Manufacturing Multi-value Chain Product Demand Forecasting Based on KPCA-PSO-SVM : -Taking the Demand of the Ring Network Cabinet as an Example

Abstract

With the growth of social demand and economy, the ring network cabinet (RNC) has become the key link of the last kilometer of distribution. Forecasting the demand of RNC from the perspective of enterprises can provide suggestions for the production plan and business decisions of enterprises. Starting from the manufacturing multi-value chain, this paper fully taps the effective information of supply chain, production chain and marketing chain as the input variables of demand forecasting. Considering too many factors involved, this paper uses KPCA to reduce the data dimension, and uses the support vector machine optimized by particle swarm optimization (PSO) to predict the demand of A ring-main unit manufacturing enterprises. In order to verify the validity of the model, ARMA, SVM and PSO-SVM are selected to compare the models. The results show that the KPCA-PSO-SVM adopted in this paper has higher prediction accuracy and efficiency. According to the prediction results, this paper gives corresponding decision-making suggestions.