基于不同质量特性指标的抽检方案研究
发布时间：2018-09-07 22:17
【摘要】：产品的好质量可以给企业在全球市场的竞争中带来良好的信誉.为了保证所生产的产品的质量,企业需要对原材料、零部件以及产品进行抽样检验.但在实际制造过程中,不同的原材料、零部件或产品往往有不同的质量特性,如何根据各自的质量特性制定合适的抽样检验方案,是一个非常实用而又迫切需要解决的问题.最近Sherman提出的重复分组抽检方案因所需的抽样量更少、节约检验成本和时间而倍受国内外学者的关注.本文分别以产品的寿命性能指数、变差系数和可靠寿命为质量指标对不同寿命分布模型制定出相应的抽样检验方案.具体工作如下：1.对于望大型特性值的产品,可以用寿命性能指数来评估产品的寿命性能.本文首次将寿命性能指数作为质量参数应用到质量抽样检验中.由于寿命性能指数与不合格率(或合格率)之间存在紧密的关系,因此可以通过控制产品的寿命性能指数来控制产品的不合格品率.文中首先推导出定数截尾样本下指数分布产品的寿命性能指数的估计量,进而提出基于寿命性能指数的一次抽样检验方案,并求得方案的抽样量和接受系数.其次结合重复分组抽样的思想,分别对瑞丽分布和Weibull分布的产品提出基于寿命性能指数的重复分组抽样方案,并给出不同质量水平和风险水平下的方案参数值,以便使用方可以据此做出接收或拒收一批产品的判断,生产方则可据此控制产品的不合格率,达到提高产品质量的目的.2.变差系数是统计质量管理中的一个很重要指标,可以用来检查产品批质量的稳定性等.当正态分布的均值和方差均未知时,变差系数的估计量服从非中心t分布,本文采用变差系数为质量参数,以变差系数的估计量为检验统计量,结合两阶段抽样、重复分组抽样、多个相依状态抽样以及多状态重复分组抽样等方法对正态分布产品制定出相应的抽样检验方案.为了降低检验成本,通常希望需要的抽样量最小.因此在求解方案参数值时,选取平均抽样量的最小值作为目标函数,以保证所制定的抽样方案是经济有效的.最后基于OC曲线和抽样量两个准则对提出的抽样方案进行比较分析,结果表明所提出的四种抽样方案均优于佟毅提出的一次抽样检验方案,而多状态重复分组抽样方案所需抽样量最少.多个相依状态抽样方案和多状态重复分组抽样方案不仅考虑当前批产品的样本信息,还考虑前面批次的样本结果,非常适用于连续批产品的检验.3.对于对称型寿命分布产品,可靠性抽样方案通常采用平均寿命作为寿命指标.而大部分寿命分布并非对称型,用平均寿命无法充分描述分布的中心趋势.因此,本文采用可靠寿命作为产品的寿命指标,对Lomax分布产品在截尾情形下制定出抽检方案.首先假设形状参数已知、产品批量很大,在消费者风险给定的情形下确定方案所需的最小样本量.其次,对方案的OC值和生产方风险进行讨论,得到可靠寿命的真实值与期望值的比值同生产方风险之间的关系；最后通过实例说明所提抽检方案是简单易行的.4.当受控过程的均值发生波动时,过程方差是均值的线性函数,即过程变差系数为常数,此时可用EWMA-CV控制图来监控过程的变化.而重复抽样的方法可以减少抽样量,节约抽样的成本和时间.因此本文将重复抽样的思想与EWMA-CV控制图相结合,利用变差系数的估计量构造EWMA统计量,由此提出一种具有内外双重控制限的控制图来监控过程的变差系数.当过程发生较小偏移时新控制图的平均运行链长变短,而且漂移越大平均运行链长越短,从而说明所提出的控制图对大漂移或小漂移都有很好的检出效果.
