多等级批质量计数抽样检验理论与方法研究

The determination of the lot quality’s classification has been proven to have important applications for food classification management, inspection and quarantine in fruits for export and import as well as epidemic control. At present, the majority of the researches and applications of sampling test assume that the lot quality only has two levels (acceptance or rejection), and most of them mainly focus on the case of product quality with a single index. The research on the lot quality with multiple classifications, especially the case of multiple product quality indexes, has not been carried out. . This project will consider the lot quality determination as the objective, aiming at Sampling Inspection by Attributes, which has the features of low cost and high efficiency, introduce the run and frequency elements and establish the models of Sampling Inspections by Attributes for Lot Quality with Multiple Classifications. Furthermore, the performance analysis of these schemes can be made by applying the Finite Markov Chain Embedding Approach and optimization design of these schemes can also be achieved considering the average sample size, the risk of misjudging the lot quality classification as well as the loss from it. Additionally, based on data which are obtained from the scheme implementation, the estimation of the lot quality distribution can be gained. Besides, the research can make effective use of historical data and gain a more accurate conclusion by using the Bayesian Method, which is introduced in all phases of schemes including performance analysis, optimization design as well as the estimation of lot quality classification. This project will provide a complete theory and method system for the development and implementation of multiple classification lot quality sampling inspection plans by attributes with optimal parameter combination.

批质量所处等级判定，对食品分级管理、进出口水果检验检疫、流行病控制等具有重要的应用价值。目前大部分抽样检验的研究和应用，均以批质量分为两等级（接收和拒收）为假设，并且研究主要集中在产品质量具有单一指标的情形。对于批质量为多等级，尤其是产品质量指标多元化的情形尚未开展相关研究。. 本项目将针对具有低成本、高效率特点的计数抽样检验，以批质量等级判定为目标，引入游程和频数等要素，构建多等级批质量计数抽样检验方案模型；应用有限马尔可夫链嵌入法进行方案性能分析；综合考虑平均样本量、错判批质量等级风险，以及由此带来的损失等对方案进行优化设计；基于方案实施获得的数据对批质量估计。此外，在方案的性能分析、优化设计、批质量估计的各个阶段全面引入贝叶斯方法，有效利用历史信息，所得结论更加准确。本项目将为具有最佳参数组合的多等级批质量计数抽样检验方案的制定和实施，提供完整的理论和方法体系。

批质量所处等级判定在食品分级管理、进出口水果检验检疫、流行病控制、可靠性工程等方面都有重要的应用价值。本项目按照申报书中的研究计划，构建了不同的多等级批质量抽样检验方案，针对每一种抽样检验方案，完成了相关抽样方案的模型构建、概率指标推导、方案性能分析及优化设计等问题的研究，并将其应用于具体的可靠性工程问题。基于不同的工程应用背景，本项目主要完成了以下3项研究内容：抽样检验在可靠性试验中的应用；抽样检验在可靠性冲击模型中的应用；抽样检验在可靠性维修优化中的应用。.在可靠性试验中，主要研究了基于可靠性试验的多属性产品多等级批的计数一次和序贯抽样检验方案，以及基于启动验证试验的多等级抽样检验方案，该项研究成果可以为可靠性试验及启动验证试验节约检测成本并提高检测结果的准确性；在可靠性冲击模型中，通过引入多等级、多阶段等概念，定义新的游程概念，联系抽样检验的停止规则和判定准则，设计新的抽样检验方案，通过结合可靠性工程背景，该项研究成果可以应用于可靠性系统建模，提高系统可靠度；在可靠性维修优化中，以电动汽车和电力系统为背景，构建具体的优化模型，并运用坐标循环算法、分支定界法、现代智能算法等对优化模型进行求解。.本项目为具有最佳参数组合的多等级批质量计数抽样检验方案的制定和实施，提供完整的理论和方法体系。