In this paper, we will assess the performance of a data-driven anomaly detection algorithm, the Inductive Monitoring System (IMS), which can be used to detect simulated Thrust Vector Control (TVC) system failures. However, the ability of IMS to detect these failures in a true operational setting may be related to the realistic nature of how they are simulated. As such, we will investigate both a low fidelity and high fidelity approach to simulating such failures, with the latter based upon the underlying physics.

Anomaly detection is the identification of abnormal system behavior, in which a model of normality is constructed, with deviations from the model identified as “abnormal”. Complex high-integrity systems typically operate normally for the majority of their service lives, and so examples of abnormal data may be rare in comparison to the amount of available normal data. Anomaly detection is particularly suited for Intelligent Fault diagnosis of such systems since it allows previously-unseen or poorly understood modes of failure to be correctly identified.

Differential protection is a popular method to protect aircraft generators against winding faults. Traditional relay-based systems have a limited capability to distinguish between differential current resulting from a winding fault, and the one resulting from measurement noise or current saturation, resulting in false alarms and unnecessary equipment shutdown. Modern aircraft generators are monitored and controlled by advanced generator control units, and therefore, sophisticated signal processing algorithms can be implemented to enhance the differential protection performance.

We analyze potential causes of anomalies, as they vary from incipient system failures to malfunctioning sensors, operating the asset in unusual regions, using inappropriate anomaly detection models, etc. For each cause, we follow the PHM cycle, creating an anomaly resolution action. Within this systematic approach, we focus on one of the most neglected causes for anomalies: the inadequate accuracy of anomaly detection models. We describe a hybrid approach based on a fuzzy supervisory system and an ensemble of locally trained auto associative neural networks (AANN’s).