面向检测实验室的测量系统分析理论与应用研究
|
摘要
在开放的市场中,商品的质量需要给予验证。这就需要各类有资格的检测实验室为其服务,在国家或行业法律、法规和规章的框架内开展各类检测活动,出具检测报告。检测报告的准确性和公正性对于社会来说至关重要,而检测结果的有效性在很大程度上取决于测量系统的质量水平。因此综合运用各种管理手段和统计方法以确保出具高质量(准确、可靠、及时)的检测数据和报告,成为检测实验室的核心任务。测量系统分析理论着重研究测量系统准确性和可靠性,将理论成果应用于检测实验室,对提高检测实验室的检测能力和公信力以及评判的科学性具有非常重要的意义。本文基于测量系统分析理论研究,结合检测实验室工作性质、特点与检测数据,将测量系统分析的理论成果应用于检测实验室内的不同测量系统中,所做的主要工作如下：
1、通过分析比较,指出检测实验室现有质量保证手段的不足,提出将测量系统分析理论应用于检测实验室的构想；在深入分析测量系统分析理论的基础上,结合检测实验室的工作性质和特点,将测量系统分析分为定性和定量两种类型进行分析。
2、定量型测量系统R&R分析的研究与改进：介绍了三种常见方法：极差法、均值极差(?)-R法和方差分析ANOVA法；总结与比较这三种方法各自的优缺点；对(?)-R法进行了改进,提高了该方法的分析精度；提出根据检测项目的分布情况结合操作者的能力选用不同的R&R方法。
3、定量型测量系统稳定性分析方法的改进：通过自制的样品或直接使用现有的不同规格的标准样品代替原来的样品采集方法,以降低实验室购置与储存大量标准样品的费用；针对上述两种样品替代法给出测量数据的统一化处理方法,并通过t检验与折线图结合实例验证处理方法的合理性；选用单值-极差控制图代替常用的(?)-R控制图,并加入EWMA(指数加权滑动平均)统计量与控制线形成一种综合型控制图,结合实例对比显示出综合型控制图的优越性。
4、定性型测量系统理论分析：针对目前所采用一致性分析方法的不足,试将广泛应用于医学统计的ROC曲线法应用于检测实验室：建立相应的ROC模型,选用次序logistic函数作为ROC模型中的关联函数,采用广义估计方程估算模型中的参数：最后通过实例验证了其有效性。
Quality of product is the attentive focus in the open market. It's necessary to verify the product quality, which requiies different kinds of qualified test laboratories to service it. The task of the test laboratories are types of testing activities within the framework of various national or industry laws, rules and regulations, and provide test reports. The accuracy and impartiality of the reports are vital for the whole society, while the success of test results is determined by the measurement data quality. Therefore, through a variety of management tools and statistical methods to ensure the high-quality (accurate, reliable, in time) test data and reports are the core mission and ultimate goal for test laboratories. Measurement System Analysis (MSA) is a very effective method to evaluate the measurement system. Therefore, it has great significance to research how to apply the achievement of the theory to the detection laboratories. Based on the principle of the theory of MSA, and combined with the inherent characteristics of test laboratories, this paper studies on how to apply the theory of MSA to different measurement systems in the test laboratories, the main tasks are as follows:
1. Through working in the test laboratories to learn the content of the work, and to find the lackages of current quality assurance tools. Propose the idea of applying the MSA theory to the test laboratories. Based on the thorough analysis of MSA, combined with the common points in the work content of the test laboratory, the measurement systems is divided into two kinds of types:the quantitative type and the qualitative type;
2. For quantitative type of MSA, the paper studies the several evaluation methods in R & R analysis which are Range method, X--R method and ANOVA variance analysis.By summarizing the advantages and disadvantages of these three methods, the paper propose the improvement measures.It also proposes that the test laboratories could choose the suitable R & R analysis methods on the basis of the distribution of test projects;
3. In the analysis phase of measurement system's stability, using self-made samples instead of the standard samples or using the existing different specifications standard samples to replace. Universalize the data. Verify the unification's rationality through the T test and the line chart. According to the characteristics of test laboratories, X-Rs control charts is selected to achieve stability analysis and EWMA is superimposed on I control chart to achieve combined control, which can obviously enhance the small-offset control to the stability of measurement system;
4. Considering the deficiency of existing analysis methods of consistency analysis in qualitative-type MSA, ROC curve method which is widely used in medical statistics is applied to the evaluation of measurement system. First, build a corresponding ROC model, the sequence logistic function is chosen as correlation function, and then use generalized estimating equation to estimate the model parameters.
