Product Quality And System Test Design And Optimization

This thesis studies the method based on the SEA (select-experiment-adapt)learning model. As a typical complex system, product quality system is mainly composed of man, machines, materials, methods, measure and environmental elements, and therefore we should use the system ideas and methods to solve the problem of quality improvement.Deming proposed the PDSA (plan-do-study-action) learning model on the basis of the PDCA (plan-do-check-action) learning model proposed by W.A.Shewhart. The cycle of PDCA-PDSA pointed out that the plan could be continuous improved through studying on the implementation of the results and adjusting the plan. However, PDCA-PDSA model cannot explain other features of the complex adapt system. To meet this situation, the U.S. quality management expert Thomas Pyzdek proposed a new model: select - experiment- adapt.According the theory of SEA models, this solution will be divided into three phases in this thesis. The first phase is to select factors, including the quality objectives and test factors. The second phase is the experimental design and modeling. Experiments are implemented based on practical situation and the corresponding models are established. This phase studies the application of orthogonal polynomial regression in rubber formulation experiment and the application of Interpretive Structural Model in Multi-response Modeling Problems. The third stage is the optimization problem. Because the optimization of the single objective is relatively simple and methods are mature, this thesis mainly studies optimization of multi-response. A new method based on the quality loss function and the function of satisfaction is proposed in this phase.