AI-based MOA fault diagnosis mechanism in wireless networks

Metal oxide arrester (MOA) is an electric power equipment used to limit overvoltage and cooperate insulation between generator substation and direct current converter station. In the actual operation of MOA, due to the abnormal leakage current will produce various faults, the significant increasing of its resistive component will cause thermal effect, and even explosion, so as to lose the protection for the MOA. Particularly, in the transition from preventive maintenance to state maintenance, the importance of its safe operation or fault diagnosis is undoubted. Since the 1990s, fault diagnosis of MOA in operation has been very popular, and it is still popular today. After decades of unremitting efforts, we have accumulated rich experience in fault diagnosis of MOA, and also explored some new fault diagnosis methods and means. With the rise of wireless network technology, its network construction, maintenance, and other aspects have incomparable advantages of wired. Therefore, adopting wireless online monitoring to track MOA status can greatly improve the flexibility, real-time, and accuracy of fault diagnosis, and reduce the error and cost of wired data transmission. In this study, the monitoring module can realize the real-time transmission of MOA operation characteristics to the control terminal through wireless networks, judge the running status and insulation performance of the MOA by the change of the basic component of the resistance current, and then diagnose the fault of the MOA by the proposed weighted k-nearest neighbors algorithm. The simulation results show that the proposed method has very high precision and accuracy in fault diagnosis of MOA, as well as high recall and F1.

Automated summary

The metal oxide arrester (MOA) is a device used to limit overvoltage and protect electric power equipment. Wireless online monitoring and fault diagnosis methods can improve its flexibility, real-time performance, and accuracy.