Jixie qiangdu (Jan 2023)
MULTI-OBJECTIVE ROBUST OPTIMIZATION DESIGN OF COMPLIANT HINGE BASED ON BP NEURAL NETWORK (MT)
Abstract
In order to improve the robustness of the compliant hinge, genetic algorithm and BP neural network methods are introduced to optimize the parameters of the compliant mechanism. Orthogonal experiments are used to select training parameters and test parameters, a BP neural network model is established, and by using the nonlinear fitting ability of the neural network and the global search and optimization ability of genetic algorithm. The flexibility and natural frequency signal-to-noise ratio are used to find the global optimum in the selection range for single and multi-objectives. It is not only limited to the permutation and combination of selected factor levels, but also provides a new solution to improve the compliance hinge robustness. The experimental results show that the comprehensive evaluation function of the compliant hinge is better, which achieves the purpose of robust optimization design, and proves the effectiveness of the method.
