基于遗传算法优化神经网络的直升机参数选择

在直升机初步设计阶段估算其基本参数是很重要的.神经网络的通用性和精度比传统的估算方法有更多的优势,但是在应用神经网络时存在如何选择合适的网络结构和隐层节点数目等一些困难.应用遗传算法优化三层神经网络结构和连接权重,并将优化得到的网络应用于直升机参数选择中.该方法不但可以给出一个最优的神经网络结构和连接权重,而且降低了估算误差,具有及时应用最新数据学习的能力.此外,该方法易于在直升机设计系统中得到应用.

Helicopter Sizing Based on Genetic Algorithm Optimized Neural Network

It is very important to estimate the basic parameters in helicopter preliminary design. Neural Network (NN) has the advantages in estimating accuracy and generalization over traditional methods. However, there are some difficulties in using NN, e.g., how to select a proper network structure and the number of hidden layers. In this paper, structure and connection weight of a three-layer NN are optimized by genetic algorithm, and the op- timized network is applied to helicopter sizing. The proposed method can not only give an optimal NN structure and connection weight, but also reduce the prediction error and has the capability of self-learning when the lat- est data are available. Furthermore, this method can be easily applied to helicopter design systems.