Short-Term Load Forecasting Method using WaveNet based on Optimized Variational Mode Decomposition

Enhancing the accuracy of load forecasting holds significant importance for ensuring the safe and stable operation of distribution networks. This paper proposes a short-term load forecasting model by combining the Optimized Variational Mode Decomposition (OVMD) algorithm with the WaveNet algorithm. The decomposition algorithm is employed to extract different frequency trend components from the load time series, forming historical load temporal feature maps. These maps strengthen the load temporal features, improving the interpretability of external features in relation to load trend variations. Additionally, considering external meteorological and temporal features such as dates, multidimensional temporal features are constructed. The WaveNet model learns from these temporal features and produces load forecasting results. Finally, experiments on load forecasting for 10kV busbars in distribution networks validate the effectiveness of the proposed model.