Application of Machine Learning Technique Using Support Vector Machine in Wind Turbine Fault Diagnosis

Abstract Wind energies are one of the most used resources worldwide and favours economy by not emitting harmful gases that could lead to global warming. It is a cost-efficient method environmentally friendly. Hence, explains popularity wind energy production over years. Unfortunately, minor fault be contagious affecting nearby components, then more complicated problem might arise, which may costly. Thus, this article conducted machine learning technique, support vector (SVM) monitor health turbine system classifying class healthy data faulty data. Some SVM types were experimented with, including Linear, Quadratic, Cubic, Fine Gaussian, Medium Coarse Gaussian. Then these models trained under different validation schemes cross-validation, holdout validation, re-substitution as an approach evaluate performance each model. In end, Cubic proven outperformed other provision 10-fold cross-validation with accuracy 98.25%. The result showed has best while Linear least among models. Hence choosing default value preferred final product diagnose in systems.