Abstract: When dealing with computationally expensive black-box engineering optimization problems, the genetic algorithm is inefficient. In order to improve engineering optimization efficiency, combining Kriging surrogate optimization and physical programming, this paper proposes a multi-objective surrogate optimization algorithm based on Kriging and physical programming. When dealing with multi-objective problems, this paper uses physical programming to convert multi-objective problems into single-objective problems, and then uses Kriging surrogate optimization to solve the problem. Two multi-objective numerical examples and one engineering example are used to verify the proposed algorithm. The results show that the proposed algorithm can find Pareto optimal solutions in accordance with preference settings, and the algorithm is more efficient.

Key words: Kriging model, physical programming, surrogate optimization, multi-objective optimization

摘要： 遗传算法处理高耗时且具有黑箱性的工程优化问题效率不足。为了提高工程优化效率，结合Kriging代理优化和物理规划，提出了基于Kriging和物理规划的多目标代理优化算法。在处理多目标问题时，使用物理规划将多目标问题转换成单目标问题，再使用Kriging代理优化对单目标问题进行求解。通过两个多目标数值算例和一个工程实例对提出的算法进行验证。结果表明，提出的算法能够求出符合偏好设置的Pareto最优解，且算法的效率更高。

关键词: Kriging模型, 物理规划, 代理优化, 多目标优化