一种基于田口-遗传算法确定的神经网络及其应用

THE NEURAL NETWORK BASED ON TAGUCHI-GENETIC ALGORITHM AND ITS APPLICATION

为了解决神经网络结构和参数难以确定的问题,提出用工程领域内的田口设计方法和人工智能领域的遗传算法相结合的方法,用于同时确定神经网络结构和参数。将该算法应用到了小麦期货的收盘价短期预测上,结果表明,算法有较好收敛能力,预测是可行和有效的,有着良好的应用前景。

For solving the problem of difficult determination in parameters and structures of neural networks,in the paper it presents a new algorithm to determine the parameters and structures of neural networks. The algorithm combines the Taguchi method in engineering and genet- ic algorithm in Artificial Intelligence. The application of the new algorithm on futures forecast of wheat shows a good convergence, feasibility and validity,which indicates that this algorithm is in good prospect.