Short-Term Load Forecasting In Thimphu and Phuentsholing Regions Using Machine Learning and Deep Learning Techniques

Accurate short-term load forecasting is essential for efficient power system operation, energymanagement, and electricity pricing. Traditional statistical methods, such as seasonal autoregressiveintegrated moving averages with exogenous variables (SARIMAX), often fail to capture theintricate and dynamic patterns of electricity demand. Addressing the knowledge gap in developingcountries, particularly Bhutan, this research explores advanced machine learning and deep learningtechniques to enhance short-term load forecasting (STLF) accuracy in the Thimphu andPhuentsholing regions of Bhutan, characterized by unique electricity demand patterns due topopulation growth, industrial and commercial activities, and supply constraints. We evaluatedSARIMAX, Support Vector Regression (SVR), Long Short-Term Memory Networks (LSTM),Convolutional Neural Networks (CNN), and hybrid CNN-LSTM architectures. On single-stepSTLF, we analyzed day-ahead aggregated load forecasts and 1-hour-ahead load forecasts based onhistorical load data over five years (2018-2022) for both Thimphu and Phuentsholing regions. Inday-ahead aggregated load forecasting, the hybrid CNN-LSTM outperformed all other models withMean Absolute Percentage Error (MAPE) values 2.332 ± 0.075% for Thimphu and 3.216 ± 0.036% for Phuentsholing, while also achieving the best MSE, RMSE, and R2 metrics. For 1-hour-ahead forecasting, the CNN model achieved the lowest MAPE of 3.224 ± 0.06% in Thimphu and the hybrid CNN-LSTM model achieved a best MAPE of 3.687 ± 0.027% for Phuentsholing.Careful preprocessing, optimal feature engineering, and hyperparameter tuning were performed forall forecasting types. The findings demonstrate that data-driven approaches significantly enhanceforecasting accuracy, providing valuable insights for energy planners to manage resources andmaintain the power grid, preventing blackouts and other disruptions.