Integrated modular avionics anomaly detection based on symbolic time series analysis

Abstract

Traditional avionics systems are federated architecture, and they are gradually replaced by integrated module avionics (IMA), which can share hardware and software resources within one cabinet. As IMA in civil aircraft becomes more popular, the maintenance, safety and supportability have gradually revealed their importance. In order to ensure the safety operation of the system, it is essential to implement prognostics and health management (PHM) to detect anomalies in time so that the real-time prognostics can be achieved. In this paper, an IMA anomaly detection method based on symbolic time series analysis is proposed. Through the study of failure modes of IMA system, a simulation experiment system was designed to acquire data which can reflect the health status of IMA. The experiment data is symbolized to build upon the D-Markov model and then the anomaly can be measured. These results show that STSA can effectively detect the anomaly of IMA. Besides, this method is able to detect the anomaly that can't be detected by the threshold, which is of great value to guarantee the normal operation.