【作者】 赵志彪；

【导师】 刘明周；

【作者基本信息】 合肥工业大学 ， 工业工程， 2013， 博士

【摘要】 复杂机械产品是由多种类型零、部件按照一定装配序列、在相应工艺参数作用下，聚合成具有多项技术要求的混联结构装配体。复杂机械产品装配过程属于串、并行相结合的多工序制造过程，在制品向下游工序流转的过程中，以其为载体的各类型质量特征不断传递、累积，形成产品各项质量指标，即最终产品的质量特性是由分布于装配过程各道工序中、具有一定耦合规律的质量特征所共同决定。因此，从全局角度出发，探讨不同类型质量特征间耦合规律，设计优化控制策略，实现装配质量在线优化和误差累积动态补偿，对减少复杂机械产品装配过程质量波动、提高装配质量稳定性有着重要的现实意义。本文从装配质量形成机理出发，将研究对象从装配尺寸推广到转矩、力等非尺寸要素，基于模型驱动研究理念，探讨装配过程中不同类型质量特征变迁、融合规律和相互作用机制，在此基础上，结合上游工序实例化数据，对装配过程在线质量工况信息做出及时响应、调整，实现装配误差累积动态补偿和装配质量主动控制，为提高复杂机械产品装配质量稳定性提供理论依据与技术支持，论文的研究内容和创新点如下：1）提出质量控制点和质量属性的概念，构建基于网络流的质量控制点-质量属性相关性模型，在识别影响质量属性的关键质量控制点基础上，实现装配质量特征约束关系的显式表达，为进一步探讨装配质量主动控制和在线优化提供研究基础。2）通过建立装配质量控制点公差分配模型，定量表征不同类型质量控制点间的非线性耦合关系，为质量控制点在线优化提供策略支持，该优化策略为不依赖历史样本的模型驱动控制模式，能够适应当前多品种、小批量的柔性制造环境。为了提高求解质量，将混沌机制和双阶段变异策略嵌入多目标粒子群算法用于模型解算，保证了帕累托最优集的多样性。3）构建具有自主协调能力的装配质量数据链模型，集成了装配资源、装配工艺、智能控制策略等方面的信息，实现装配过程中在制品质量状态的“可观”、“可控”，能够根据装配质量工况信息做出实时调整，解决装配过程中误差累积动态补偿问题；通过模型动态行为模拟、推演，为系统可达性分析提供依据，并为在线质量控制系统开发奠定基础。4）基于统计理论对传统装配模式和本文在线优化策略下产生的装配质量数据序列进行分布性态验证、方差齐次性分析以及均值比对，证明了优化方法能够有效提高复杂机械产品装配质量稳定性；利用模糊关联分析计算样本序列与标准序列之间的平均隶属度和贴近度，进一步佐证了先前论断。5）开发面向复杂机械产品装配过程在线质量控制系统，将智能计算、遥感测控、工业数据传输网络等相关技术无缝集成，为提高装配过程质量稳定性提供技术保障。更多还原

【Abstract】 Complex product, defined as “hybrid structure assembly” with technical requirements, iscomposed of many kinds of parts according to the corresponding process parameters and certainassembly sequence. Complex product assembly process is a serial-parallel-hybrid-multi-stagemanufacturing process.In the process of WIP (Working In Process)flowing to the downstream,product quality attributes are accumulated by the various types of quality characteristics of WIP. Asa result, the quality characteristics of the final product are determined by the coupling qualitycharacteristics in each assembly procedure. Therefore，to realize the online optimization ofassembly quality and dynamic compensation of accumulated errors,from the global perspective, itis significant and pratical to explore the coupling laws in different types of quality characteristicsand design the control optimization strategy for improving the assembly quality stability. Toimprove the assembly quality stability of complex product, from the sight of assembly qualityformation mechanism, the research objects are extended from the assembly sizes to other kinds ofelements such as torque, force. Based on the model-driven philosophy, the integration law and theinteraction mechanism among the different types of the quality characteristics in assembly processare studied, considering the instantiated data of the upstream quality control points, the timelyresponse and adjustment of the on-line quality information are carried out.The research contents and innovations of the dissertation are as follows:1) The concept of quality control points and quality attributes are proposed.Based on networkflow theory, the Correlation-Model between quality attributes and quality control points isestablished, and the constraint relationships among the assembly quality characteristics areexpressed by this model, Correlation-Model can lay the foundation of future research of theassembly quality active control and online optimization.2) The tolerance distribution model of quality control points oriented to assembly quality ispresented to characterize the nonlinear coupling relationships among different types of qualitycontrol points,which provides policy support for online optimization of the quality control points.The optimization strategy is a pattern of model-driven control mode which can well adapt to thesmall-batch flexible manufacturing environment due to being independent of historicalsamples.The chaos theory and two-stage variation strategy are adopted to improve multipleobjective particle swarm optimization (MOPSO). The diversity of pareto optimal set is ensured.3) The assembly quality data chain model with the ability of self-coordination isestablished.The assembly resources, assembly process and intelligent control strategies are integrated into this model to achieve the quality state of WIP of "visualization "and" controllability".The dynamic compensation of error accumulation in assembly process is resolved by real-timeadjustment according to the real-time quality information. The simulation of the model dynamicbehavior is the basis of the system reachability analysis and the development of the online qualitycontrol system.4) To prove the optimization method can effectively improve the assembly quality stability ofcomplex product, two groups of assembly quality data sequences, which are generated from theoptimization strategy and the traditional assembly mode,respectively, are compared by thestatistical theory including the verification of distribution state, variance homogeneity analysis andaverage contrast. To verify the previous assertion, the average membership grade and closenessdegree between the sample sequence and the standard sequence are calculated by the fuzzyassociation analysis.5) To provide technical support for improving the quality stability of the assembly process, theonline quality control system for complex mechanical assembly process is developed, integratedseamlessly with the intelligent computing, telemetry and industrial data transmission networks inthe online quality control system.更多还原