LFPKMPPM - Lubrication Failure Prediction of Key Mechanical Parts for Predictive Maintenance

An appropriate predictive maintenance strategy is significant for reducing the maintenance cost of large mechanical equipment, and the key lies in early failure behaviour monitoring and prediction of key mechnical parts. The growing demand for wind power in Europe incurs an exponential rise in maintenance costs. To reduce it and control the energy price, predictive maintenance strategies are becoming more important than ever before. As the key support component in a wind turbine, the heavy-loaded bearing supports the most load and is the most vulnerable part (this accounts for 76% of mechanical failures). Evidently, the earlier warning of most unrecoverable failures remains a blind spot due to unreliable technologies for monitoring and predicting early lubrication failure. Focusing on this, the project aims at fundamental research including i) how to develop an ultrasonic measurement method for online monitoring of lubrication-health related variables and ii) how to dynamically characterize and predict lubrication failure. Specifically, this project employs the ultrasonic reflection phenomenon to develop: i) an online simultaneous measurement method of the minimum oil film thickness and surface roughness by acoustic finite element simulation and a mapping model of echo features; ii)real-time lubrication state characterization enabled by fuzzy pattern recognition; and iii) a data-model fusion prediction method of lubrication states with a self-updated Stribeck curve. Overall, this project highlights i) the simultaneous measurement of oil film thickness and surface roughness; ii) the identification of lubrication state and the prediction of lubrication failure in heavy-loaded roller bearings, and iii) the elimination of difficulties that obstruct reliable operation and maintenance of large power generation equipment.