论文部分内容阅读
针对产品质量改进中的多响应稳健性优化问题,提出了一种基于贝耶斯分析的递阶优化方法。首先基于满意度函数方法求出初始优化解,然后通过贝耶斯分析评价解的稳健性,以初始优化解为起始搜索点进行稳健性寻优;针对现有稳健最优解可靠性较差的情况,给出了两种改进策略下仿真数据的产生方法,利用贝耶斯预后验分析来对未来改进措施的效果进行定量评价。该方法可以实现最优解和稳健解的权衡,降低算法的复杂度并提高寻优效率,且适用于响应曲面模型回归项不一致的情况。算例表明,对多响应优化问题进行贝耶斯分析能有效找到稳健最优解,并可以为后续实验改进提供依据。
Aiming at the multi-response robustness optimization in product quality improvement, a hierarchical optimization method based on Bayesian analysis is proposed. Firstly, the initial optimal solution is obtained based on the satisfaction function method. Then, the Bayesian analysis is used to evaluate the robustness of the solution and the initial optimal solution is used as the initial search point for robustness optimization. In view of the poor reliability of the existing robust optimal solutions , Two methods for generating simulation data are given. The Bayesian prognostic analysis is used to quantitatively evaluate the effect of future improvement measures. This method can realize the trade-off between the optimal solution and the robust solution, reduce the complexity of the algorithm and improve the efficiency of the optimization, and is suitable for the inconsistent regression of the response surface model. The example shows that the Bayesian analysis of multiple response optimization problems can find a robust optimal solution efficiently and provide a basis for further improvement of experiments.