论文部分内容阅读
针对现有的多响应优化方法只注重某种最优性指标(例如满意度)的构建,而忽略了因子容差扰动的影响,提出了一种稳健的满意度函数法,对多响应曲面问题进行稳健优化。该方法在传统的满意度函数法的基础上,根据设计因子的容差区域定义总体满意度的稳健性度量指标,并引入混合遗传/模式搜索算法对该指标进行最大化寻优,得到因子的稳健最优水平。实例表明,随着因子容差区域的逐渐变大,该方法更容易寻找到稳健可行域中的解。与传统方法相比,所获得的稳健最优解对因子容差具有更好的抗干扰性,从而更具应用价值。
In view of the fact that the existing multi-response optimization method only focuses on the construction of some kind of optimality index (such as satisfaction degree), ignoring the influence of factor tolerance disturbance, a robust satisfaction function method is proposed to solve the multi-response surface problem Make a robust optimization. Based on the traditional method of satisfaction function, this method defines the robustness measure of overall satisfaction according to the tolerance region of design factors, and introduces the hybrid genetic / pattern search algorithm to optimize the indicator to get the factor Robust optimal level. The example shows that as the factor tolerance region becomes larger, the method can find solutions in robust feasible domain more easily. Compared with the traditional method, the obtained robust optimal solution has better anti-interference to the factor tolerance and thus has more application value.