基于BP-GA算法的环肋锥柱壳多目标优化设计

本文以超大潜深的潜艇耐压壳结构为研究对象,利用排水量表达式估算潜艇主尺度。通过Ansys软件的Apdl语言建立环肋锥柱壳的有限元模型,并分析计算耐压壳的强度及稳定性。以肋骨间距、耐压壳厚度和肋骨尺寸作为离散设计变量,以结构重量、总体失稳临界压力作为优化目标,实现基于神经网络和遗传算法的环肋锥柱壳多目标优化设计。在Matlab平台上,首先用拉丁超立方体抽样,再用BP神经网络建立起样本点和目标函数之间的映射关系,构建神经网络代理模型,最后调用多目标优化函数gamultiobj进行优化。优化结果表明,利用BP神经网络和遗传算法相结合进行复杂模型环肋锥柱壳的多目标优化,效率较高,精度较好,达到较理想的优化效果。

Multi-objective optimization design of ring-stiffened cone-cylinder shell based on BP-GA algorithm

In this paper,a ultra-deep submarine pressure shell is taken as the research object,and the main scales of the submarine are estimated by using the displacement expression.The finite element model of the ring stiffened cone shell was established by the APDL language of Ansys software,and the strength and stability of the pressure shell were analyzed.The stiffener spacing,the pressure shell thickness and the stiffener size are taken as discrete design variables.The structural weight and the overall instability critical pressure are used as the optimization targets to realize the multi-objective optimization design of the ring stiffened cone shell based on neural network and genetic algorithm.Using the Matlab,Latin hypercube is firstly sampled,and then the BP neural network is used to establish the mapping relationship between the sample points and the objective function,and the neural network proxy model is constructed.Finally,the multi-objective optimization function gamultiobj is called to optimize.The optimization results show that BP neural network cooperated with genetic algorithm in solving the multi-objective optimization of complex ring stiffened cone shells are good in efficiency and precision.