Prediksi Kebutuhan Energi Listrik Kota Medan Menggunakan Metode Jaringan Syaraf Tiruan Model Algoritma Backpropogation

Abstract

This study aimed to predict the electrical energy demand at PT PLN (Persero) UP3 Medan for the next two years using the Artificial Neural Network (ANN) with the Backpropagation model. The purpose was to assist the company in preparing accurate planning for electricity supply. The research used historical data from 2020 to 2024, which included eleven input variables such as the number of customers, total power capacity, and electricity consumption in various sectors. The ANN model was designed in MATLAB R2023b with an architecture consisting of an input layer, one hidden layer, and an output layer. Data normalization was applied using the min-max method, and the network was trained using the Levenberg–Marquardt algorithm. The training process reached the target error within eight epochs, producing a Mean Square Error (MSE) of 0.0387 and a Mean Absolute Percentage Error (MAPE) of 20,49 %, indicating high prediction accuracy. The results show that the predicted electricity demand in 2025 is 1.000 MWh, in 2026 is 1.519 MWh, and in 2027 is 1.550 MWh. The variable with the greatest influence on demand growth was the number of customers in the business sector, followed by total installed power. In conclusion, the ANN Backpropagation model provided highly accurate predictions and could be implemented as a decision support tool for future electricity demand planning at PT PLN (Persero) UP3 Medan.