现代不确定度评定方法及应用

发布时间：2018-03-16 10:32

本文选题：现代不确定度　切入点：贝叶斯方法　出处：《合肥工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：不确定度作为表征测量结果的重要参数指标,越来越受到社会各个领域的重视。随着时代的不断进步、科技的高速发展以及人们对产品质量的高度重视,现代不确定度评定方法和理论应运而生。研究现代不确定度评定方法,对完善现代不确定度理论,促进现代不确定度的广泛应用具有重要意义。分析现代不确定度理论的前提下,对贝叶斯不确定度评定方法进行了系统研究;基于共轭贝叶斯原理,提出融合历史信息和当前样本信息的不确定度分量实时、连续更新方法;针对无信息、共轭先验贝叶斯方法的局限性,提出最大熵原理的贝叶斯不确定度评估方法,引入最优化算法和计算机编程实现不确定度的优化估计;通过模拟仿真,验证了所提出方法的有效性。以贝叶斯动态预测原理和模型为基础,研究了贝叶斯动态不确定度评定与预测方法。分析了动态随机过程特征和类型,重点建立了各态历经随机过程动态不确定度预测模型;讨论了非各态历经随机过程动态不确定度评定方法;通过测量实例分析,为实际动态不确定度评定提供了普遍意义的指导。鉴于蒙特卡洛方法在不确定度评定中的应用优势,提出了蒙特卡洛不确定度验证方法;通过模拟仿真实例分析,验证了贝叶斯动态不确定度评定与预测方法的可操作性,保证了不确定度评定与预测结果的可靠性。重点关注了不确定度在产品检验中的应用,研究了不确定度影响下的产品检验合格性判定方法;建立了单一产品检验合格判定误判率计算模型;针对批量产品检验,提出了全数检验产品合格判定误判风险评估方法。通过产品检验实例分析,综合运用所提出理论,为基于不确定度的产品检验合格判定方法提供了细化指导,为产品供求双方协商决定产品的合格性提供了科学依据。
[Abstract]:As an important parameter index to characterize the measurement results, uncertainty has been paid more and more attention by various fields of the society. With the development of the times, the rapid development of science and technology and the high attention to the quality of products, people pay more and more attention to the quality of products. Modern uncertainty evaluation methods and theories come into being. It is of great significance to promote the wide application of modern uncertainty. Based on the analysis of modern uncertainty theory, the evaluation method of Bayesian uncertainty is studied systematically, which is based on conjugate Bayesian principle. A real-time and continuous updating method for the uncertainty components of historical information and current sample information is proposed, and a Bayesian uncertainty evaluation method based on maximum entropy principle is proposed to evaluate the uncertainty of Bayes, which has no information and conjugate prior Bayes method. The optimization algorithm and computer programming are introduced to realize the optimal estimation of uncertainty, and the effectiveness of the proposed method is verified by simulation, which is based on the Bayesian dynamic prediction principle and model. The evaluation and prediction methods of Bayesian dynamic uncertainty are studied, the characteristics and types of dynamic stochastic processes are analyzed, and the prediction models of dynamic uncertainty of ergodic stochastic processes are established. This paper discusses the evaluation method of dynamic uncertainty of non-ergodic random processes, and provides a general guidance for the evaluation of actual dynamic uncertainty through the analysis of measurement examples. In view of the advantages of Monte Carlo method in the evaluation of uncertainty, The verification method of Monte Carlo uncertainty is put forward, and the feasibility of Bayesian dynamic uncertainty evaluation and prediction method is verified by simulation analysis. The reliability of uncertainty evaluation and prediction results is ensured. The application of uncertainty in product inspection is focused on, and the method of product qualification determination under the influence of uncertainty is studied. This paper establishes a model for calculating the misjudgment rate of a single product's qualified judgment, and puts forward a risk assessment method for the batch product's qualification judgment and misjudgment. Through the analysis of a product inspection example, the author synthetically applies the proposed theory. It provides detailed guidance for the method of product qualification based on uncertainty, and provides scientific basis for both supply and demand parties to decide the conformity of product through negotiation.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG801

【参考文献】