Combination forecasting based on SVM and neural network for urban rail vehicle spare parts demand

Abstract

Spare parts are indispensable resources to ensure equipment the normal operation and continuous production, especially for urban rau vehicles. When the spare parts storage is insufficient, the equipment can't be replaced or repair ed in time, which can cause serious loss. Therefore, it is important to forecast the demand of the urban rail vehicle spare parts. A combination forecasting method based on SVM and neural network for urban rail vehicle spare parts demand is proposed to accurately forecast the demand of urban rail vehicle spare parts. First, the paper builds two kinds of single forecasting model of SVM and neural network according to the historical consumption data of spare parts and the main factors affecting the use of spare parts. Second, the paper establishes the nonlinear combination forecasting model based on neural network according to the two kinds of single forecasting model. Finally, the paper validates the accuracy of the combination forecasting model by examples and compares the results of two kinds of single forecasting models. The consequence shows that spare parts consumption is forecasted by combination forecasting model agree with the actual spare parts consumption and the result of combination forecasting model is more ideal.