Wind Turbine Failure Risk Assessment Model Based on DBN

* Energy & Power Engineering School, North China Electric Power University China, Beijing, ([email protected]) ** Energy & Power Engineering School, North China Electric Power University China, Beijing, ([email protected]) Abstract As wind turbine is mainly composed of two strongly coordinated mechanisms, the transmission mechanism and the energy conversion, fault propagation characteristics and waveforms are fairly complex. Traditional Analysis Method of Kinetic Model, Expert System and Superficial Learning Model are effective in characteristic representation and failure analysis, but their prediction based on risk assessment is not adequately accurate. Furthermore, for big data which is multi-scale, heterogeneous, multi-source, modeling and training through those methologies is difficult. This paper proposes to apply DBN Depth Learning Theory to failure risk assessment of wind turbine. The experience from analysis on mechanical characteristics and image ones is deeply structured as the working machanism of human brain. Experiment indicates that the characteristic analysis method and failure risk assessment model based on DBN discussed in this paper has better performance in prediction accuracy and evolution ability than traditional solutions.