一种融合?KPCA?和?BP?神经网络的用水总量预测方法

针对影响用水总量的相关用水因子的不确定性和非线性多维特点，论文研究并提出了一种融合KPCA和思维优化BP神经网络的用水总量预测方法。首先运用相关系数法确定预测因子，然后利用核主成分分析（KPCA）对所述预测因子进行降维处理，解决数据之间的非线性特征，最后采用BP神经网络建立用水总量预测模型，同时采用思维进化学习算法优化BP神经网络的权值和阈值。该方法在国家统计局的2007-2016年度开放统计用水数据中实验，通过实验比较，该模型的相对预测误差小于5%，结果表明，融合 KPCA和思维优化BP神经网络的用水总量预测模型能很好的预测未来用水总量。

Water consumption prediction approach using KPCA and BP neural network

In response to the uncertainty and non-linear multidimensional characteristics of the related factors affecting the water consumption, this paper studies and proposes an approach for forecasting the water consumption by fusing KPCA and mind evolutionary BP neural network. First, Using Pearson correlation coefficient method to determine predictors, and then using kernel principal component analysis (KPCA) to perform dimension reduction process for predictors, solving the non-linear characteristics between the data. Finally, the water consumption prediction model is established by using BP neural network, at the same time, the mind evolutionary algorithm is used to optimize weights and thresholds of the BP neural network. This approach is tested in the open statistical water consumption data of the national bureau of statistics from 2007 to 2016. Through experimental comparison, the relative prediction error of this model is less than 5%. The result shows that the water consumption prediction model fusing KPCA and mind evolutionary BP neural network can predict the future water consumption well.