论文部分内容阅读
针对BP神经网络的特点提出一种基于递阶遗传算法的四层BP神经网络的结构设计模型及应用。现有的BP训练方法只能训练BP网络的权重和阈值,网络的结构得预先用某种方法确定。利用很好设计的递阶遗传算法能够把网络的结构、权重和阈值同时通过训练确定。以经济系统中的人口时间序列数据进行训练和测试,与传统的BP网络预测模型相比较,结果证明该模型的预测精确度是令人满意的,提出的方法是可行的。
Aiming at the characteristics of BP neural network, a structural design model of four-layer BP neural network based on hierarchical genetic algorithm and its application are proposed. The existing BP training method can only train the weights and thresholds of the BP network. The structure of the network must be determined in advance by some method. Using a well-designed hierarchical genetic algorithm, the structure, weights and thresholds of the network can be determined simultaneously by training. Compared with the traditional BP network prediction model, the results show that the prediction accuracy of the model is satisfactory and the proposed method is feasible.