Data Driven Damage Diagnosis and Prognosis

This proposal is aimed at demonstrating the applicability of a method for data-based, online, real-time monitoring of machine health state and predicting imminent failures. The data driven prognostic system will be based on a new, general state-space basedapproach to parameter tracking in dynamical systems. This method is applicable to systems where the parameters drift at a slower rate than the observable dynamics measured by sensors. This method will be applied to a gray-scale health monitoring andimminent failure prediction in the Airborne Laser (ABL) subsystems. In particular, the initial focus of this work will be on the ABL beam control systems. The main objective of this proposal is to identify specific critical component(s) of the beam controlsystem and demonstrate the ability to predict its failure. This will be accomplished by developing enabling software technologies that will utilize readily available operating data and sensor measurements to monitor systems in real-time so that theincipient damage can be tracked and time-to-failure can be predicted, completed with error estimates. To assist users in analyzing variables associated with the component damages, an unsupervised neural network is used to classify the measurement data anda computer visualization program will show high-dimensional data patterns. Our innovative dynamical-system-based diagnosis and prognosis software system developed in Phase I and II has a great potential for commercial success. The immediate market willbe the defense industry in the United States. Our core technologies can also be applied to process equipment monitoring (e.g., sensors and valves), aircraft equipment monitoring (e.g., engines), power generators, etc.