1、通过分析比较,指出检测实验室现有质量保证手段的不足,提出将测量系统分析理论应用于检测实验室的构想；在深入分析测量系统分析理论的基础上,结合检测实验室的工作性质和特点,将测量系统分析分为定性和定量两种类型进行分析。
2、定量型测量系统R&R分析的研究与改进：介绍了三种常见方法：极差法、均值极差(?)-R法和方差分析ANOVA法；总结与比较这三种方法各自的优缺点；对(?)-R法进行了改进,提高了该方法的分析精度；提出根据检测项目的分布情况结合操作者的能力选用不同的R&R方法。
3、定量型测量系统稳定性分析方法的改进：通过自制的样品或直接使用现有的不同规格的标准样品代替原来的样品采集方法,以降低实验室购置与储存大量标准样品的费用；针对上述两种样品替代法给出测量数据的统一化处理方法,并通过t检验与折线图结合实例验证处理方法的合理性；选用单值-极差控制图代替常用的(?)-R控制图,并加入EWMA(指数加权滑动平均)统计量与控制线形成一种综合型控制图,结合实例对比显示出综合型控制图的优越性。
4、定性型测量系统理论分析：针对目前所采用一致性分析方法的不足,试将广泛应用于医学统计的ROC曲线法应用于检测实验室：建立相应的ROC模型,选用次序logistic函数作为ROC模型中的关联函数,采用广义估计方程估算模型中的参数：最后通过实例验证了其有效性。
Quality of product is the attentive focus in the open market. It's necessary to verify the product quality, which requiies different kinds of qualified test laboratories to service it. The task of the test laboratories are types of testing activities within the framework of various national or industry laws, rules and regulations, and provide test reports. The accuracy and impartiality of the reports are vital for the whole society, while the success of test results is determined by the measurement data quality. Therefore, through a variety of management tools and statistical methods to ensure the high-quality (accurate, reliable, in time) test data and reports are the core mission and ultimate goal for test laboratories. Measurement System Analysis (MSA) is a very effective method to evaluate the measurement system. Therefore, it has great significance to research how to apply the achievement of the theory to the detection laboratories. Based on the principle of the theory of MSA, and combined with the inherent characteristics of test laboratories, this paper studies on how to apply the theory of MSA to different measurement systems in the test laboratories, the main tasks are as follows:
1. Through working in the test laboratories to learn the content of the work, and to find the lackages of current quality assurance tools. Propose the idea of applying the MSA theory to the test laboratories. Based on the thorough analysis of MSA, combined with the common points in the work content of the test laboratory, the measurement systems is divided into two kinds of types:the quantitative type and the qualitative type;
2. For quantitative type of MSA, the paper studies the several evaluation methods in R & R analysis which are Range method, X--R method and ANOVA variance analysis.By summarizing the advantages and disadvantages of these three methods, the paper propose the improvement measures.It also proposes that the test laboratories could choose the suitable R & R analysis methods on the basis of the distribution of test projects;
3. In the analysis phase of measurement system's stability, using self-made samples instead of the standard samples or using the existing different specifications standard samples to replace. Universalize the data. Verify the unification's rationality through the T test and the line chart. According to the characteristics of test laboratories, X-Rs control charts is selected to achieve stability analysis and EWMA is superimposed on I control chart to achieve combined control, which can obviously enhance the small-offset control to the stability of measurement system;
4. Considering the deficiency of existing analysis methods of consistency analysis in qualitative-type MSA, ROC curve method which is widely used in medical statistics is applied to the evaluation of measurement system. First, build a corresponding ROC model, the sequence logistic function is chosen as correlation function, and then use generalized estimating equation to estimate the model parameters.
