Day ahead load forecasting for IESCO using Artificial Neural Network and Bagged Regression Tree

Abstract

Load Forecasting (LF) plays a vital role in the planning, operation and control of power system in deregulated environment. It facilitates the utility in effective decision making related to different critical tasks such as dispatch scheduling of generating capacity, maintenance scheduling, reliability analysis, input/output supply balance, command and control of power system etc. This paper presents an Artificial Neural Network (ANN) and Bagged Regression Tree (BRT) based short term LF model which uses the MATLAB neural network fitting tool to predict day ahead load for Islamabad Electric Supply Company (IESCO), Pakistan. These models use past data of temperature and system load as input of the system. The ANN and BRT are trained on the hourly data provided by IESCO. Extensive simulations and error analysis show the effectiveness and stability of the proposed forecasting models.