论文部分内容阅读
社会物流总成本(WLCB)序列同时具有线性和非线性的特征,直接对WLCB预测,传统预测、神经网络方法均产生很大的误差。本文提出:对WLCB数据作两步预处理,逐步消除WLCB的线性特征,在只具非线性特征的预处理后数据基础上,建立BP预测模型。分别与建立在原始数据及只作一步预处理的数据的BP预测模型进行比较,实验表明:两步预处理后不含线性特征的BP模型预测准确率大大提高,从而证实了改进NN模型用于WLCB预测的有效性。
The total cost of social logistics (WLCB) sequence has both linear and non-linear characteristics and directly causes great errors to the WLCB prediction, the traditional prediction and the neural network method. In this paper, the WLCB data is preprocessed in two steps and the linear features of WLCB are gradually eliminated. Based on the preprocessed data with only nonlinear features, a BP prediction model is established. Compared with the BP prediction model based on the original data and the one-step preprocessing data, the experimental results show that the prediction accuracy of the BP model without linear features after two-step preprocessing is greatly improved, thus confirming the improvement of the NN model for Validity of WLCB prediction.