ANN Based Load Forecasting Model for Short Term Planning: A Case Study of Ota Community in Nigeria

Abstract

Load forecasting is fundamentally crucial to the efficient and effective operations of power systems, as accurate load forecast results aid in keeping risks incurred during decision-making processes to a minimum, and also lead to reductions in costs associated with power plant operations. Hence, this paper focuses on short-term load forecasting for a 33/11-kV transmission sub-station in Ota, Ogun State, Nigeria, using an artificial neural network (ANN). The study uses five neural network input parameters, such as days of the week, time of the day in hours, working days, weekends, and total daily load data from two previous weeks. The resulting output parameters after several training (using the Bayesian Regularization (BR) algorithm in the MATLAB ANN toolbox), validation, and testing sessions are the load data for the next two weeks. The performance of the developed model is evaluated using regression plots and the mean absolute deviation (MAD) as well as mean squared error (MSE) indices. Values of 0.993, 0.025, and 0.0025 for regression, MAD, and MSE, respectively, are obtained.