联合监控均值和方差的CUSUM控制图优化设计

发布时间：2018-07-08 10:06

本文选题：统计过程控制 + 累积和控制图　；参考：《上海交通大学》2014年硕士论文

【摘要】：为了降低质量损失和次品废品的出现，目前制造业的发展需求更高精度的质量检测，当生产过程出现问题时能够快速发现并加以改进。质量偏移分为两种形式，一种为质量特性值的均值偏移问题，代表生产过程远离目标值；另一种为方差偏移，代表生产过程幅度过大。然而在现代生产过程中，我们要求生产过程既要接近目标值，又不能有太大的波动幅度，因此在大部分情况下都需要同时监控生产过程中的均值和方差。传统的休哈特控制图在检测微小偏移上效果并不明显，本文将检测力度更优的累积和控制图引入联合监控均值和方差偏移的控制图方法中，利用马尔可夫链方法来求得新控制图的平均链长，对比文献中各种控制图的平均链长数据以及更换变量数值改进控制图。结果证明新的控制图无论在检测均值或方差的偏移中都有良好的改进效果，，尤其是在小偏移方面有很大的提升。文章在新的模型上继续改进，引入变动抽样间隔的方法，通过长短抽样间隔设置，使得检测微小偏移的力度进一步提高。接着本文考虑到在SPC的监控过程中，有时候很难预测过程的偏移的大小和规格，这使得我们在设定监控参数的时候不能根据固定偏移量来优化结果。本文将休哈特和累积和控制图整合模型应用于联合监控均值和方差偏移的研究中，整合模型在大偏移监控中更加敏感，增大了控制图的应用范围。本文的建模和改进研究结果可以大大提高生产监控能力，降低质量损失。文章最后对研究的不足和未来的发展做了总结。
[Abstract]:In order to reduce the quality loss and the appearance of defective products, the development of manufacturing industry requires higher precision quality detection, which can be quickly detected and improved when problems occur in the production process. Mass migration can be divided into two forms, one is the mean deviation of the quality characteristic value, which represents the production process away from the target value, and the other is the variance deviation, which represents the excessive amplitude of the production process. However, in the modern production process, we require that the production process should be close to the target value without too much fluctuation, so in most cases, it is necessary to monitor the mean value and variance in the production process at the same time. The traditional Shewhart control chart is not effective in detecting small migration. In this paper, the cumulative and control chart, which is better for detecting, is introduced into the control chart method of joint monitoring of mean and variance shifts. The average chain length of the new control chart is obtained by using the Markov chain method. The average chain length data of various control charts in the literature and the numerical improvement control charts of changing variables are compared. The results show that the new control chart has a good improvement in the detection of mean and variance migration, especially in the small migration. In this paper, the new model is further improved, and the method of variable sampling interval is introduced. Through the setting of long and long sampling interval, the detection of small offset is further improved. Then this paper considers that it is very difficult to predict the deviation size and specification of SPC in the process of SPC monitoring which makes it impossible for us to optimize the results according to the fixed offset when we set the monitoring parameters. In this paper, Shewhart and cumulative sum control chart integration model are applied to the study of joint monitoring mean and variance migration. The integration model is more sensitive to large migration monitoring and increases the application scope of control chart. The modeling and improving results of this paper can greatly improve the production monitoring ability and reduce the quality loss. At the end of the article, the deficiency of the research and the development of the future are summarized.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F273.2;TB114.2

【参考文献】