Active Power Control of a Natural Gas/Fuel Oil Turbine Power Plant with Adaptive Neuro-Fuzzy Inference System-Based on Modern Controllers

Abstract

The settings of conventional controllers, which are commonly used in power plants, are determined based on system characteristics during the establishment phase, and, thus are unable to adjust to changing system dynamics over the life of the plant. To avoid this unsatisfactory state, the parameters of the controllers used in electric power plants should be self-adapting. As a result, the goal of this paper is to determine the optimal parameter settings using Ant Colony Optimization-based PI controller (ACO-PI), Fuzzy Gain Scheduled PI (FGPI), Adaptive Neuro-Fuzzy Inference System based PI (ANFIS-PI) and, conventional PI controller for proportional-integral (PI) controllers that will adapt to the changing dynamics of a natural gas/fuel oil turbine power plant's system model and, comparing the system output signal's transient responses. The results show that the ANFIS-PI controller outperforms the other methods.