Online Fault Diagnosis of Electromagnetic Launch System via Time Series Anomaly Detection

As a special nonperiodic transient system, the electromagnetic launch system realizes the conversion of ultrahigh power of energy in a few seconds, which is harmful when the system fails. It is urgent to study the online fault diagnosis method of the system to stop the launch in time. Fault diagnosis based on online detection of abnormal waveform of time series in launch period is an important direction to solve the problems. Compared with traditional waveforms anomaly detection, the time series data points of electromagnetic launch system are very large, the time distortion is serious, and the abnormal waveform characteristics are not obvious. Therefore, the traditional methods can not realize online anomaly detection and location. This article analyzes the characteristics of electromagnetic launch time series and proposes a novel named FWSSP-TSAD anomaly detection method. To verify the performance of the proposed method, multiple discharge tests were conducted based on an electromagnetic launch system, and the obtained PFN voltage time series dataset was used as an algorithm input. The results show that the proposed algorithm accurately identifies all abnormal waveforms and extracts all abnormal sub waveforms, achieving fault diagnosis and localization. The average calculation time is less than the window time, which meets the requirements of online fault diagnosis.