基于代理模型的箭体连接结构极值响应分析与优化

Extreme response analysis and optimization of rocket connection structure based on surrogate model

螺栓法兰连接结构在航空航天等工程领域中广泛应用,其力学性能在不同工况和装配情况下十分复杂.由于拉压刚度差异,含连接结构的箭体动力学响应呈现明显的非线性特征.因此,考虑不同连接参数及工况下的连接非线性动力学响应,对结构优化设计有着重要意义.本文针对以双线性弹簧表征螺栓法兰连接非线性的箭体等效动力学模型,基于径向基函数(RBF)神经网络和响应面法分别建立其连接面处的极值响应代理模型,对比发现RBF神经网络模型在较高精度上可以实现对动响应极值的预测及分析;同时分析了不同载荷参数及刚度变化对连接结构动响应极值的影响;最后,利用RBF神经网络代理模型,开展了连接面加速度极值响应与螺栓弹簧力最小化为目标的连接结构参数优化.

The mechanical properties of bolted flange connection structures used widely in aerospace engineering and other fields are complex under different working environments and assembly conditions.Due to the different stiffness of in tension and compression,the dynamic response of the rocket demonstrates nonlinear features.So,it is important for structural optimization to consider the dynamic response of nonlinear connection under different structural parameters and working conditions.In this paper,surrogate models based on RBF neural network are proposed to predict and analyze the extreme responses of the connection part of the equivalent structure of the rocket,in which nonlinearity of the bolted flange connection is characterized by bilinear springs.It is found that RBF neural network model predicts and analyzes the extreme dynamic responses accurately.Meanwhile,the influence of different load parameters and variation of stiffness on the extreme dynamic responses is analyzed.Finally,using the RBF neural network surrogate model,the parameter optimization of the connection structure is carried out to minimize the extreme acceleration responses and the bolt spring forces of the connection.