Optimization Design Of The Optimum Process Mean Based On Loss Function

As the increasingly fierce competition in various industries and the improvement of Chinese living standards, only providing the eligible products and services have not guarantee organizations’success and development. How to design and produce high quality, low cost and short development cycle product, has become a problem urgently to be solved in enterprises. In addition to increase research investment to improve the technological content of products, on the other hand, it also needs continuous quality improvement. The optimization design of process mean, which is an important supporting technology in the continuous quality improvement activities, mainly is used in the design phase of product /process. Without the need of increasing investment or little investment, it can not only reduce the production costs, increase the market competitiveness, but also achieve the goal of product quality improvement.In the paper, taking the determination of the optimal process mean and response optimization design as the subjects of the research, by means of response surface method, quality loss function, the optimization modeling, two layers of nested optimization algorithm and empirical research, we systematically study the problems of different marking strategies of manufactures have a different emphasis degree on the internal costs and the external costs, and because of the interference of uncertainty factors, the validity of the prediction model about the distributed parameter will be reduced. Some main results in this paper as summarized as follows.(1) The optimization design of process mean based on the unequal weighted loss costs. An optimization model for the process mean is proposed based on the unequal weighted loss costs, which is caused by different marketing strategies of manufactures have a different emphasis degree on the internal costs and the external costs. Firstly, response surface models are built for the process mean and dispersion effects. Then, optimization model is built considering the unequal weighted loss costs between the internal costs and the external costs. Finally, the optimal process mean and parameter levels are selected under minimizing the expected total cost. The characteristic of this method is that, it makes results more flexible because manufacturers will select the appropriate optimal process mean and parameter levels according to their types of the corresponding marketing strategies and the degree of attention to the internal costs and the external costs.(2) The optimization design of process mean based on confidence intervals about process distributed parameters. Generally, the process distributed parameters are unknown. Because of the uncertainty interference factors always exist in industrial design, there is a certain deviation between the parameter estimates with the real values, when only use the limited experimental data to build response surface models for this parameters. Thus, putting forward from the perspective of the confidence intervals of model parameters, combining with robust design principle, and based on the method of minimization of the worst strategy, two layers of nested optimization model is built to determine the optimal process mean and parameter levels with imprecise process distributed parameters. The results of the case study presented that, if not considering the interference of uncertainty factors, manufacturers will likely make a mistake in selecting the optimal process mean and parameter levels, and may even underestimate the production budget.This paper further enrich the research content the optimal process mean and parameter levels in the field of quality design, and it provides some new research ideas and solutions to select the optimal process mean in the design phase of product/process. On this basis, the thesis also discusses some issues which deserve further research in the future.