【作者】 高嘉爽；

【导师】 方洪渊；

【作者基本信息】 哈尔滨工业大学 ， 材料加工工程， 2009， 博士

【摘要】 焊接技术是先进制造技术的重要组成部分,焊接技术水平直接影响着制造业的整体能力和水平。随着先进制造技术的发展,实现焊接产品制造的自动化、柔性化与智能化成为必然趋势。在实际生产之前,有必要对焊接产品进行设计,综合利用计算机技术、网络技术、并行工程、分布式人工智能对焊接产品进行协同优化,对实现焊接产品结构优化具有重要意义。本文根据以上需求,基于对焊接加工过程特点的分析,提出综合运用智能体(Agent)技术及有限单元分析(FEA)技术来构建焊接加工过程协同设计系统(Welding Processes Collaborative Design System, WPCDS),以实现焊接过程的协同优化。在系统的对单工序内各个参数协同优化策略及焊接过程多工序加工协同优化策略进行研究的基础上,建立了基于Agent的WPCDS,并对系统中的若干核心Agent进行了分析与研究,构建了相应的Agent及基于网络的仿真环境。本研究从焊接有限元分析特点出发,提出遗传算法与有限元计算相耦合的方法对单工序内部各项设计变量进行优化,并以某油箱焊接过程为例进行了模拟,结果表明:这种耦合方法可以高效、科学地获得优化参数组合。考虑到工程产品往往经过多道工序加工,所以分析了多工序加工时的各项设计变量的优化。提出了焊接分析过程的三层体系结构,即设计层、工序层和工艺层。在此分层模型基础上,建立了各层的功能模型,定义了工艺活动模型,并对工序类型进行了划分并相应地确定了优化策略,分析表明:该优化策略可以克服传统CO方法过程复杂、优化规模大、非线性计算困难等缺点,适合焊接加工的特点,实现各个设计变量协调、动态、并行地进行优化。将以上策略应用于某油箱结构的焊接参数优化,以焊后各单元节点最大变形量最小为适应度判断依据,优化结果表明:该策略可以实现在统一的调控机制下进行多项有限元任务的协同并行计算,可以获得使焊接变形降低的优化参数组。基于上述的优化策略,设计了WPCDS框架,该框架包括5个子模块和1个加工设备库,并以一个典型的分析过程为例说明协同设计步骤。进一步,采用MaSE方法对WPCDS进行了目标分析、用例分析、角色定义、系统部署。构建出的MAS(Multi-Agent System)能够使焊接过程设计过程更加柔性,能充分利用Agent的自治性、动态性、异构性等特性,完成焊接加工过程各项有限元计算任务的协同并发优化设计。针对焊接工序的特点,重点研究了WPCDS中的焊接有限元计算模块,开发出面向焊接过程的特有的Agent。设计并开发了焊接接头库Agent、焊接预装配Agent、焊道填充模拟Agent。实例模拟发现:焊接接头库Agent可简化焊接建模过程,提高接头设计的效率;焊接预装配Agent可以使用户省去乏味和复杂的前处理过程,将抽象的条件加载转化为逼真的虚拟焊接操作,为虚拟制造技术在焊接领域中应用提供了思路;在理论上给出“节点动态松弛法”的证明之后,设计了焊道填充模拟Agent,该方法解决了生死单元法及静态单元法在模拟焊道填充时常出现的问题。以上的这些Agent扩充了WPCDS的功能。由于跨学科、跨文化的设计队伍远程合作的需要,以及高效合理利用异地分布的计算资源的需要,设计了基于网络的WPCDS,该系统建立在JADE平台中,根据要完成的任务为Agent划分物理计算资源,建立协同优化仿真环境,定义必要的本体。研究表明:网络环境下的协同设计可以为实现设计人员之间以及设计、分析软件之间的数据、信息和知识的共享提供条件,也能为实现多种已有设计和分析软件的松散集成,满足开放式系统的快速设计和优化提供可能。更多还原

【Abstract】 As a key techonolgy in advanced manufacturing field, welding techonolgy has affected the facrication industry markedly. With the imporovement of advanced technologies, automatization, intelligence and flexibility have become the inevitable trend in welding techonogly. Furthermore, it is necessary to design welded processes before manufacturing, in which collaborative optimization by using computer technology, net technology, information technology and distributed artificial intelligence is very important for welding structure optimization.Based on welding processing characteristics, Welding processes collaborative design system (WPCDS) using agent technology and finite element analysis (FEA) was proposed to make welding processes optimized collaboratively. In the thesis, according to the collaborative strategy of multi-parameter in single process and multi-process in continuous manufacturing, WPCDS is built on basis of agent technology and some other critical agents were discussed and researched, furthermore, net simulation environment was also built correspondingly.A coupled method of genetic optimization and FEA is proposed for optimizing multi-parameter in single process considering characteristics of welding FEA, and its validity and rationality were proved by optimizing an oil tank welded structure by this coupled method.Optimization of parameters in multi-process is necessary to be studied for all engineering product must be produced by multi-pass processes. A three layers structure including designing layer, processing layer and parameters layer was proposed to describe these processes; function model of each layer and production model were defined respectively, in addition, those processes were classified and corresponding optimization strategy was discussed. The analysis shows: the complex and difficulty of traditional CO method can be overcame by using this new strategy; those designing parameters are no more dependent, static and experienced, but the internal relationship of variable process parameters and process stage can be discovered during collaborative optimization; technology research groups and experts can cooperate better. The strategy is used to optimize welding parameters of oil tank structure with the fittness function criterion of maximum deformation of all nodes being minimized, the results show that multi-FEA tasks are done collaboratively using a unified controlling mechanism and a group of better welding parameters are found.Based on collaborative strategy mentioned above, the framework including five modules and one machine database of WPCDS was built, and was interpreted by a typical design process. Furthermore, the multi-agent system (MAS) prototype for welding processes collaborative design was studied in detail based on agent technology and MAS was built using MaSE method, main steps are as follows: goal analysis, use case analysis and role definition. Because agent is autonomous, dynamic and heterogeneous, the designing process of welding product becomes flexible. Obviously, it is seen that FEA tasks can be finished collaboratively and welding product can be designed rapidly by WPCDS.Welding FEA module in WPCDS was studied emphatically and welding joint database agent, pre-welding assembly agent and welding filler simulation agent were proposed because they are important for WPCDS based on agent technology. The utility of those agents for oil tank welded structure shows that spending time in building model of welding FEA is discreaesed sharply and the computing efficiency is improved using welding joint database agent; the loadcase definition related with welding assembly is easy and vivid by using pre-welding assembly agent, welding VRML(virtual reality modeling language) model is operated according to user click mouse or keyboard, not tedious menu operations anymore; welding filling simulation agent discussed based on the theory of“node dynamic relaxation method”is a better method to simulate weld filling process comparing with “deactive element method”and“quiet element method”. Thus, WPCDS is extended by three special agents mentioned above.Net environment for WPCDS was built because different disciplines and study groups need remote cooperation; even different local computer also need integration. On JADE platform, according to the different tasks of agents, an amount of physical computer resources are divided, collaborative simulation environment is set up, necessary ontologies are defined. So it is useful for the data sharing, information sharing and technology sharing from different analysis and design software, and it is also helpful for the loose integration of engineer softwares, the reality of open system for rapid designing and optimization.更多还原