CAREER: A Generic Scheme for Graph Topology Optimization

这个职业发展项目将研究解决拓扑设计问题的第一个通用方案。这种方法结合了优化、图论、数学编程、人工智能以及形状和图形语法的各个方面。虽然传统的优化通常用于在已经详细的设计中找到尺寸和参数的最佳值，但拓扑优化可以在设计过程的早期使用，以确定本构元素的配置。考虑到部件类型的丰富性和多样性，以及较短的设计周期带来的压力，配置部件可能会很繁琐。这类拓扑问题在工程设计中十分普遍。例如，金属板零件的设计，道路的位置和连接，以及化工厂中加工机的选择和连接。因此，提出了一种自动选择和配置元素的设计综合方法。研究活动将与教育计划相结合，该计划包括以项目密集型方法教授机械元素、面向对工程感兴趣的高中生和本科生的基于网络的档案袋工具，以及开发两门跨学科的设计研究生课程。预计当新方法用于解决实际工程设计问题时，研究的更广泛影响将变得显著。由于该方法增加了表示能力，提高了搜索过程的效率，因此可以在计算上解决更具雄心的设计问题。通过开发和实现用于解决此类搜索问题的通用模块，研究人员将能够实现其特定的表示和评估代码，以便在拟议的框架内工作。。

This career-development project will investigate the first generalized scheme for solving topological design problems. This approach combines aspects of optimization, graph theory, mathematical programming, artificial intelligence, and shape and graph grammars. While traditional optimization is typically used to find the best values for dimensions and parameters within an already detailed design, topology optimization can be used earlier in the design process to determine a configuration of constitutive elements. Configuring components can be tedious given the abundance and variety of component types and the pressures imposed by shorter design cycles. Such topological problems are prevalent in engineering design. Examples include the design of sheet metal components, the location and connection of roadways, and the choice and connection of processors in a chemical plant. Thus, a design synthesis method is proposed to automatically choose and configure elements.The research activities will be integrated with an educational plan that includes a project-intensive approach to teaching machine elements, a web-based portfolio tool geared towards high-school students interested in engineering and undergraduate students, and the development of two interdisciplinary graduate courses in design. It is expected that the broader impacts of the research will become significant when the new method is used to solve real engineering design problems. Since the method proposes to increase the representation abilities and improve the efficiency of the search process, more ambitious design problems can be addressed computationally. By developing and implementing generalized modules for solving search problems such as these, researchers will be able to implement their specific representation and evaluation code to work within the proposed framework. .