A Review of Methods for Intermittent Fault Feature Recognition

Intermittent failures are a primary cause of No Fault Found (NFF) problems. They can cause equipment to stop functioning, resulting in serious accidents. Intermittent faults are highly random and unrepeatable, making detection difficult, which seriously affects the reliability and safety of the equipment. Whether it is a traditional fault or an intermittent fault, the fault data are mostly time series in nature. The data anomalies in time series caused by intermittent faults can be regarded as random anomalies, and there exists rich local feature information within the fault interval. Therefore, it is necessary to analyze the time series data of faults to identify the local feature information that indicates obvious intermittent faults. This paper reviews the existing intermittent fault feature recognition methods and analyzes the advantages, disadvantages, and scope of application of various methods. It also reviews feature recognition methods for time series and identifies suitable local feature recognition methods. The future development of solutions to the intermittent fault problem is also discussed.