Prediction Of Feature Matching Between End-of-life Vehicle Recycling And Remanufcturing Manugacturers And Producer Service Demand

With the rapid development of the automobile industry in China, theend-of–life vehicle recycling and remanufacturing industry has also faceda rapidly increase. However, most of disassembling and recyclingmanufacturers are small and medium-sized enterprises and stay in lowmanagement and technical level. It is necessary to develop producerservices industry in order to improve the situation. To better the servicelevel, the thesis works on how to recommend the producer services basedon some features of the producer. This work not only can helpmanufacturers to choose the services, but can help service providers toidentify the potential demands.The content of the thesis can be divided to three parts. First, thecustomer survey questionnaire is designed and conducted in order toanalyze the demand and influencing factors for producer services. Second,on the basis of ascertaining the feature parameters and collecting datathrough questionnaire, principal component analysis is used to eliminateinformation overlapping of the raw data and reduce the input dimension.Then, a prediction model of BP neural network is proposed and finallyused for training, testing data and predicting manufacturers’ servicedemand. Finally, due to the limitations of the BP neutral network, we builda genetic algorithm-BP neutral network mixed model and particle swarmalgorithm-BP neutral network mixed model to further optimize the BPneutral network. The two mixed models go through repeatedly training andtesting and could successfully improve the prediction accuracy, better forecast the demand extent of various producer services.