引文
[1]鄢国强,牛兴荣,张斌.检测实验室与实验室认可[M].理化检验-物理分册,2004,40(1)：49-63.
[2]Measurement management systems—Requirements for measurement processes and measuring equipment[S], ISO10012:2003.
[3]丛黎明,丁钢强,张双凤.ISO/IEC17025实验室管理体系应用指南[M].浙江大学出版社,2006：36-52.
[4]夏偕田,孟小平.检测实验室管理体系建立指南第二版[M].化学工业出版社,2007：15-16,27-32.
[5]王建跃,张立军.国家认可检测实验室规范化建设与研究[J].中国公共卫生管理,2002,18(3)：190-192.
[6]肖建平.浅谈国家实验室认可中的“期间核查”[J].广船科技,2004,4：39-41.
[7]赵颖.检测机构设备和标准物质核查问题的探讨[J].现代测量与实验室管理,2004,4：41-42.
[8]蒋晓光,林忠,陈兆君,张玉廷.实验室仪器设备的期间核查[J].科技资讯,2007,34：101.
[9]万颖,朱茂英.食品菌落总数测量不确定度评定[J].阜阳师范学院学报(自然科学版),2006,23,2：48-49.
[10]Eagle A. R. A Method for Handling Errors in Testing and Measuring [J]. Industrial Quality Control,1954,3:10-14.
[11]F. E. Grubbs. On Estimating Precision of Measuring Instruments and Product Variability [J]. Journal of the American Statistical Association,1998.43:53-56,
[12]Mandel J. Repeatability and Reproducibility [J]. Journal of Quality Technology, 1972,4(2):74-85.
[13]Jay L. Bucher, Rules, Keep it Internal [J], Quality Progress,2005,38(3):76-77.
[14]F. E. Grubbs. Error of Measurement Precision, Accuracy and the Statistical Comparison of Measuring Instruments [J]. Technometrics,1973,15(2):273-278.
[15]Donald S. Ermer, Appraiser Variation in Gage R&R Measurement Studies [J], Quality Progress,2006,39 (3):77-79.
[16]Wessel N. van Wieringen, Edwin R. van den Heuvel. A Comparison of Methods for the Evaluation of Binary Measurement Systems [J], Quality Engineering,2005,17(4):495-507.
[17]刘通.测量系统分析与测量不确定度分析比较[J].中国计量,2007,1：90-91.
[18]王珏.测量过程控制技术的工程应用研究[J].工业计量,2004,14(1)：13-16.
[19]张黎.测量系统能力分析与统计过程控制[J].航空制造技术,2005,(7)：86-88.
[20]李晓东.测量系统分析在完善测量管理体系中的应用研究[D].江苏：南京理工大学,2007.
[21]白旭.测量系统分析(MSA)在计量工作中的应用[J].计量与测试技术,2007,34(9)：58-59
[22]于荣波,付玉霞,尹燕功.用MSA方法校准激光测距仪[J].北京工业职业技术学院学报,2006,2(5)：102-105.
[23]刘智敏.实验室认可中的不确定度和统计分析[M].中国标准出版社,2007
[24]王建辉.实验室质量管理与资质认定[M].黄河水利出版社,2007：87-90.
[25]胡仪之.QS9000质量体系中的测量系统分析[J].中国计量,2006,8：24-27.
[26]王霄锋.QS9000参考手册学习与理解-测量系统分析[M].清华大学出版社,2004：1-2,41-50,105-106.
[27]唐中一,俞磊,倪伟.测量系统分析(MSA)及其软件设计[J].现代电子技术,2007,23：126-128.
[28]江晓.R&R法：一种科学的测量系统分析方法[J].浙江万里学院学报,2003,16(2)：126-129.
