Advanced prognosis and health management of aircraft and spacecraft subsystems

Abstract

Beacon Exception Analysis for Maintenance (BEAM) has the potential to be an efficient and effective model in detection and diagnosis of nominal and anomalous activity in both spacecraft and aircraft systems. The main goals of BEAM are to classify events from abstract metrics, reduce the telemetry requirements during normal and abnormal flight operations, and to detect and diagnose major system-wide changes. This thesis explores the mathematical foundations behind the BEAM process and analyzes its performance on an experimental dataset. Furthermore, BEAM's performance is compared to analysis done with principal component transforms. Metrics are established where accurate reduction of observable telemetry and detection of system-wide activities are stressed. Experiments show that BEAM is able to detect critical and yet subtle changes in system performance while principal component analysis proves to lack the sensitivity and at the same time requires more computation and subjective user inputs. More importantly, BEAM can be implemented as a real-time process in a more efficient manner.