Abstract
Planetary journal bearings are enablers for wind turbine gearbox torque density and reliability increase due to their compactness and potentially unlimited lifetime. They are designed to withstand the load conditions during wind turbine operation. Despite their general robustness, abnormal events such as particle contamination, strong overload or operation without sufficient oil supply may be harmful to the bearings. In these cases, damage can occur quickly and with little warning time. Such spontaneous failure leads to turbine downtime and cost-intensive repair work on the wind turbine drive train. Thus, reliable load and condition monitoring systems, which allow the detection of critical operating states before damage occurs, would be beneficial. For journal bearings in wind turbine gearboxes, no commercially available monitoring system exists to date. The existing studies on journal bearing condition monitoring are limited to experiments on component test rigs or small gearboxes, and their transferability to full-size systems has yet to be proven. This work presents the results of a system test with an 850 kW wind turbine gearbox equipped with planetary journal bearings and a novel condition monitoring system based on the measurement of surface acoustic waves. First, the journal bearing design, including the sensor setup, is explained. Second, the test campaign layout is presented. The gearbox is tested under load conditions specific to wind turbines, and the condition monitoring signals are examined in detail. An algorithm based on a machine learning model is presented for evaluating the monitoring signals and predicting the friction state of the bearings. Finally, the practical feasibility and quality of the monitoring approach for planetary journal bearings presented in this work is discussed.