with dynamic neural network models

Abstract

This paper investigates applications of dynamic neural network (DNN) models for adaptive load frequency controller designs in power systems. The proposed dynamic neural network models have lag dynamics and dynamical elements such as delayers or integrators in their processing units. They only differ in activation functions. The first uses sigmoid functions, the second uses standard fuzzy systems and the third uses non-orthogonal mother wavelets as activation functions. Each DNN model is connected between two area power systems. The input signals of the DNN models are the area control errors (ACE). The outputs are the control signals for two area load frequency control. Adaptation is based on adjusting the parameters of each for load frequency control. This is done by minimizing the cost functional of load frequency deviations. In simulations for each DNN model, comparative results are obtained for damping the frequency due to a load disturbance effect applied to a two area power system