Wind turbine failure risk assessment model based on DBN

摘要

As wind turbine is mainly composedof two strongly coordinated mechanisms, the transmission mechanismand the energy conversion, fault propagation characteristics and waveforms arefairly complex. Traditional Analysis Methodof Kinetic Model, Expert Systemand SuperficialLearning Model are effective in characteristicrepresentation 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 throughthose methologiesis difficult. This paper proposesto apply DBN Depth Learning Theoryto failure risk assessmentof wind turbine. The experience from analysis onmechanical characteristicsand image onesisdeeply structuredas the working machanism of human brain.Experiment indicatesthat the characteristicanalysis method and failure risk assessment model based on DBNdiscussed in this paperhas better performance inprediction accuracyand evolution abilitythantraditional solutions.