基于神经网络和遗传算法的卫星结构参数优化

针对卫星结构要求总质量轻小、一阶固有频率较高的特点,建立了卫星结构多目标优化的数学模型,通过BP神经网络和遗传算法相结合进行参数优化,并编制了相应的计算程序,既利用了神经网络的非线性映射、网络推理和预测功能,又发挥了遗传算法的全局优化特性,得出了合理的优化结果,与传统的结构优化方法相比,此方法效率较高,精度良好.

Parameters Optimization of Satellite Structure Based on Artificial Neural Networks and Genetic Algorithms

Considering the weight and the first eigenfrequency of the satellite structure,a multi-objective optimization design model of the satellite structure is presented.Using combination of BP neural networks and genetic algorithm,a program is designed for the structure optimization.This method uses not only the nonlinear mapping,network reasoning and predicting of the neural networks,but also the global optimal characteristics of genetic algorithm,which gains a reasonable optimization effect.The example shows that the technique has better calculating rate and precision than those with traditional optimization methods.