Fault Diagnosis Method for Converter Stations Based on Fault Area Identification and Evidence Information Fusion

DC converter stations have a high voltage level, a long transmission distance, and complex internal equipment, and contain power electronic devices, which seriously endanger the stable operation of the system itself and the active distribution network at the receiving end when faults occur. Accurate fault analysis and diagnosis are critical to the safe and stable operation of power systems. Traditional fault diagnosis methods often rely on a single source of information, leading to issues such as insufficient information utilization and incomplete diagnostic scope when applied to DC transmission systems. To address these problems, a fault diagnosis method for converter stations based on preliminary identification of the fault range and the fusion of evidence information of the switch signal and electrical quantity is proposed. First, the preprocessing of converter station sequential event recording (SER) events and a statistical analysis of event characteristics are completed to initially determine the range of the fault.Then, a fuzzy Petri net model and a BP neural network model are constructed on the basis of the fault data from a real-time digital simulation system (RTDS), and the corresponding evidence information of the switch signal and electrical quantity are obtained via iterative inference and deep learning methods. Finally, on the basis of D-S evidence theory, a comprehensive diagnosis result is obtained by fusing the switch and electric evidence information. Taking the fault data of a DC converter station as an example, the proposed method is analyzed and compared with the traditional method, which is based on single information. The results show that the proposed method can reliably and accurately identify fault points in the protected area of the converter station.