[29]谢少锌.测量系统及其分析方法综述与分析[J].电子质量,自动化与仪器仪表卷：18-20.
[30]王静龙,梁小筠.定性数据分析[M].华东师范大学出版社,2004：
[31]AIAG. Measurement Systems Analysis Reference Manual [M]. Chrysler. Ford. GM printed,1995.
[32]Chrysler. Ford. GM.质量体系要求QS9000(第三版)[S].Chrysler. Ford. GM printed,1998
[33]王霄锋,佘俊.均值-极差法分析评价人-零件间的交互作用[J].汽车工程,2004,26(1)：111-114.
[34]徐兰,韩之俊.测量系统性能评价的方差分析方法[J].统计与决策,2008,8：156-157.
[35]栾江.数据质量控制：数据预处理研究、设计与实现[D].四川：四川大学,2004：
[36]W. Linna Kenneth, H. Woodall William. Effect of measurement error on Shewhart control charts. Journal of Quality Technology [J],2001.33(2):213.
[37]Ming-Nan Chen JrJung Lyu. Gauge capability studies for attribute data. Quality and Reliability Engineering International [J],2008.24(1):71-82.
[38]Douglas C. Montgomery. Introduction to Statistical Quality Control.3rd ed[M].1996:John Wiley and Sons.
[39]孙静.接近零不合格过程的有效控制[M].清华大学出版社,2005.
[40]Jing Sun. Control Charts Based on the Number of Consecutive Conforming Items between Two Successive Nonconforming Items for the near Zero-Nonconforming Processes [J]. Total Quality Management.2006, Vol.11, pp.235-250.
[41]何桢,赵燕,邹峰.属性值测量系统分析的方法研究与应用[J].工业工程,2008,11(1)：82-86.
[42]J. Cohen, A coefficient of agreement for nominal scales [J], Ed uc. Psychol. Meas, 1960,20(1):37-46.
[43]翁颖.计数性量具的系统分析法[J].2006,(7)：29-32.
[44]施锡铨,范正琦.数据分析与统计建模[M].上海人民出版社,2007.
[45]J. L. Fleiss, Statistical Methods for Rates and Proportions,2nd edition [M], New York: John Wiley&Sons,1981.
[46]刘开珊.简介检验一致性的可靠实用方法[J].湖北预防医学杂志,1991,2(3)：38-43.
[47]赵庶娴,李效伟,刘东邰,鲁建.利用假设试验分析法评价计数型测量系统[J].中国计量,2008,3：91-93.
[48]张建国.ROC曲线分析的基本原理以及在体质与健康促进研究中的应用[J].体育科学,2008,28(6)：62-66.
[49]Goodenough DJ, Rossmann K, Lusted LB (1974) Radiographic applications of receiver operating characteristic (ROC) curves [J]. Radiology:110,89-95.
[50]张克坚,扬振华.应用ROC曲线图评价检验项目的临床准确性[J].江西医学检验,1996,17(2)：66-68.
[51]张克坚,扬振华.应用ROC曲线图评价检验项目的临床准确性(续完)[J].江西医学检验,1999,17(3)：189-191.
[52]李康,郭祖超,胡琳等.有序分类数据的累积比数模型分析方法[J].中国卫生统计,1993,10(4)：35-38.
[53]Vanagas G. Receiver operating characteristic curves and comparison of cardiac surgery risk stratification systems [J]. Interactive Cardiovascular and Thoracic Surgery:3,319-322.
[54]缪凡,童峰.复杂抽样数据的logistic回归分析方法及其应用[J].中国卫生统计,2008,25(6)：577-579.
[55]王洋,徐刚锋,左东广.基于多元Logistic回归分析的匹配算法性能评估[J].Infrared Technology,2008,30 (10):563-566.
[56]Nancy L. Leech, Karen C. Barrett, George A. Morgan. SPSS for Intermediate Statistics Use and Interpretation [M]. Publishing House of Electronics Industry,2009.
[57]林杰斌,林川雄,刘明德.SPSS12统计建模与应用实务[M].中国铁道出版社,2005.
