不确定条件下复杂产品性能增强设计理论、方法及其应用研究

发布时间：2018-03-30 00:26

本文选题：复杂产品　切入点：性能增强设计　出处：《浙江大学》2017年博士论文

【摘要】：针对复杂产品设计具有层次性、耦合性、模糊性、迭代性、认知性等固有特点,将性能设计与复杂产品设计过程进行融合,系统性地提出了不确定条件下复杂产品性能增强设计理论和方法,对期望性能辨识、行为性能均衡、结构性能适配以及预测性能评估等关键技术进行了深入研究,并根据研究成果开发了复杂产品性能设计系统集成平台并实际工程应用,取得了良好的成果,验证了本文方法的有效性与可行性。本文的研究内容主要包括:第一章概述了复杂产品性能设计的内涵理解及其相关技术的研究现状,并讨论了产品性能增强设计与不确定设计的研究现状,提出了不确定条件下复杂产品性能增强设计理论及方法,同时阐述了本文的研究内容和总体结构框架。第二章提出了基于状态感知的产品期望性能解析辨识技术。根据性能在产品全生命周期逐层演化的特点,构建了期望性能闭环解析的机理模型;提出了不确定条件下期望性能正向递推解析与状态感知反馈的期望性能反向计算方法,并采用模糊积分对意图重要度函数进行补偿修正,实现了需求域与过程域影响下的性能重要度非线性融合辨识及度量。第三章提出了基于解耦计算的产品行为性能均衡求解技术。基于模糊不确定的产品行为关联准则数据,对产品构型的耦合强度进行计算;采用设计结构矩阵聚类对产品进行初次解耦规划并计算解的行为单元性能指数;以初次解耦行为单元为基础,构建了以质量稳定性、维修成本与系统可用性为目标以及相关阈值为约束的行为性能模型;提出了基于随机搜索的离散粒子群算法对性能模型进行多目标优化求解,并采用熵权理想点排序法识别最优结果,从而获得性能均衡的单元方案。第四章提出了基于约束传递的产品结构性能模糊适配技术。首先对结构性能的约束信息空间进行约简并以隶属度的方式对约束空间中的信息进行一致性转换;以关联相似度为依据,提出了面向约束处理的二次过滤技术,实现性能约束驱动下的功能域到结构域的关联映射;构建了以成本、质量与物理相容性为目标及相关阈值为约束的数学模型,提出了基于约束水平的离散差分进化算法对带约束的性能适配过程进行多目标优化,同时采用一种约束满足偏差最小的方案决策方法选取结构性能最优的适配结果。第五章提出了鲁棒学习的产品预测性能可信评估技术。首先对样本输入数据进行预处理,采用互信息估计对设计变量进行筛选以降低预测模型的规模,采用k最近邻法对数据样本的异常点进行识别以提高其精度;对隐式的性能参数一设计变量响应模型进行回归拟合获得稳健的性能校核预测模型;提出基于Bootstrap统计推断的性能可信度分析方法,实现了预测结果对数据不确定性的处理,并通过灵敏度分析识别影响性能的关键设计变量。第六章将论文的研究成果与实际应用相结合,开发了复杂产品性能设计系统集成平台并应用于锻压装备的产品设计过程中,验证了所提出方法和技术的可行性与有效性。第七章对论文进行了总结,归纳了论文工作的主要研究成果与创新点,并对未来的研究工作进行了展望。
[Abstract]:With the level of product design, aiming at the complex coupling, fuzzy, iterative, cognitive inherent characteristics, the design and performance of complex product design process integration, systematically put forward the uncertain complex product performance under the condition of enhanced design theory and method, the expected performance of behavior identification, the balance of performance, structural performance adaptation and key technology of forecast performance evaluation conducted in-depth research, development and performance of complex product design system integration platform and the practical engineering application according to the research results, and achieved good results, verify the feasibility and effectiveness of the proposed method. The main contents of this paper include: the first chapter summarizes the research status quo of the connotation of performance the understanding of complex product design and related technology, and discusses the research status of product performance enhancement design and uncertainty, put forward under the condition of uncertainty complex The product performance and enhance the design theory and method, and expounds the research content and framework of this paper. The second chapter puts forward the state aware product identification technology based on expected performance analysis. According to the characteristics of the product life cycle layer evolution, constructs the mechanism model of expected closed-loop analytical; put forward the conditions of uncertainty under the expected performance of recursive forward calculation method and analysis of the expected performance of reverse push perception of the state feedback, and the fuzzy integral to importance function is compensated and corrected, the identification and measurement of fusion performance importance nonlinear demand domain and process domain under the influence. The third chapter puts forward the product performance behavior equilibrium calculation based on decoupling technology. Product behavior relevance data fuzzy uncertainty based on the calculation of coupling strength on product configuration; using design structure matrix clustering The initial planning of the products and decoupling performance index calculation unit of behavior of solutions to the initial behavior; decoupling unit as the basis to construct the stability of quality, maintenance cost and availability of the system as the goal and the threshold for the behavior performance model constraint; discrete particle swarm optimization algorithm was proposed based on random search on the performance of multi-objective optimization model to solve, and the entropy weighting ideal point method to identify the optimal result, so as to obtain the unit scheme performance equilibrium. The fourth chapter puts forward the structural performance constraint fuzzy adaptation technique based on constraint information. The first space on the structure properties and reduction to the membership of the information in the space constraint consistency transformation based on relevance similarity; as the basis, puts forward two constraint handling filtering technology, realize the performance constraint driven function domain to the domain of the customs Joint mapping; construction of the cost, quality and physical compatibility as the goal and related mathematical model for threshold constraints, the proposed discrete difference constraint level based on genetic algorithm with constraints on the performance of the adaptation process of multi-objective optimization, using a constraint satisfaction scheme decision method of minimum deviation of selecting optimal performance structure the results of adaptation. The fifth chapter puts forward the performance prediction of trustworthy evaluation technology of robust learning products. The first sample input data were preprocessed by mutual information estimation are selected to reduce the prediction model of the scale of the design variables, using k nearest neighbor method of data outlier identification in order to improve the precision of performance check; the performance parameters of the implicit response model of design variable regression fitting robust prediction model; analyze the performance reliability of statistical inference is proposed based on the Bootstrap Method, realized the prediction results for data processing of uncertainty, and the sensitivity analysis of key design variables influence the performance of the recognition. The sixth chapter will research combining with the practical application of product design, process design system for complex product development integration platform and its application in forging press equipment, verified the feasibility of the proposed methods and techniques and effective. The seventh chapter summarizes the thesis, summarizes the main research results and innovation, and the future research work is prospected.

【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TB472

【相似文献】