A novel decomposition-based ensemble broad learning system for short-term load forecasting

Abstract

Load forecasting of the power system plays a key role in the production planning and actual operation scheduling of power systems. However, as the power system becomes larger and more complex, it is very important to propose a forecasting model with high accuracy and low computational cost. In this paper, a novel short-term load forecasting model is proposed, which combines the empirical wavelet transform (EWT) and the broad learning system (BLS). The advantage of EWT is that it can decompose the signal into multiple local frequency bands, select the local wavelet function adaptively, and overcome the modal mixing problem caused by the discontinuity of the signal time-frequency scale. While, the advantage of BLS is that it allows the model to learn more by stretching the width, and uses ridge regression to make it less time consuming and faster. To verify the effectiveness, the model is compared with some other state-of-the-art methods, and several performance estimations indicate that the model has high accuracy and low computational cost.