A State Evaluation Method for Distribution Network Secondary Equipment Based on Whole Process Integration and Deep Learning

Aiming at the problem of state evaluation of distribution network secondary equipment, a method integrating whole process integration and deep learning is proposed. Firstly, the whole process integration technology breaks the data barriers in all stages of the equipment life cycle, collects multi-source heterogeneous data including operation parameters, environmental data, maintenance records, etc., and transmits them to the processing center. The Transformer model is used for data fusion processing. Through operations such as local and global feature extraction and multi-head self-attention mechanism, combined with combined pooling and attention layer, effective features are obtained. Then, the trained and optimized model is applied to actual evaluation, and the equipment state results are output to provide decision support for operation and maintenance. In the case analysis, monitoring data from multiple regions in China are selected, and the data set is divided according to 7:2:1. The experimental results show that the proposed method has the shortest training time and the highest evaluation accuracy under different data volumes, effectively improving the accuracy and reliability of the state evaluation of distribution network secondary equipment.