[58]邹莉玲,沈其君,陈峰等.ROC曲线下面积的ML估计与假设检验[J].中国公共卫生,2003,19(1)：127-128.
[2]Measurement management systems—Requirements for measurement processes and measuring equipment[S], ISO10012:2003.
[3]丛黎明,丁钢强,张双凤.ISO/IEC17025实验室管理体系应用指南[M].浙江大学出版社,2006：36-52.
[4]夏偕田,孟小平.检测实验室管理体系建立指南第二版[M].化学工业出版社,2007：15-16,27-32.
[5]王建跃,张立军.国家认可检测实验室规范化建设与研究[J].中国公共卫生管理,2002,18(3)：190-192.
[6]肖建平.浅谈国家实验室认可中的“期间核查”[J].广船科技,2004,4：39-41.
[7]赵颖.检测机构设备和标准物质核查问题的探讨[J].现代测量与实验室管理,2004,4：41-42.
[8]蒋晓光,林忠,陈兆君,张玉廷.实验室仪器设备的期间核查[J].科技资讯,2007,34：101.
[9]万颖,朱茂英.食品菌落总数测量不确定度评定[J].阜阳师范学院学报(自然科学版),2006,23,2：48-49.
[10]Eagle A. R. A Method for Handling Errors in Testing and Measuring [J]. Industrial Quality Control,1954,3:10-14.
[11]F. E. Grubbs. On Estimating Precision of Measuring Instruments and Product Variability [J]. Journal of the American Statistical Association,1998.43:53-56,
[12]Mandel J. Repeatability and Reproducibility [J]. Journal of Quality Technology, 1972,4(2):74-85.
[13]Jay L. Bucher, Rules, Keep it Internal [J], Quality Progress,2005,38(3):76-77.
[14]F. E. Grubbs. Error of Measurement Precision, Accuracy and the Statistical Comparison of Measuring Instruments [J]. Technometrics,1973,15(2):273-278.
[15]Donald S. Ermer, Appraiser Variation in Gage R&R Measurement Studies [J], Quality Progress,2006,39 (3):77-79.
[16]Wessel N. van Wieringen, Edwin R. van den Heuvel. A Comparison of Methods for the Evaluation of Binary Measurement Systems [J], Quality Engineering,2005,17(4):495-507.
[17]刘通.测量系统分析与测量不确定度分析比较[J].中国计量,2007,1：90-91.
[18]王珏.测量过程控制技术的工程应用研究[J].工业计量,2004,14(1)：13-16.
[19]张黎.测量系统能力分析与统计过程控制[J].航空制造技术,2005,(7)：86-88.
[20]李晓东.测量系统分析在完善测量管理体系中的应用研究[D].江苏：南京理工大学,2007.
[21]白旭.测量系统分析(MSA)在计量工作中的应用[J].计量与测试技术,2007,34(9)：58-59
[22]于荣波,付玉霞,尹燕功.用MSA方法校准激光测距仪[J].北京工业职业技术学院学报,2006,2(5)：102-105.
[23]刘智敏.实验室认可中的不确定度和统计分析[M].中国标准出版社,2007
[24]王建辉.实验室质量管理与资质认定[M].黄河水利出版社,2007：87-90.
[25]胡仪之.QS9000质量体系中的测量系统分析[J].中国计量,2006,8：24-27.
[26]王霄锋.QS9000参考手册学习与理解-测量系统分析[M].清华大学出版社,2004：1-2,41-50,105-106.
[27]唐中一,俞磊,倪伟.测量系统分析(MSA)及其软件设计[J].现代电子技术,2007,23：126-128.
[28]江晓.R&R法：一种科学的测量系统分析方法[J].浙江万里学院学报,2003,16(2)：126-129.
[29]谢少锌.测量系统及其分析方法综述与分析[J].电子质量,自动化与仪器仪表卷：18-20.
