Abstract:

Traditional control charts did not indicate the real time of the process changes which were change-point. A fuzzy-statistical clustering approach was given to estimate the time of changes based on the fuzzy cluster and statistical theory. The proposed approach has a good result of change -point compared with traditional Shewhart control chart method and FCM method to estimate the change-point of different control charting methods with or without variable sampling strategies in a good deal of simulation. Finally, the method was used in the engine cylinder manufacturing process. The effectiveness and feasibility of this study were tested.

Key words: statistical process control, change-point estimation, fuzzy set theory, fuzzy clustering, variable sampling control chart

摘要：


针对传统控制图不能解决生产过程发生变化的确切时间点即变点估计的问题,基于模糊聚类和统计学理论知识提出了变点估计的混合模糊统计聚类方法。在模拟仿真试验下,验证了该方法无论在固定抽样策略还是在可变抽样策略下,与传统的修哈特控制图法和FCM方法相比,对于控制图的变点估计都有着良好的效果。最后将该方法运用于发动机缸体制造的过程控制之中,检验了其有效性和可行性。



关键词: 统计过程控制, 变点估计, 模糊集理论, 模糊聚类, 可变抽样控制图