Study of the Performance of the Fuzzy Controller and the Neuro-Fuzzy Inference-Based Controller When Using in Regulating the Frequency of the Electrical Grid

Abstract

Load frequency controlling is considered as an important issue for determining grid reliability and stability. This study provides a comprehensive framework for analyzing and improving frequency regulation performance in a thermal generation unit by comparing two intelligent control methodologies: the conventional fuzzy controller and the neuro-fuzzy inference-based controller (ANFIS).

     We began by simulation of a linear mathematical model of the thermal plant. The model was implemented using the MATLAB/Simulink environment to verify its dynamic characteristics and generate reliable training data for the ANFIS controller. The fuzzy controller was designed using frequency error as an input and the control signal to the turbine valve as an output. The ANFIS model was trained using data extracted from the system response with a PID controller, enabling it to learn the nonlinear relationships between inputs and the control signal autonomously and adaptively.

     Both strategies were tested on the thermal model under various load change scenarios. Results indicate that the fuzzy controller fails to achieve stable performance under different operating conditions. In contrast, ANFIS provides better response, superior damping of frequency oscillations, and greater adaptability to nonlinearities and sudden load changes.