Electricity Consumption Modelling in Kendari using The Backpropagation Method on The Artificial Neural Network

Kendari is the capital city of Southeast Sulawesi province with significant population growth. Population growth has resulted in higher demand for electricity, while Kendari city electricity source is only obtained from diesel power plants, of course, the amount of electricity available is limited compared to other power plants. For this reason, it is necessary to model the consumption of electricity in the city of Kendari, as well as forecasting peak loads at certain times of the day. In fact, the assumption of the data cannot be fully fulfilled, even though the assumption has been handled. For this reason, forecasting using the backpropagation method in artificial neural networks provides a solution when the assumptions of statistical data are not fulfilled. The purpose of this article is to model electricity consumption in Kendari city using the back propagation method on artificial neural networks. The best modelling results use a network structure with 10 input layer units and 4 hidden layer units with the smallest mean square error of 0.000145. The highest average peak load on a daily basis occurs at 8.00 PM is 75,593 MWh, while the lowest electricity load that occurs at 05.00 AM is 64,203 MWh. Forecasting of the mean of electricity consumption the next week will produce 69,079 MWh on Monday, 69,381 MWh on Tuesday, 68,550 MWh on Wednesday, 69,124 MWh on Thursday, 68,110 MWh on Friday, 67,927 MWh on Saturday, and 68,833 MWh on Sunday.