基于灰色关联和ABC-BP神经网络的叶绿素a浓度预测  

Prediction of Chlorophyll-a Concentration Based on Grey Relational Analysis and ABC-BP Neural Network

作　　者：胡志洋 李翠梅[1] 薛天一 HU Zhi-yang;LI Cui-mei;XUE Tian-yi(School of Environmental Science and Engineering,Suzhou University of Science and Technology,Suzhou 215009,China)

机构地区：[1]苏州科技大学环境科学与工程学院,江苏苏州215009

出　　处：《水电能源科学》2021年第1期55-58,共4页Water Resources and Power

基　　金：国家自然科学基金项目(51109153)。

摘　　要：为预测太湖梅梁湾叶绿素a浓度,建立了基于灰色关联和ABC-BP神经网络的叶绿素a浓度预测模型,即通过灰色关联分析选取总磷、CODMn、水温、pH值、悬浮质为BP神经网络的输入变量,采用人工蜂群(ABC)算法优化网络权值与阈值,构造基于ABC算法优化的BP(ABC-BP)神经网络模型,预测出2014年1月~2015年12月的梅梁湾叶绿素a浓度。结果表明,经ABC算法优化后,BP网络模型预测梅梁湾叶绿素a浓度的最大绝对误差从3.54μg/L减小到1.28μg/L,最大相对误差从41.57%减小到20.62%,平均相对误差从8.83%减小到3.31%,可以提高梅梁湾叶绿素a浓度预测的准确性。The method based gray correlation analysis and ABC-BP neural network was adopted to predict the chlorophyll-a concentration in Meiliang Bay of Taihu Lake.The ABC-BP model was established to predict the chlorophyll-a concentration in Meiliang Bay from January 2014 to December 2015.The total phosphorus,CODMn,water temperature,the value of pH,and suspended solids were selected as the input variables of the BP neural network through gray correlation analysis,and the artificial bee colony(ABC)algorithm was used to optimize the BP weights and thresholds.The results show that after the optimization of ABC algorithm,the maximum absolute error of chlorophyll-a concentration in Meiliang Bay decreases from 3.54μg/L to 1.28μg/L,the maximum relative error decreases from 40.57%to 20.62%,and the average relative error decreases from 8.83%to 3.31%,which can improve the accuracy of chlorophyll-a concentration prediction in Meiliang Bay.

关 键 词：叶绿素A浓度 灰色关联分析 人工蜂群算法 BP神经网络 预测 

分 类 号：X824[环境科学与工程—环境工程]