A Novel ANFIS Controller for LFC in RES Integrated Three-Area Power System

Abstract

This paper presents a novel Adaptive Neuro-Fuzzy Inference System (ANFIS) model for Load Frequency Control (LFC) with an expanded input configuration, incorporating the integral of the area control error (ACE) alongside the traditional ACE and its derivative. This additional input captures historical ACE trends, enhancing the ANFIS control performance. The ANFIS training dataset, comprising ACE error, its derivative, and integral, is generated using a PID controller tuned by a variant of Particle Swarm Optimization (PSO) algorithm called an Adaptive Dynamic Inertia Weight Acceleration Coefficient (ADIWACO). Its evaluation on a three-area power system with renewable energy sources (RES) includes comparative analysis with PID, traditional 2-input ANFIS, Fuzzy Logic, and Artificial Neural Network (ANN) controllers. Simulation results demonstrate the superior performance of the proposed 3-input ANFIS controller in terms of performance metrics, consisting of overshoot, undershoot, settling time, steady-state error, and Integral Time Absolute Error (ITAE). Notably, the proposed ANFIS model shows a significant 75.89% improvement in ITAE value over the traditional 2-input ANFIS when communication delays and governor dead band constraints are considered, underscoring the significant impact of the additional input. System parameters variation of ±25%, further confirms the controller's robustness to uncertain model parameters. This study contributes to advancing real-world application of ANFIS controllers for LFC in interconnected power systems integrated with the two most widely developed renewable resources, namely solar and wind power plants.