[Abstract]:In order to ensure the quality of the products produced, enterprises need to sample the raw materials, parts and products. But in the actual manufacturing process, different raw materials, parts or products often have different quality characteristics, how to according to each Recently Sherman's repeated grouping sampling scheme has attracted much attention from scholars both at home and abroad because it requires less sampling and saves inspection cost and time. The specific work is as follows: 1. Life performance index can be used to evaluate the life performance of products with large characteristic values. This paper applies life performance index as a quality parameter to quality sampling inspection for the first time. There is a close relationship between the life performance index and the unqualified rate (or the qualified rate), so the unqualified rate of the product can be controlled by controlling the life performance index of the product. Secondly, combined with the idea of repeated grouping sampling, the repeated grouping sampling scheme based on life performance index for products with Rayleigh distribution and Weibull distribution is proposed, and the scheme parameters under different quality and risk levels are given, so that the scheme can be used accordingly. 2. Variation coefficient is an important index in statistical quality management, which can be used to check the stability of product batch quality. When the mean and variance of normal distribution are unknown, variation coefficient The estimator obeys the non-central t-distribution. In this paper, the variance coefficient is used as the quality parameter, and the variance coefficient estimator is used as the test statistic. In order to ensure that the sampling scheme is economical and effective, the minimum of the average sampling quantity is chosen as the objective function in solving the parameter value of the scheme. Finally, the proposed sampling scheme is compared and analyzed based on OC curve and sampling quantity, and the results show that the proposed four sampling schemes are effective. All the sampling schemes are superior to the one-time sampling scheme proposed by Tong Yi, and the multi-state repeated grouping sampling scheme requires the least sampling amount. 3. For symmetrical life distribution products, the average life is usually used as the life index in reliability sampling scheme. However, most of the life distribution is not symmetrical, so the average life can not fully describe the central trend of distribution. Firstly, assuming that the shape parameters are known and the batch size is large, the minimum sample size required for the scheme is determined under the given consumer risk. Secondly, the OC value of the scheme and the risk of the producer are discussed, and the relationship between the ratio of the true value of the reliable life to the expected value and the risk of the producer is obtained. An example shows that the proposed sampling scheme is simple and feasible. 4. When the mean value of the controlled process fluctuates, the process variance is a linear function of the mean value, i.e. the process variation coefficient is constant. In this case, the EWMA-CV control chart can be used to monitor the change of the process. In this paper, the idea of repetitive sampling is combined with EWMA-CV control chart to construct EWMA statistics by estimating the variation coefficients. A control chart with internal and external double control limits is proposed to monitor the variation coefficients of the process. The average running chain length of the new control chart becomes shorter and the drift becomes larger and the average running is larger. The shorter the chain length, the better the detection effect of the proposed control chart is for large drift or small drift.
【学位授予单位】：西安电子科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：O212.2
本文编号：2229570
[Abstract]:In order to ensure the quality of the products produced, enterprises need to sample the raw materials, parts and products. But in the actual manufacturing process, different raw materials, parts or products often have different quality characteristics, how to according to each Recently Sherman's repeated grouping sampling scheme has attracted much attention from scholars both at home and abroad because it requires less sampling and saves inspection cost and time. The specific work is as follows: 1. Life performance index can be used to evaluate the life performance of products with large characteristic values. This paper applies life performance index as a quality parameter to quality sampling inspection for the first time. There is a close relationship between the life performance index and the unqualified rate (or the qualified rate), so the unqualified rate of the product can be controlled by controlling the life performance index of the product. Secondly, combined with the idea of repeated grouping sampling, the repeated grouping sampling scheme based on life performance index for products with Rayleigh distribution and Weibull distribution is proposed, and the scheme parameters under different quality and risk levels are given, so that the scheme can be used accordingly. 2. Variation coefficient is an important index in statistical quality management, which can be used to check the stability of product batch quality. When the mean and variance of normal distribution are unknown, variation coefficient The estimator obeys the non-central t-distribution. In this paper, the variance coefficient is used as the quality parameter, and the variance coefficient estimator is used as the test statistic. In order to ensure that the sampling scheme is economical and effective, the minimum of the average sampling quantity is chosen as the objective function in solving the parameter value of the scheme. Finally, the proposed sampling scheme is compared and analyzed based on OC curve and sampling quantity, and the results show that the proposed four sampling schemes are effective. All the sampling schemes are superior to the one-time sampling scheme proposed by Tong Yi, and the multi-state repeated grouping sampling scheme requires the least sampling amount. 3. For symmetrical life distribution products, the average life is usually used as the life index in reliability sampling scheme. However, most of the life distribution is not symmetrical, so the average life can not fully describe the central trend of distribution. Firstly, assuming that the shape parameters are known and the batch size is large, the minimum sample size required for the scheme is determined under the given consumer risk. Secondly, the OC value of the scheme and the risk of the producer are discussed, and the relationship between the ratio of the true value of the reliable life to the expected value and the risk of the producer is obtained. An example shows that the proposed sampling scheme is simple and feasible. 4. When the mean value of the controlled process fluctuates, the process variance is a linear function of the mean value, i.e. the process variation coefficient is constant. In this case, the EWMA-CV control chart can be used to monitor the change of the process. In this paper, the idea of repetitive sampling is combined with EWMA-CV control chart to construct EWMA statistics by estimating the variation coefficients. A control chart with internal and external double control limits is proposed to monitor the variation coefficients of the process. The average running chain length of the new control chart becomes shorter and the drift becomes larger and the average running is larger. The shorter the chain length, the better the detection effect of the proposed control chart is for large drift or small drift.
【学位授予单位】：西安电子科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：O212.2
【参考文献】
相关期刊论文 前1条
1 林金官;对数正态分布和Weibull分布变差系数的抽样验证方案[J];工科数学;1998年03期
，本文编号：2229570
本文链接：https://www.wllwen.com/jingjilunwen/jiliangjingjilunwen/2229570.html
