LOAD FORECASTING OF CUMULATIVE ENERGY REQUIREMENTS OF COMERCIAL CLUSTERS FOR YEAR 2035 USING ARTIFICIAL NEURAL NETWORK

Abstract

Load forecasting is a key aspect of power system design, operation, and maintenance. To obtain an accurate forecast model which is very vital for the development of any power sector capacity expansion plan be it on generation, transmission and distribution, regression (0.0139) and mean square error (0.99791) were used as criteria for selecting the optimal model for load forecasting. The higher the value of regression, the better the value of the accuracy of the load forecast. In the same vein, the lower the value of the mean square error, the better the accuracy of the load forecast. The load forecast using Backpropagation artificial neural network (BPANN) obtained load forecast in the year 2035 to be 85998MW. This forecast is necessary to provide an increase in generation capacity to meet 85836.99MW of load by 2035. Taking 2021 as base year, the total load demand, which comprises the industrial, commercial, and residential customers were estimated and used to forecast the total load requirements of the customers in year 2035. The ten experimental results obtained by creating ten different topologies of BPANN validated the choice of 30 hidden layer neurons for use in the load forecasting.