PERAMALAN BEBAN LISTRIK BULANAN BERBASIS ARTIFICIAL NEURAL NETWORK PADA PT. PLN UP3 TANJUNG PINANG

Electricity is a primary source of energy in both industrial and household sectors. The increase in the number of electricity consumers in the Riau Islands between 2020 and 2021 reflects a continuously growing demand. Therefore, forecasting electrical load is crucial to maintain the stability of the electrical power system and prevent energy wastage and power outages. This research utilizes the Artificial Neural Network (ANN) method to forecast monthly electrical loads at PT. PLN UP3 Tanjung Pinang. The research findings indicate that the adjustment of parameters, such as epoch and batch size, significantly influences the accuracy of the forecasts. The ANN-Backpropagation model is capable of providing estimates that closely align with actual data, with a low Mean Square Error (MSE) of 1.31e+12 and a high level of predictive accuracy, demonstrated by a Mean Absolute Percentage Error (MAPE) of 0.14%. In the context of electricity load planning and management, this model delivers accurate estimates, with the smallest percentage difference being 0.0217%. An analysis of the stable growth trends in electrical loads highlights the potential of utilizing the ANN-Backpropagation model as an effective tool for forecasting and managing electrical loads in the future.