基于BP神经网络与遗传算法的减振器优化设计

摘要

The equality of stiffness in three directions of the system means that the non-rotational vibration＇s inherent frequency of the three elastic axis is almost the same, which is the main goal of rubber absorber structure optimization design. By considering the factors of damping, stiffness, and boundary conditions etc, the finite element simulative model was established by using ANSYS software. With the help of the data generated by finite element a- nalysis, trained; orthogonal training sample data sheet was created; BP（ Back ） neural network was successfully and the nonlinear mapping function from design variables to system frequence was constructed. Aiming at the equality of stiffness in three direction of the system, the neural network model was optimized through the genetic algorithm. Finally, the best absorber design parameters was obtained.%使减振系统三向等刚度，即3个弹性主轴方向上的非回转振动固有频率基本一致，是对橡胶减振器进行结构优化设计的主要目标。在综合考虑了阻尼、刚度，边界条件等影响因素的基础上，在ANSYS中建立了参数化有限元仿真模型。利用有限元分析得出的数据建立了正交样本数据表，完成了BP神经网络的训练，建立了设计参数到系统频率的映像关系。以系统的三向等刚度为优化目标，利用遗传算法对神经网络模型寻求最优解，得到了最优的减振器设计参数。