Detection and Identification of Abnormal Topology Relationships in Three-Phase Power System Transformers Using ResNet Model

The power industry is a key foundational energy industry for socio-economic development. In recent years, with the rapid development of the national economy, the electricity consumption has sharply increased, the power supply scale has rapidly expanded, and the demand for electricity services has also been increasing. With the construction of smart grids, the electricity information collection system has achieved full coverage and can obtain a large amount ofbasic electricity information data. In the actual production process, higher requirements have been put forward for the accuracy of various indicators for stable system operation and electricity customer information. To this end, we introduce a new method based on the ResNet (Residual neural network, hereinafter referred to as ResNet) model. This method extracts features directly from raw data through end-to-end learning, and performs anomaly detection and recognition through deep residual networks. We conduct extensive experiments on actual three-phase power system datasets, and the results show that the proposed method has higher accuracy and better generalization ability compared to traditional methods. It can effectively detect and identify anomalies in the topology relationship of three-phase power system transformers, providing strong support for the safe and stable operation of the power system.