Agent-Based Manufacturing Systems

9622779赖特该奖项为创建基于代理的制造系统提供资金。其关键特征是将创建制造软件“代理”，以提高精度和生产速度。今天，互联网上有大量用于工程设计、分析、规划和制造的软件包。但是，它们存在于异类分布式系统中，通常没有到单个设计器的特征明确的链接。例如，这样的远程服务可能是为一个商业CAD系统配置的，而个人设计师正在处理另一个系统。或者，服务可能是一个专门的制造设施，设计者想要“查询和获取关于它的信息”。第一个研究任务将为这些服务创建一个标准的“包装器”，使用公共语言、表示和互连。第二项任务将创造一个协作环境，以便将所有视角纳入设计。第三项任务将创建满足权衡和解决冲突的方法。如果成功，这项研究将使美国制造商能够以低成本进行生产，同时解决在尺寸、形状、表面光洁度和零件完整性方面的严格限制。例如，伯克利以前的研究为消费电子行业中使用的注塑模具的生产创建了一个集成的CAD/CAM系统。在“基于代理的制造”中，一个代理将就如何在注塑模具内获得最佳表面光洁度提供建议，而另一个代理将补偿模具偏差，从而创建更精确的部件。代理商将提取目前在互联网上分发的知识，并将其纳入设计和规划。这将避免制造过程中的模棱两可和错误，从而提高生产。

9622779 Wright This award provides funding to create Agent-Based Manufacturing Systems. The key feature is that manufacturing software "agents" will be created to enhance precision and production speed. Today, a vast number of software packages for engineering design, analysis, planning and fabrication are available on the Internet. However, they exist in a heterogeneous distributed system, usually with no well-characterized link to an individual designer. For example, such remote services might be configured for one commercial CAD system while the individual designer is working on another. Or, the service might be a specialized manufacturing facility, that the designer would like to "query and get information about". The first research task will create a standard "wrapper" for such services, employing common languages, representations and interconnections. The second task will create a collaborative environment so that all points-of-view are brought into the design. The third task will create methods to satisfy trade-offs and resolve conflicts. If successful, this research will enable US manufacturers to produce at low cost while addressing the stringent constraints on dimension, form, surface finish and part integrity. As one example, previous research at Berkeley created an integrated CAD/CAM system for the production of injection molds used in the consumer electronics industry. With "agent-based manufacturing", one agent will advise on how to obtain the best surface finish inside the injection molds, and another agent will compensate for tooling deflections, thus creating more accurate components. Agents will extract the knowledge that is now distributed on the Internet and incorporate it into design and planning. This will avoid ambiguities and mistakes during manufacturing, thus improving production.