Title: AI-Driven Short-Term Forecasting of Renewable Energy Generation: Empirical Evidence and Operational Implications for Vietnam
Authors: Broustail Chloe Dan Tam , Ha Nguyen Manh
Volume: 9
Issue: 12
Pages: 147-154
Publication Date: 2025/12/28
Abstract:
Vietnam's rapid expansion of solar and wind energy has intensified the need for accurate short-term forecasting to support grid stability, reduce curtailment, and improve operational efficiency. Traditional statistical approaches often fail to capture the nonlinear and highly variable nature of renewable power generation under Vietnam's diverse and rapidly changing meteorological conditions, highlighting the need for more advanced forecasting tools. This study evaluates the effectiveness of artificial intelligence models such as ANN, LSTM, and CNN-LSTM in forecasting renewable energy output using publicly documented datasets from Vietnamese solar power plants, including the TTC1 facility in Taiiy Ninh and a 50 MW plant in Central Vietnam. Following a structured methodological framework comprising data collection, preprocessing, model development, and performance evaluation, forecasting accuracy was assessed using MAE, RMSE, and MAPE. The results show that LSTM models achieve strong predictive performance, with MAPE values of 3-5% under measured weather inputs and 8-10% under forecast-driven conditions, while CNN-LSTM architectures offer marginally higher accuracy at the expense of computational cost. Overall, the findings demonstrate that AI-based forecasting provides a reliable and practical solution for enhancing renewable energy integration in Vietnam and underscore the importance of improving meteorological data quality to further strengthen operational forecasting and policy planning.