Implementasi Adaptive Neuro-Fuzzy Inference System untuk Prediksi Beban Listrik Jangka Pendek di Gardu Induk 150 kV Tanjung

Abstrak

Electric load fluctuations are influenced by the dynamics of human activities, making short-term load forecasting a crucial aspect of efficient power system management. This study focuses on the 150 kV Tanjung Substation, which supplies electricity to various sectors, including the tourism sector—specifically the three Gili Islands: Gili Air, Gili Meno, and Gili Trawangan, which are prioritized to operate without interruption. Given the high energy demand, the reliability of electricity supply is a key factor in ensuring business continuity. The objective of this study is to develop a short-term load forecasting model using the Adaptive Neuro-Fuzzy Inference System (ANFIS) method and to evaluate its forecasting accuracy using the Mean Absolute Percentage Error (MAPE). The forecast is based on four months of historical load data from 2024. The analysis shows that the optimal ANFIS model is achieved using three input variables from the previous day's load data, with data divided into 9, 4, and 3 sets for training, testing, and validation, respectively. The average MAPE value for this configuration is 3.18% on weekdays and 4.76% on holidays. These results indicate that the ANFIS model is effective for short-term load forecasting at the 150 kV Tanjung Substation.

Kata kunci

• ANFIS
• Mape
• Short term load forecasting
• Tanjung substation
• tourism sector