基于灰色关联和BP神经网络的地下水PH值预测研究  

Prediction of Groundwater PH Value Based on Grey Correlation and BP Neural Network

作　　者：孙国营[1] 孙新杰[1] 陕振沛[1] 龙尚银 SUN Guo-ying;SUN Xin-jie;SHAN Zhen-pei;LONG Shang-yin(Department of Mathematics and Information Engineering,Liupanshui Normal College,Liupanshui 553004,China)

机构地区：[1]六盘水师范学院数学与计算机科学学院,贵州六盘水553004

出　　处：《数学的实践与认识》2020年第14期147-155,共9页Mathematics in Practice and Theory

基　　金：贵州省教育厅自然科学基金资助(黔教合KY字[2017]260号);六盘水市科技计划项目(5202020180415);贵州省教育厅基金项目(黔教合KY字[2018]387号);六盘水师范学院教学内容与课程体系改革项目(LPSSYjg201917)。

摘　　要：通过熵值法确定地下水各指标的权重,通过平均值灰色关联法以及权重灰色关联法求解各指标与PH值之间的关联度,然后通过BP神经网络算法,选取关联度排名前3,6和9的指标预测地下水的PH值,并和通过选取所有指标预测的地下水PH值进行比较,发现结合灰色关联算法的预测值要比传统BP神经网络算法的预测值更精确,而且选取指标数不同,结合平均值灰色关联法和结合权重灰色关联法的BP神经网络算法预测的精确度和优劣程度也略有不同.The entropy method is used to determine the weight of each indicator of groundwater.The correlation between the indicators and the PH value is solved by the mean gray correlation method and the weight grey correlation method.Then,the BP neural network algorithm is used to select the top 3,6 and The index of 9 predicts the pH value of groundwater and compares it with the groundwater pH value predicted by selecting all indicators.It is found that the predicted value combined with the gray correlation algorithm is more accurate than the predicted value of the traditional BP neural network algorithm,and the number of selected indicators is different.The accuracy and pros and cons of BP neural network algorithm combined with mean gray correlation method and weighted gray correlation method are also slightly different.

关 键 词：熵值法 PH值 平均值灰色关联法 权重灰色关联法 BP神经网络 

分 类 号：P641.12[天文地球—地质矿产勘探] TP183[天文地球—地质学]