Application of Status Monitoring of Wind Turbines Based on Relevance Vector Machine Regression
Based on the single kernel function relevance vector machine(RVM) models，a multiple load-forecasting model has been established and simulated with several compound kernel functions, including Gauss kernel, Laplace, linear compounded by Gauss and Laplace, Gauss and polynomial kernel. Each model gained comparatively reasonable results in simulation .Moreover, multi linear-compound kernel RVMs performed better than single kernel RVMs in terms of most evaluating indicators, which prove that RVM is an appropriate machine learning method in monitoring status of components of wind turbines.
Weiguo Pan, Jianxing Ren and Yongguang Li
J. P. Sun and L. T. Hu, "Application of Status Monitoring of Wind Turbines Based on Relevance Vector Machine Regression", Advanced Materials Research, Vols. 347-353, pp. 2337-2341, 2012