Research on Demand Prediction Method for Wind Power Spare Parts Based on Small Sample Data
The global installed capacity of wind power is continuously increasing,and China's wind power industry has also experienced rapid development.The vigorous development of the wind power industry has brought huge challenges to the operation and maintenance(O&M)management.The demand prediction of spare parts in wind power O&M management is an important factor in determining the O&M cost,especially the prediction of spare parts under complex conditions is an urgent problem facing the wind power O&M manage-ment.This paper focuses on the small-sample characteristics of wind power O&M data and conducts research on small-sample spare parts demand prediction methods in the wind power industry.Based on the characteristics of spare parts demand in the wind power industry,dis-crete wavelet transform is selected to extract the trend and detailed features of historical demand data of spare parts,and an appropriate time window is chosen to transform the time series data type into a time series pairs form for supervised learning.A prediction algorithm GA-dwtSVR based on a hybrid genetic algorithm and support vector regression is proposed,and verification experiments are carried out.The experimental results show that compared with traditional prediction methods,GA-dwtSVR has better performance in spare parts demand prediction for wind power and can meet the prediction needs of wind power spare parts based on small-sample data.
Wind power equipment operation and maintenanceSpare parts forecastingGenetic algorithmSupport Vector RegressionWavelet transform
万方数据
维普
