Health assessment of wind turbine gearbox via parallel ensemble and fuzzy derivation collaboration approach

To simultaneously identify the specific single or compound faults and assess the health state of the wind turbine gearbox accurately, a novel parallel ensemble and fuzzy derivation collaboration approach is proposed based on parallel light gradient boosting machine (Parallel-LightGBM) and group individual evaluation (GIE). In this approach, Parallel-LightGBM is constructed to diagnose more than one fault of wind turbine gearbox, and the fault probability distribution is mastered and translated into the membership function to quantify the fault state objectively. Then, a fuzzy derivation method named GIE is devised to estimate the impact of different detected faults on gearbox’s behavior, and determine the fault risk weights based on the expert perspectives. Meanwhile, an extended ratio system (ERS) is modified to define the health indicator by combining the fault status membership and corresponding risk weights, thus formulate the approach of parallel ensemble and fuzzy derivation collaboration for global health assessment. The effectiveness of the proposed approach was verified on a compound fault test platform for the wind turbine gearbox. The results show that the proposed method is competitive with other existing methods in terms of situation adaptability and performance stability.