Towards Rational Simulation-Based Engineering Optimization

尽管优化在工程和产品开发中得到广泛应用，但为基于仿真的工程优化选择高效的优化策略仍然是一个公开的挑战。对于一定数量的计算资源，建模，仿真和优化（MSO）过程的效率，直接影响的性能和质量的技术产品的设计结果。该项目的目标是开发一种方法框架，在有限的计算资源约束的背景下分析基于模拟的优化问题和算法，以提高整个MSO过程的特定问题的效率。该框架将包括：（一）新的和现有的问题表征和适应度景观分析方法;（二）系统优化算法基准方法，以及（三）新的基于问题特征的算法选择和分析方法。虽然该框架将使用新生成的基于有限元仿真的工程优化问题或技术相关性的案例研究进行开发和测试，但这些方法将以一般方式制定，以便它们可以应用并扩展到其他基于仿真的问题。该框架将使建模，分析和理解的策略，资源有限的MSO过程，以提高其计算效率，通过合理的方法和算法选择。

Despite the wide-spread use of optimization in engineering and product development, the selection of efficient optimization strategies for simulation-based engineering optimization is still an open challenge. For a fixed amount of computational resources the efficiency of the-Modeling, Simulation and Optimization (MSO)-process, directly influences the performance and quality of the resulting design of technological products. The objective of this project is the development of a methodological framework to analyze simulation-based optimization problems and algorithms in the context of bounded computational resource constraints order to improve the problem-specific efficiency of the overall MSO-process. The framework will be composed of: (i) new and existing methods for problem characterization and fitness-landscape analysis; (ii) systematic optimization algorithm benchmarking approaches, and (iii) new problem feature-based algorithm selection and analysis approaches. Although the framework will be developed and tested using newly generated case-studies of Finite Element Simulation-based engineering optimization problems or technological relevance, the methods will be formulated in a general way, such that they can be applied and extended to other simulation-based problems. The framework will enable to model, analyze and understand strategies for resource bounded MSO-processes in order to improve their computational efficiency by rational method and algorithm selection.