An Integrated Energy Load Forecasting Method Based on Multi-Source Metering Data

With the rapid development of integrated energy harvesting technologies, multi-source metering data provide rich support for user energy load forecasting. However, these data are characterized by high dimensionality, nonlinearity and strong dynamics, and the traditional methods are deficient in feature extraction and long sequence dependent modeling. For this reason, this paper proposes a user energy load forecasting method based on the improved Informer model. Firstly, we design a hierarchical data fusion framework to clean, extract and fuse features from multi-source data to construct a high-quality input feature set suitable for load forecasting; secondly, we introduce relative position encoding and inflated causal convolution in the Informer model to optimize the modeling ability of time-dependence relationship and improve the forecasting efficiency. The experimental results show that the proposed method outperforms the existing methods in terms of prediction accuracy and computational efficiency, and is able to effectively capture the dynamic change patterns of customer loads, providing technical support for comprehensive energy management and optimization.