盐湖水化学类型的人工神经网络判别方法

Artificial Neural Network (ANN) Method for Predicting of Hydrochemical Types of Salt Lakes

研究了作为典型径向基函数网络之一的概率神经网络在盐湖水化学类型分类预测中的应用,验证了该方法的可靠性,得到了满意的分类预测结果。实验结果和网络结构分析表明,概率神经网络方法比熟知的反向传播算法(BP)网络要好。概率神经网络的研究应用为化学模式识别提供了一个新工具。

The prediction of hydrochemical types of salt lakes by probability artificial neural network model, which is one of the typical Radial basis function networks was studied. The good classing and predicting results were obtained. The average accuracy for predicting of hydrochemical types of salt lakes was 91.0%. Both the experimental results and structure analysis of neural networks indicated that the probability artificial neural network method is much better than the back-propagation (BP) neural network method . In fact, this study provides a new tool for chemical pattern recognition.