提高预测精度的ELMAN和SOM神经网络组合

Using combination of ELMAN and SOM neural networks to enhance prediction precision

针对预测样本数量有限的问题,提出了对训练样本和要预测的样本先聚类、后分别训练和预测的方法。利用网络特性,对复杂信息进行预先分类,使后续信息处理和映射更精确迅速,采用ELMAN神经网络和SOM神经网络的组合提高预测精度。通过对天气和疾病的预测仿真实验表明,该方法增强了网络的局部泛化能力,预测精度高于BP网络和单一采用EMAN网络或SOM网络的精度。

Combination of ELMAN and SOM neural networks can used to enhance the prediction precision. A new method of training and predicting of samples is developed. In this new method, the training and predicting are divided to two steps: clustering at first and then train and predict the samples at the clustered areas. This method is applied to weather and disaster prediction. Simulation results show that this method improved the ability of local generalization of the network and the prediction precision is higher than normal BP network or just one of ELMAN and SOM networks.