Research On Manufacturing Cost Prediction Of Single Piece And Mini-batch Product Based On Improved LS-SVM

With the continuous growth of the personalized needs of customers and development of the information technology, the product development strategy which takes the product family as a unit and the product platform as the foundation play an increasingly important role in the whole product life cycle of the enterprise. The enterprises of single piece and mini-batch production use this strategy to meet the diversified and personalized needs of customers. Product designing and developing are the key components to the whole life cycle of product and the product cost management, also is the indispensable part of management activities for single piece and mini-batch production enterprises. However, there are some problems about single piece and mini-batch product, such as complex in structure, long production cycle, difficult in development and production. How to effectively predict manufacturing cost of single piece and mini-batch product in the design phase of product development is relation to the market competitiveness of products, cost control in production phrase and the quick response rate for customs in orders management process. Therefore, manufacturing cost prediction in the design stage has become an important part for enterprise cost management activities, even one of essential part of business in the enterprise. So, the thesis takes single piece and mini-batch product as the research object, constructing a predicting method and a model for manufacturing cost of single piece and mini-batch product based on the improved LS-SVM. The thesis is mainly focus on the following aspects:Firstly, an overall research on product manufacturing cost prediction was made. On the basis of product manufacturing cost meaning and the importance of manufacturing cost prediction in design stage, the limitation of current methods was analyzed, the characteristics of single piece and mini-batch product was proposed; the research scope was defined and the key problems were analyzed. Then, the overall plan was proposed, which concluded the product modularization and structure decomposition, the analysis of influence factors and the prediction model based on improved LS-SVM. At last the implementation process and technique framework were described.Secondly, the manufacturing cost prediction model of single piece and small batch based on the improved LS-SVM was proposed. On the basis of product family, product platform and modular theory, the product was modeled and decomposed in structure; and the influence factors was analyzed and screened; the prediction model based on improved LS-SVM was proposed via the sample similarity weighting method, and the theory of the improved LS-SVM was explained.Thirdly, the solving process and validation method of the improved LS-SVM prediction model were analyzed. The basic process of the improved LS-SVM prediction model was given, and the important steps were explained. The necessity of predicting model validation, prediction model of evaluation index and calculation method were analyzed.Finally, the research results of this thesis were applied to one company's new product manufacturing cost prediction. Then the modularization and structure decomposition, data collection and the sample determination, acquisition and screening of influence factors of product manufacturing cost was conducted; and also the new products' manufacturing cost was calculated via the prediction model, and then the analysis on its application effects was carried out.