Dynamic evaluation of wind turbine health condition based on Gaussian mixture model and evidential reasoning

Condition-based maintenance is an effective way to reduce operation and maintenance cost of wind turbine. Highly complex and non-stationary operational conditions of wind turbine pose a challenge to conventional condition monitoring technique. Thus, a systematic multi-parameter health condition evaluation framework that considers the dynamic operational conditions is proposed. After characteristic parameter selection and Gaussian mixture model based multi-regime modeling, evidential reasoning is developed to evaluate the health condition of wind turbine. The proposed approach shows good health condition evaluation performance not only on the parameter level but also on the component and system level. Case studies indicate the effectiveness and potential applications of the proposed method for the wind turbine health condition evaluation.