W-RBF神经网络模型在中药材价格预测中的应用

Application of W-RBF neural network model in Chinese herbal medicine price forecast

中药材作为一种特殊的农产品,其价格态变化过程呈现出高度复杂的非线性特征,增加了中药材价格预测的难度.通过研究影响中药材价格的主要因素,并在分析传统价格预测方法的基础上,针对中药材价格变化具有随机性、突变性和非线性的特点,通过小波和RBF神经网络结合构建一个中药材价格预测模型(W-RBF).采用W-RBF神经网络模型对白芍价格进行预测,并将预测结果与RBF神经网络模型预测结果做了比较.实验结果说明,W-RBF神经网络模型预测准确率明显提高,比RBF神经网络模型更具优越性.

Chinese herbal medicine is a special agricultural product, and its price state change shows the characteristics of highly complicated nonlinear process, which increases the difficulty in predicting its price. This paper studies the main price-influencing factors, analyzes the traditional price prediction method, and, taking into consideration the fact that the price changes randomly, suddenly and non-linearly, builds a Chinese herbal medicine price forecast model which combines wavelet with the RBF neural network（W-RBF）. The W-RBF neural network model makes a forecast to the price of radix paeoniae alba, and compares the prediction results to the RBF neural network model predictive results. The analysis results show that the W-RBF neural network model is superior to RBF neural network model, and that the predictions are obviously improved in accuracy.