Online Real-Time Tribology Failure Detection System

The investigation of the coating friction as a function of time is important to monitor the ball bearing heath. Despite the importance of the subject mater, there is a crucial lack of information in the literature about coating life and friction force in ball bearings as coating wear of progressively increases. Here we propose to develop a strategic space vehicle health monitoring system that will identify potential and/or imminent lubrication problems, analyze these parameters in real time, and provide direct input so that these problems are mitigated prior to failure. We will set up a lab experiment environment with a universal microtribometer and acoustic emission sensors measuring the signals associated with wear and the changes that tend to occur as a function of time. Friction force and acoustic signal will be measured with respect to the bearing condition. To capture the dynamic nature of friction evolution, we propose to extract the temporal transient features from the sensing data and develop Hidden Markov Models with four distinct states associated with four operation conditions of the ball bearing. Our system uniquely combine both physics-based and stochastic models for the online diagnosis.