数据与机理驱动的精密机械装配性能精准调控方法研究

The requirements of comprehensive physical properties such as the service performance and reliability of precision machinery are continuously increased. The lack of a deep understanding of the relationship between geometrical error, assembly adjustment process and physical properties is the key problem to guarantee and improve the performance of precision machinery. The assembly performance adjustment method is investigated in this project based on inspection data from three aspects: data description, construction of the hybrid analysis model, the adjustment and control strategy of assembly performance. The proposed method overcomes the theory difficulty of mechanism research method when facing the cross scale and strong coupling problems in complex assembly process. The representation method of multi-source heterogeneous inspection data based on assembly process description is studied. The assembly process information model is constructed to implement the data mining of the inspection data. The hybrid analysis model integrating data and mechanism analysis model is explored. The method of revealing the generation of assembly process performance from inspection data is investigated. The adjustment decision, assembly process planning and inspection optimization in the aim of the assembly performance is studied in this project. An innovative concept of precision assembly driven by data and mechanism is proposed in this project, which is helpful to realize the knowledge discovery and intelligent decision technology in precision mechanical assembly process. This study provides theory and technology support for precision assembly of mechanical system oriented to intelligent manufacturing.

精密机械服役性能、可靠性与性能保持性等综合物理性能要求不断提高，而缺乏对几何误差、装调工艺与产品物理性能关联关系的深入认识是制约精密机械装配性能精准调控与性能保障的关键问题。本项目从装配检测数据描述、混合分析模型构建、装配性能调控三个层面开展基于检测数据的精密机械装配性能精准调控方法研究，克服机理出发的研究方法在面对复杂装配过程的多因素、跨尺度、强耦合理论难题。研究基于装配工艺描述的多源异构检测数据表征方法，通过构建装配过程信息模型，对检测数据进行数据挖掘，探索数据与机理分析模型融合的混合分析模型构建方法，探究由数据表象揭示装配过程性能衍生规律的方法途径。研究以综合性能为目标的装调工艺决策、装配工艺规划、检测工艺优化方法。项目提出数据与机理驱动的精准装配这一创新构想，有助于精密机械装配工艺知识发现与智能决策技术实现突破，为切实推进面向智能制造的机械系统精准装配提供理论与技术支撑。

复杂精密机械装配性能、可靠性与性能保持性要求的不断提升，对装配过程几何误差、装配工艺与产品装配性能之间的关联关系挖掘、装配性能溯源与调控策略的准确分析提出了更高要求。本项目以构建数据与机理驱动的精密机械装配性能分析、预测与调控方法，揭示装配过程几何-物理因素的形成与演变规律为目标，结合运动机构、固定连接结构等精密机械装配问题开展理论研究与技术开发。针对测量数据分析问题，提出了一种解决欠定多工位装配系统偏差诊断问题的方法。将偏差传递模型与数据统计分析技术相结合，建立了增强贝叶斯层次模型解决欠定系统偏差辨识的推理问题，为装配系统的质量控制与测量方案规划提供了一个有效的算法工具。针对装配性能精准预测问题，建立了机构运动学基础上的装配误差分析模型，获得了组成杆件尺寸、铰链间隙等误差因素与装配精度之间的解析关系，打破刚体和理想体假设，建立了复杂形貌数据描述方法和刚柔耦合的装配精度分析模型。在批量样本数据基础上，进一步通过数据驱动方法建立装配性能预测模型，提升了性能预测精度与效率。针对装配性能调控问题，重点构建装配性能调控约束描述与求解算法。采用数据挖掘与多目标优化相结合的方法建立复杂杆系定量装调优化算法，构建了采用粒子群算法进行装配相位优化的装调策略，使装配同心度、装配性能多目标达到设计要求。项目以卫星天线展开机构装配精度与可靠性、航空发动机转子装配同心度为典型对象开展系统的研究工作，所形成的装配误差建模、分析预测、工艺优化方法为现场装配提供了指导，以算法为基础的原型软件应用于实际产品现场装配调整中。