The diagnosis method for abnormal current in converter transformer core and clamps based on neural networks and control chart method
YU Yang1, CHEN Yi1, ZHANG Bao2, ZHONG Jinlong1, WANG Jing3
1. State Grid Jiangsu Electric Power Co., Ltd EHV Branch Company, Nanjing 211102; 2. Changzhou Power Supply Branch, State Grid Jiangsu Electric Power Co., Ltd, Changzhou, Jiangsu 213004; 3. Changzhou Jintan District Power Supply Branch, State Grid Jiangsu Electric Power Co., Ltd, Changzhou, Jiangsu 213004
Abstract:The online monitoring system for converter transformers is a system that evaluates the transformer condition based on characteristic parameters. This system records various parameters during the operation of the converter transformer, including gas composition in the oil, SF6 gas pressure in the bushings, and leakage current in the core clamps. Among these, the core clamp current is a crucial indicator for determining the grounding condition and insulation strength of the transformer core. However, the internal electromagnetic environment of a converter transformer is quite complex during operation, and due to the limitations of sensor precision and operational conditions, the traditional threshold-based abnormal diagnosis methods for core clamp current have a high false alarm rate, posing challenges for refined operation and maintenance. This paper establishes a neural network-based core clamp current prediction model and uses the error between the predicted values and the online values as the observation metric. An abnormal diagnosis method for core clamp current based on the control chart method is proposed. The feasibility of this method is validated using core clamp current data from a ±800kV converter station. The experimental results show that the proposed method can avoid false alarms and accurately diagnose true alarms.
於杨, 陈意, 张豹, 仲金龙, 王静. 基于神经网络和控制图法的换流变铁心夹件电流异常诊断方法[J]. 电气技术, 2025, 26(4): 65-72.
YU Yang, CHEN Yi, ZHANG Bao, ZHONG Jinlong, WANG Jing. The diagnosis method for abnormal current in converter transformer core and clamps based on neural networks and control chart method. Electrical Engineering, 2025, 26(4): 65-72.