[30]王静龙,梁小筠.定性数据分析[M].华东师范大学出版社,2004：
[31]AIAG. Measurement Systems Analysis Reference Manual [M]. Chrysler. Ford. GM printed,1995.
[32]Chrysler. Ford. GM.质量体系要求QS9000(第三版)[S].Chrysler. Ford. GM printed,1998
[33]王霄锋,佘俊.均值-极差法分析评价人-零件间的交互作用[J].汽车工程,2004,26(1)：111-114.
[34]徐兰,韩之俊.测量系统性能评价的方差分析方法[J].统计与决策,2008,8：156-157.
[35]栾江.数据质量控制：数据预处理研究、设计与实现[D].四川：四川大学,2004：
[36]W. Linna Kenneth, H. Woodall William. Effect of measurement error on Shewhart control charts. Journal of Quality Technology [J],2001.33(2):213.
[37]Ming-Nan Chen JrJung Lyu. Gauge capability studies for attribute data. Quality and Reliability Engineering International [J],2008.24(1):71-82.
[38]Douglas C. Montgomery. Introduction to Statistical Quality Control.3rd ed[M].1996:John Wiley and Sons.
[39]孙静.接近零不合格过程的有效控制[M].清华大学出版社,2005.
[40]Jing Sun. Control Charts Based on the Number of Consecutive Conforming Items between Two Successive Nonconforming Items for the near Zero-Nonconforming Processes [J]. Total Quality Management.2006, Vol.11, pp.235-250.
[41]何桢,赵燕,邹峰.属性值测量系统分析的方法研究与应用[J].工业工程,2008,11(1)：82-86.
[42]J. Cohen, A coefficient of agreement for nominal scales [J], Ed uc. Psychol. Meas, 1960,20(1):37-46.
[43]翁颖.计数性量具的系统分析法[J].2006,(7)：29-32.
[44]施锡铨,范正琦.数据分析与统计建模[M].上海人民出版社,2007.
[45]J. L. Fleiss, Statistical Methods for Rates and Proportions,2nd edition [M], New York: John Wiley&Sons,1981.
[46]刘开珊.简介检验一致性的可靠实用方法[J].湖北预防医学杂志,1991,2(3)：38-43.
[47]赵庶娴,李效伟,刘东邰,鲁建.利用假设试验分析法评价计数型测量系统[J].中国计量,2008,3：91-93.
[48]张建国.ROC曲线分析的基本原理以及在体质与健康促进研究中的应用[J].体育科学,2008,28(6)：62-66.
[49]Goodenough DJ, Rossmann K, Lusted LB (1974) Radiographic applications of receiver operating characteristic (ROC) curves [J]. Radiology:110,89-95.
[50]张克坚,扬振华.应用ROC曲线图评价检验项目的临床准确性[J].江西医学检验,1996,17(2)：66-68.
[51]张克坚,扬振华.应用ROC曲线图评价检验项目的临床准确性(续完)[J].江西医学检验,1999,17(3)：189-191.
[52]李康,郭祖超,胡琳等.有序分类数据的累积比数模型分析方法[J].中国卫生统计,1993,10(4)：35-38.
[53]Vanagas G. Receiver operating characteristic curves and comparison of cardiac surgery risk stratification systems [J]. Interactive Cardiovascular and Thoracic Surgery:3,319-322.
[54]缪凡,童峰.复杂抽样数据的logistic回归分析方法及其应用[J].中国卫生统计,2008,25(6)：577-579.
[55]王洋,徐刚锋,左东广.基于多元Logistic回归分析的匹配算法性能评估[J].Infrared Technology,2008,30 (10):563-566.
[56]Nancy L. Leech, Karen C. Barrett, George A. Morgan. SPSS for Intermediate Statistics Use and Interpretation [M]. Publishing House of Electronics Industry,2009.
[57]林杰斌,林川雄,刘明德.SPSS12统计建模与应用实务[M].中国铁道出版社,2005.
[58]邹莉玲,沈其君,陈峰等.ROC曲线下面积的ML估计与假设检验[J].中国公共卫生,2003,19(1)：127-128.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |