Fault classification with Gauss-Newton optimization and real-time simulation

Anomaly diagnostics and fault classification with prognostics is an active research topic, and real-time detection of anomalies and their classification has remained a critical challenge to be overcome. We developed an innovative, model-driven anomaly diagnostic and fault characterization system for electromechanical actuator (EMA) systems to mitigate catastrophic failures. The efficacy of the Model-based Avionic Prognostic Reasoner (MAPR) approach has been proven in real time using test data acquired from a MIL-STD-1553 testbed. Receiver operating characteristic (ROC) curves are generated as a result of this study to show the tradeoff between sensitivity and specificity. Results of model optimization and fault classification are also presented. This real-time processing will enable enhancements in flight safety and condition-based maintenance (CBM). Once this system is completely mature, flight safety will be improved by allowing the on-board flight computers to read from the MAPR and update their control envelope based on its evaluations of the hardware health, reducing damage propagation, decreasing maintenance time, and increasing operational safety.