制造过程失控趋势模式识别和变点估计研究及应用
发布时间：2018-07-26 19:23
【摘要】：如何对制造现场获取的统计数据进行分析，方法之一是对制造过程中的质量特征值进行统计过程控制。统计过程控制就是对制造过程数据的收集、整理、分析，通过客观定量的方法来分析产品制造过程中的质量问题，谋求以较少的资源消耗来提高产品质量，是质量管理中最有用的工具。在现实的生产过程中，失控的产生是不可避免的，同时造成失控的原因十分复杂，统计过程控制可以找出制造过程的部分确定性，尽可能保证加工过程的质量特征处于一个可以接受的水平之上。当前，在复杂制造环境下，由于生产过程高度自动化以及产品的复杂性，质量问题的来源更加广泛，使要解决如何找出制造过程中的失控原因的问题更加迫切。如何在计算机集成制造的背景下，应用统计过程控制自动识别制造过程失控趋势模式，为用户提供高效率、高精度的过程受控、失控的信息，并且给出失控原因及针对失控原因的解决办法是当前研究的一个热点。 本文从制造过程质量稳定性出发，以统计过程控制为背景，探讨制造过程中引起失控的各种系统性因素，并结合发动机缸体的生产实际，对其制造过程统计数据进行模式识别和变点估计，实现了失控过程的真正预测，为提高制造过程质量稳定性提供理论依据与技术支持。 论文首先总结了制造过程失控趋势模式识别及变点估计的经典理论，建立了失控趋势分析的理论体系，提出在统计过程中对失控趋势采用模糊神经网络进行模式识别，采用模糊聚类分析来进行变点估计，以有效的提高质量稳定性。其次建立了控制图特征的提取方法，设计基于特征的神经网络模式识别器，通过对特征的定义完成了由样本函数进行的特征提取，根据不同过程模式的特征，以自动的识别出六种失控模式。然后基于模糊聚类理论以及统计方法，，提出一种新的模糊统计聚类方法来处理实际中的变点问题，并将该方法应用于不同类型控制图中，结果证明，本文所提出的该方法无论在固定抽样策略还是可变抽样策略下，对于控制图的变点估计都有着良好的效果。并且将基于特征的失控趋势模式识别和基于模糊聚类的变点估计应用于某公司缸体加工过程的质量控制中，结果表明该方法能很好地预测失控过程。最后基于IDEF和UML开发了面向制造过程的失控趋势分析系统OCRS，并将所建立的产品质量监控模型、数据采集等相关技术无缝集成，为提高制造过程质量稳定性提供技术保障。
[Abstract]:One of the methods to analyze the statistical data obtained from the manufacturing field is to control the quality characteristic values in the manufacturing process. Statistical process control is the collection, collation and analysis of manufacturing process data, and through objective and quantitative methods to analyze the quality problems in the process of manufacturing products, in order to improve the quality of products with less consumption of resources. Is the most useful tool in quality management. In the actual production process, the production of runaway is inevitable, at the same time, the cause of runaway is very complex, statistical process control can find out some certainty of the manufacturing process. Ensure as much as possible that the quality characteristics of the process are above an acceptable level. At present, in the complex manufacturing environment, because of the high automation of the production process and the complexity of the product, the source of the quality problem is more extensive, so it is more urgent to solve the problem of how to find out the cause of the runaway in the manufacturing process. Under the background of computer integrated manufacturing, the statistical process control is applied to identify the trend pattern of manufacturing process runaway automatically, which can provide users with high efficiency, high precision information of process control and out of control. It is a hot topic to give out the cause of runaway and to solve the problem. Starting from the quality stability of manufacturing process and taking statistical process control as the background, this paper discusses various systematic factors that cause runaway in manufacturing process, and combines with the production practice of engine cylinder block. The statistical data of manufacturing process are used for pattern recognition and change point estimation to realize the real prediction of runaway process and to provide theoretical basis and technical support for improving the quality stability of manufacturing process. Firstly, this paper summarizes the classical theory of pattern recognition and change point estimation of runaway trend in manufacturing process, establishes the theoretical system of runaway trend analysis, and puts forward that fuzzy neural network is used for pattern recognition of runaway trend in statistical process. Fuzzy clustering analysis is used to estimate the variation points to improve the quality stability. Secondly, the feature extraction method of control chart is established, and the neural network pattern recognizer based on feature is designed. The feature extraction by sample function is completed by defining the feature, and according to the feature of different process pattern, Automatically identify six out of control modes. Then, based on fuzzy clustering theory and statistical method, a new fuzzy statistical clustering method is proposed to deal with the problem of change points in practice, and the method is applied to different types of control graphs. The method proposed in this paper has a good effect on the change point estimation of the control chart both under the fixed sampling strategy and the variable sampling strategy. The feature based pattern recognition of runaway trend and the change point estimation based on fuzzy clustering are applied to the quality control of cylinder block machining process in a company. The results show that the method can predict the runaway process well. Finally, based on IDEF and UML, a trend analysis system for manufacturing process is developed, and the established product quality monitoring model, data acquisition and other related technologies are seamlessly integrated to provide technical guarantee for improving the quality stability of manufacturing process.
