基于性能稳健偏差的区间型参数稳健设计优化

提出了基于系统性能稳健偏差的稳健设计优化方法。该方法针对区间型参数非梯度性能函数的稳健设计问题，应用性能稳健偏差概念评估目标性能函数和约束性能函数的稳健性，形成了三层次稳健设计优化模型：①搜索优化设计点；②迭代策略搜索设计点的目标函数和约束函数的稳健偏差；③优化寻求给定性能偏差所对应的稳健性指数。该方法可获得同时满足约束条件和目标值稳健性要求的优化设计，并输出目标值及其稳健偏差。用三杆桁架结构稳健设计优化说明了该方法的用法和特点。

Robust Design Optimization Involving Interval Parameters Based on Robust Performance Variations

This paper present a new robust design optimization method based on robust performance variation estimation, in which only the known ranges of design parameters were required without presumed probability distributions. The robust performance variation was used to represent the robustness of design instead of the robustness index, the robustness of objective value and the robustness of feasibility in the formulation of robust design optimization. The new robust design optimization involving interval parameters included three levels. Upper level optimized objective function subject to the robustness constraints, which output a design solution and receipted the robust performance variations of objective functions and constraint functions from the middle level. The middle level conducted the search for the robust performance variations using an iterative approach, which output robust performance variations and receipted the design solution for the upper level, and output the performance variation and obtained the robustness index for the lower level. Lower level performed the search for the robustness index at a given performance variation by means of an optimization approach. By using analytical interpretation and providing illustrative examples, the RPV method shows its advantages that its process and results are easier to be understood by designers and that it enables designers to easier achieve the robustness of a design solution by simultaneously optimizing the mean performance, minimizing the performance variation and maintaining feasibility robustness and to carry out a trade-off analysis between the objective values and the robustness of design solutions because the robust performance variation is of the same physical significance as the performance function.