Research And Implementation Of The Health Model In The Fault Diagnosis Of Wind Turbines

In the face of the increasingly serious environmental pollution and energy shortage,wind energy as a clean and environmentally friendly renewable energy is getting more and more attention of the world.With the increase of the capacity of the global wind power,the frequency of the wind turbine accident increases sharply.In this background,the fault diagnosis technology of wind turbine becomes more and more important.This paper investigate the present situation of the fault diagnosis technology of wind turbine and study the related technologies in this field.Based on feature selection and machine learning theory,this paper researches the health of wind turbine,and proposes a method of feature selection and model training.First,the original SCADA data is discretized and standardized,and then the unrelated feature is filtered out according to the information gain ratio.Finally,the feature selection and model training are carried out by using genetic algorithm and logistic regression.After the experiment,the prediction ability of the model to the unknown data is up to 78.1%.