多质量指标的解耦与稳健优化设计

Decoupling of Multi-quality Characteristics and Robust Design Optimization

分析了产品的多质量特性指标与设计参数的耦合关系,提出了一种耦合性强弱的度量指标与解耦方法。在此基础上,提出一种多质量特性指标的稳健优化设计方法,基本思路是:按信息公理的要求,对每个质量特性指标进行稳健优化,减少其不确定性信息含量;按独立性公理要求,对多个质量指标进行近似解耦,以提高多个质量指标的整体稳健性。为了提高模型求解效率,用BP神经网络作为非线性随机函数的替代模型,并与遗传算法相结合,构建了一种混合智能优化算法。通过稳健设计实例,对所提出的方法进行了验证。

The coupling relations between the multi-quality characteristics and design parameters of products were analyzed. An index for measuring the coupling degree of multi-quality characteristics was defined and a robust design criterion was put forward to achieve the decoupling of characteristics approximately. A modeling method of the robust design with multi-quality characteristics was proposed. On one hand, according to the requirement of Suh＇ s first design axiom （the information axiom）, the robustness of each characteristic was optimized to reduce its contents of uncertainties; on the other hand, according to the requirement of Suh＇ s second design axiom （the independence axiom）, the characteristics of a product were decoupled approximately to improve the integrated robustness. A back-propagation neural network was designed as the substitution for the nonlinear stochastic functions in the established model. With genetic algorithm combined with stochastic simulations, an intelligent optimization algorithm was developed for solving the established model effectively. An example was presented, and the results show that this method is effective and feasible.