【学位授予单位】：合肥工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP391.4;TB497
本文编号：2147044
[Abstract]:One of the methods to analyze the statistical data obtained from the manufacturing field is to control the quality characteristic values in the manufacturing process. Statistical process control is the collection, collation and analysis of manufacturing process data, and through objective and quantitative methods to analyze the quality problems in the process of manufacturing products, in order to improve the quality of products with less consumption of resources. Is the most useful tool in quality management. In the actual production process, the production of runaway is inevitable, at the same time, the cause of runaway is very complex, statistical process control can find out some certainty of the manufacturing process. Ensure as much as possible that the quality characteristics of the process are above an acceptable level. At present, in the complex manufacturing environment, because of the high automation of the production process and the complexity of the product, the source of the quality problem is more extensive, so it is more urgent to solve the problem of how to find out the cause of the runaway in the manufacturing process. Under the background of computer integrated manufacturing, the statistical process control is applied to identify the trend pattern of manufacturing process runaway automatically, which can provide users with high efficiency, high precision information of process control and out of control. It is a hot topic to give out the cause of runaway and to solve the problem. Starting from the quality stability of manufacturing process and taking statistical process control as the background, this paper discusses various systematic factors that cause runaway in manufacturing process, and combines with the production practice of engine cylinder block. The statistical data of manufacturing process are used for pattern recognition and change point estimation to realize the real prediction of runaway process and to provide theoretical basis and technical support for improving the quality stability of manufacturing process. Firstly, this paper summarizes the classical theory of pattern recognition and change point estimation of runaway trend in manufacturing process, establishes the theoretical system of runaway trend analysis, and puts forward that fuzzy neural network is used for pattern recognition of runaway trend in statistical process. Fuzzy clustering analysis is used to estimate the variation points to improve the quality stability. Secondly, the feature extraction method of control chart is established, and the neural network pattern recognizer based on feature is designed. The feature extraction by sample function is completed by defining the feature, and according to the feature of different process pattern, Automatically identify six out of control modes. Then, based on fuzzy clustering theory and statistical method, a new fuzzy statistical clustering method is proposed to deal with the problem of change points in practice, and the method is applied to different types of control graphs. The method proposed in this paper has a good effect on the change point estimation of the control chart both under the fixed sampling strategy and the variable sampling strategy. The feature based pattern recognition of runaway trend and the change point estimation based on fuzzy clustering are applied to the quality control of cylinder block machining process in a company. The results show that the method can predict the runaway process well. Finally, based on IDEF and UML, a trend analysis system for manufacturing process is developed, and the established product quality monitoring model, data acquisition and other related technologies are seamlessly integrated to provide technical guarantee for improving the quality stability of manufacturing process.
【学位授予单位】：合肥工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP391.4;TB497
【参考文献】
相关期刊论文 前1条
1 陈希孺;变点统计分析简介[J];数理统计与管理;1991年01期
相关博士学位论文 前4条
1 程志强;基于智能方法的产品制造过程质量诊断[D];南京理工大学;2011年
2 吕泽华;模糊集理论的新拓展及其应用研究[D];华中科技大学;2007年
3 余建波;基于智能学习模型的制造过程质量控制理论与方法论研究[D];上海交通大学;2009年
4 施建中;基于模糊聚类的非线性系统辨识研究[D];华北电力大学;2012年
本文编号：2147044
本文链接：https://www.wllwen.com/guanlilunwen/gongchengguanli/2147044.html
