Application in Avionics Anomaly Detection Using Multi-Variant Hotelling T2 Statistics
This article proposes a framework of in-situ monitoring for anomaly detection of avionics, Uses the multi-variant Hotelling T2 statistics to form a quickly reference to the anomaly behavior. The proposed method can be used to solve the data driven Prognostics and Health Management problem to detect the anomaly behavior of equipment as well as the potential isolation and diagnosis of the symptoms of incoming faults. The article also gives a structure of an on-line test and monitoring system prototype, based on the legacy on-line test system, as well as developed a basic information gathering route and a multi-variant test based anomaly detection method. A fault injection based simulation experiment was formed to validate the performance of the system and method under consideration. The results shows that the proposed method has a solid mathematics fundamental and the advantages as well as good performance, worth to be spread.