QQ客服 官方微博 反馈留言

题名	基于支持向量机的多元过程质量诊断研究与应用
作者	蔡亚军
学位类别	硕士
答辩日期	2017-05-24
授予单位	中国科学院沈阳自动化研究所
授予地点	沈阳
导师	陈书宏
关键词	多元统计过程控制 支持向量机 异常诊断 主元分析 质量管理系统
其他题名	Research and application of multi-process quality diagnosis based on support vector machine
学位专业	机械电子工程
中文摘要	本文以华晨动力机械有限公司的某系列变速器装配线项目为依托，对基于 SVM 的多元过程质量监控与异常诊断方法进行了研究。主要内容如下： (1) 结合课题的研究背景和意义，简述了故障诊断技术的发展，对常用的多元质量诊断方法做了介绍，并指出存在的问题。提出了基于支持向量机的多元过程质量诊断，为多元质量诊断提供了新的思路。 (2) 介绍了主成分分析法及多元控制图的基本理论，并详细阐述建立多元过程监控的步骤与方法。 (3) 研究了基于支持向量机模型的多元过程异常在线监控与诊断方法。针对传统的多变量质量控制图可以检测失控事件，但不直接确定哪个变量或变量组导致失控信号的问题，本文提出一种基于支持向量机的模型对多变量过程中的均值阶跃异常进行在线监控与诊断。首先利用 T2 控制图良好的监控性能对生产过程进行监控，判断生产过程是否异常，当生产过程出现异常时，再用改进的 SVM 模型对均值的偏移情况进行识别，实现对生产过程监控和异常源识别的目的。为降低数据噪声和提高聚类，本文利用主元分析法（Principal component analysis，PCA）对数据进行预处理，有效提取数据特征信息；为提高模型分类准确率与效率，本文用改进的网格搜索法（Grid search，GS）对SVM参数进行寻优。通过对比准确率、效率与平均运行长度，证明所提出的模型在识别异常源上，与一般的模型相比具有优越性。 (4) 在变速器装配过程中，有许多多元过程的情况，而且这多个变量之间存在着互相关性。比如，压装工位，需要同时对压装的压力和位移进行控制，确保这两个变量在合格的范围；拧紧工位也需要对转矩和位移进行控制。针对华晨项目需要对多元生产过程进行监控并识别异常源的要求，基于本文研究成果，设计和开发了多元过程质量管理系统。该质量管理系统可以对多元质量过程实时监控与诊断，有效减少了故障率，降低了生产成本。
英文摘要	Based on a series of transmission line project of Brilliance Power Machinery Co., Ltd., this thesis studies the multi-process quality monitoring and anomaly diagnosis method based on SVM. The main contents are as follows: (1) Based on the research background and significance of the research, the development of fault diagnosis technology is briefly introduced, and the commonly used multi-quality diagnosis method is introduced, and the existing problems are pointed out. A multi - process quality diagnosis based on support vector machine is proposed, which provides a new idea for multivariate quality diagnosis. (2) Introduce the basic theory of principal component analysis and multiple control chart, and elaborate the steps and methods of establishing multi-process monitoring. (3) The method of online monitoring and diagnosis of multiple process anomalies based on support vector machine model is studied. In this thesis, a model based on support vector machine (SVM) is proposed to analyze the out-of-control events in the multivariable process. In this thesis, we propose a model based on support vector machine (SVM) Monitoring and diagnosis. First of all, use the T2 control chart to monitor the performance of the production process to monitor and determine whether the production process is abnormal, when the production process is abnormal, and then improve the SVM model to identify the deviation of the mean to achieve the production process monitoring and abnormal Source identification purposes. In order to reduce the data noise and improve the clustering, this thesis preprocesses the data by principal component analysis (PCA), and effectively extracts the data feature information. In order to improve the accuracy and efficiency of the classification, this paper uses Grid search (GS) to optimize the SVM parameters. By comparing the accuracy, efficiency and average run length, it is proved that the proposed model is superior to the general model in identifying the abnormal source. (4) In the transmission assembly process, there are many multi-process situation, and there are many variables between the cross-correlation. For example, the press-fit station needs to control the pressure and displacement of the pressure-fit to ensure that the two variables are in the qualified range; the tightening station also needs to control the torque and displacement. Based on the above research results, we design and develop a multi-process quality management system for the Brilliance project to monitor the multiple production processes and identify the abnormal sources. The quality management system can be real-time monitoring and diagnosis of multiple quality processes, effectively reducing the failure rate and reducing the production cost.
语种	中文
产权排序	1
内容类型	学位论文
源URL	[http://ir.sia.cn/handle/173321/20516]
专题	沈阳自动化研究所_智能检测与装备研究室
推荐引用方式 GB/T 7714	蔡亚军. 基于支持向量机的多元过程质量诊断研究与应用[D]. 沈阳. 中国科学院沈阳自动化研究所. 2017.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。