Research On Customer Demand Prediction And Capacity Planning For Service-oriented Manufacturing

Under the service-oriented manufacturing(SOM)pattern,the manufacturing industry improves the performance of supply chain management by increasing the value of value-added services.Although the service intervention and the customer participation are different between the SOM mode and the traditional manufacturing mode,both modes have identical operations management process.The customer demand predication and capacity planning are the two key contents associated with operations management.Futhermore,the two contents have a close relationship,i.e.,the predicated results of the customer demand are the important inputs for the capacity planning.Therefore,on the basis of traditional manufacturing operations management research,this thesis performs research on the above two key contents in the SOM mode.However,the customer demand predication and the capacity planning have their own features of the service intervention and customer participation,the approaches of demand predication and capacity planning in traditionial manufacturing are no longer suitable.Hence,this study uses the customer satisfication index as one of the inputs for the demand predication,and uses the demand predication values considering the customer satisficaton index as the inputs for the capacity planning,thereby taking into account the above features of the SOM in these decision acitivies.This paper provides scientific decision-making theory and basis for the SOM enterprises,which is of great significance to promote the transformation of traditional manufacturing enterprises and improve enterprise efficiency.The thesis uses the refrigeration service system,which is a typical SOM product,as a case study.In the thesis,the literature review,statistics,operations research and systems engineering methods are used.Specifically,the structural equation model,the least square support vector machine(LSSVM),the particle swarm optimization algorithm(PSO)and the stochastic programming method are adopted for the customer demand predication and capacity planning.The main research work of this thesis includes the following parts:(1)The customer satisfication modeling based on the structural equation.First,a framework of impact factors of customer satisfication is constructed through market survey,product analysis,literature analysis,customer interviews(internal discussion and external consultation),and the Delphi method,and the questionaires are designed.Then,the customer satisfication is formulated by the structural equation model,and solved by the partial least square.Finally,taking air conditioner compressor as an example,the customer satisfication modeling process based on the structural equation is introduced in details,and its advantage is demonstrated by comparing with the principal component analysis.(2)The product and service demand predication considering customer satisfication index.The customer demand in the service-oriented manufacturing system contains product and service demands.This part is to develop a novel customer demand prediction approach for SOM using the phase space reconstruction technique,LSSVM with hybrid kernel and the PSO.First,the prediction sample space is reconstructed by the PSR to enrich the time series dynamics.Then,the generalization and learning ability of the LSSVM is improved by hybrid kernel.Finally,the key parameters of the LSSVM are optimized by the PSO.In a real case study,the customer demand prediction of the air conditioner compressor is implemented,and the effectiveness and validity of the proposed approach and the necessity of considering customer satisfication are demonstrated by comparing with other predication approaches.(3)The production and service capacity planning based on stochastic programming with recourse.This part is to address the compressor product and service capacity planning problem with stochastic demands of customers.First,a stochastic programming model with recourse is constructed to describe the problem,and the objective function is to minimize the sum of transformation and purchase cost of production lines,inventory cost and delivery delaying cost of compressors,training cost,fire cost,and additional recruiting cost of engineers.Then,to solve the model,a given number of samples are generated by Monte Carlo simulation approach to generate the same number of optimization problems,and the average objective function value of those problems is employed to approximate the expected value of the original objective function.Finally,taking a compressor company in Shanghai as an example,compared with the capacity planning approach used by the company and the mixed integer programming in deterministic situations,the effectiveness of the stochastic programming based capacity planning approach is validated.