Electricity Demand Forecasting Using Adaptive Neuro-Fuzzy Inference System and Particle Swarm Optimization

Abstract

As a state-owned company which supplies electricity to the public, PT. PLN should continuously provide electricity in sufficient amountwith good quality and reliability. As a first step to achieve this goal, PLN should be able to estimate or predict the demand for electricity in the future. This paper proposes the combination of adaptive neuro fuzzy inference system (ANFIS) and particle swarm optimization (PSO) approaches to forecast the electricity demand. PSO is used to optimize the membership function of ANFIS by changing the parameter premise and consequent parameters on ANFIS. This proposed method is implemented in each province in Indonesia. The results indicate that the accuracy of the proposed method increased significantly compared to the standard adaptive neuro-fuzzy inferences system. It can be seen from the Mean Absolute Percentage Error (MAPE) in each method that the ANFIS is 6.8000% and the ANFIS-PSO is 5.5286%.