基于过程状态特征和广义推断的统计过程控制方法研究

In many real productive processes, process data follow non-normal distributions; tolerance and quality loss are often asymmetric. All these outstanding problems bring many difficulties and great challenge to statistical process control (SPC), thus it is very urgent to carry out researches on statistical process control to solve the above questions.. Process state characteristics are adopted to monitor the process state instead of distribution parameters, which will depict the process state more objective and accurate than the traditional distribution parameters; and the goal of this project is to establish analysis, design and evaluation methods for statistical process control. Generalized inference and randomized estimation and statistical process control theory are integrated flexibly, and process capability analysis, process state control (control chart) and simultaneous optimization of robust parameter and tolerance will be carried out. Considering the asymmetric characteristic of the process distribution and quality loss, some new and suitable process capability indices will be proposed and studied, furthermore, their confidence interval and lower confidence limits will be precisely estimated using the generalized inference and randomized estimation; distribution characteristics of process variables will be extracted from the viewpoint of distribution, and some distribution characteristics, such as probability density function and likelihood function, are taken as monitoring variable to design and analyze control charts of non-normal processes by the generalized p-value test and the VDR (vertical density representation)test; considering the non-normality of process distribution, the asymmetry of tolerances, and their intimate relationship with the process capability indices, simultaneous optimization of robust parameter and tolerance is studied based on generalized linear models.. The research results of this project will not only extend and develop the theory and method of statistical process control, but also provide effective guidance for the quality control and improvement of productive processes, service processes and so on.

实际生产过程中，大量出现的过程数据非正态性、设计公(容)差和质量损失的非对称性等突出问题，给统计过程控制理论带来了诸多困难和巨大挑战。.本项目以反映上述突出问题的过程状态特征为监控对象，以建立相应的统计过程控制分析、设计和评价方法为研究目标。将广义推断、随机估计等方法和统计过程控制理论有机融合，以过程能力分析、过程状态监控与稳健设计优化为主线开展研究。具体突出过程分布的非正态性、损失的非对称性，开展过程能力指数研究，融合广义推断和随机估计等方法，构造过程能力指数的精确区间估计；从分布角度提取分布密度、似然函数等过程状态特征，综合运用广义p值检验和VDR检验等，进行非正态控制图设计分析；综合考虑过程分布的非正态性与容差、损失的非对称性，基于广义线性模型开展稳健参数与容差并行优化设计。.本项目预期研究成果将在丰富和发展现有统计过程控制理论的同时，有效指导生产、服务等复杂过程的质量管理工作。

本项目针对过程质量控制存在的过程数据非正态性、容差非对称性与质量损失非对称性等突出问题，以反映这些突出问题的过程状态特征为研究对象，以建立相应的统计过程控制分析、设计和评价方法为研究目标。将广义推断、随机估计等方法和统计过程控制理论有机融合，以过程能力分析、过程状态监控与稳健设计优化为主线开展研究。系统开展并完成了非正态分布过程能力指数建立、过程能力指数区间估计、监控事件大小的离散控制图设计、非正态计量控制图设计和分析、稳健参数与容差并行优化设计等五方面研究，并突破了基于广义信仰推断的过程能力指数区间估计、基于平均事件间隔事件的无偏控制图设计与基于可靠性的产品质量特征容差设计方法等三个关键科学问题。本项目的研究成果已成功应用于航空、航天、电子、核工业等领域典型装备研制过程中，包括某型远程战略导弹用电液伺服阀滑阀偶件的径向加工工艺、电液伺服机构小球的氮化工艺、雷达铁氧体移相器的镀膜工艺、某型舰载机中央翼板铝合金薄壁件的表面切削工艺、高功率固体激光器针翅肋片热沉性能的工艺优化设计等，并取得了显著的工程经济效益。在上述研究基础上，出版了著作《基于状态特征的质量控制方法》，主持起草完成国家标准《GB/T 17989.4-2020 控制图第4部分：累积和控制图》与《过程管理中的统计方法—能力与性能 第4部分：过程能力估计和性能测量》，并在IJPR、JMS、CIE、RESS等国际权威期刊发表高水平学术论文47篇，授权国家专利9项，受理国家发明专利4项。

内容获取失败，请点击重试