摘要
本研究提出了一种基于遗传算法(Genetic algorithm,GA)和BP神经网络(Back propagation neural network,BPNN)的复合模型——GABP模型,以安徽省池州市某矿区及其周边为研究区,预测了土壤中p H和7种重金属元素(Cd、Pb、Cr、Cu、Ni、Hg、As)含量的空间分布,并与BPNN和反比距离权重法(Inverse distance weighting,IDW)进行了比较。研究结果表明:受采矿活动影响,研究区土壤p H和重金属含量呈显著的空间分异性;GABP复合模型的数据扩增能够有效弥补BPNN对样本数量的依赖,同时结合了地理位置和高程属性,精度评价结果显示GABP模型的平均R^(2)、r、RMSE、MAE分别是IDW和BPNN的3.03倍、2.56倍,2.93倍、2.39倍,0.85倍、0.61倍,0.79倍、0.62倍,预测精度更高。模型解决了传统空间插值方法结果中可能出现负值和边界无法插值的问题,为土壤重金属含量空间分布预测提供了一种新方法。
Based on genetic algorithm(GA)and back propagation neural network(BPNN),this study proposed a composite model:GABP model.Focusing on a mining area and its surroundings in Chizhou City,Anhui Province,the spatial distribution of soil pH value and the concentrations of seven heavy metals(Cd,Pb,Cr,Cu,Ni,Hg and As)were predicted by GABP model,and the prediction results were compared with those of BPNN and inverse distance weighting(IDW)method.The results showed that,influenced by mining activities,there was significant spatial heterogeneity in soil pH value and heavy metal concentrations in the study area.The data augmentation of GABP model effectively compensated for the dependency of BPNN on the sample size,and simultaneously incorporated geographical location and elevation attributes.The precision evaluation results indicated that the average R^(2),r,RMSE and MAE of GABP model was 3.03 times and 2.56 times,2.93 times and 2.39 times,0.85 times and 0.61 times,0.79 times and 0.62 times higher than those of IDW and BPNN,respectively,indicating a higher predictive accuracy.The proposed model can solve the issues in traditional spatial interpolation methods where negative values and boundary interpolation difficulties may occur,and provides a novel approach for predicting the spatial distribution of soil heavy metal contents.
作者
赵萍
阮旭东
刘亚风
赵思逸
孙雨
常杰
周俊
ZHAO Ping;RUAN Xudong;LIU Yafeng;ZHAO Siyi;SUN Yu;CHANG Jie;ZHOU Jun(School of Resources and Environmental Engineering,Hefei University of Technology,Hefei 230009,China;School of Resources and Environment,Anqing Normal University,Anqing,Anhui 246133,China)
出处
《土壤》
CAS
CSCD
北大核心
2024年第4期889-896,共8页
Soils
基金
国家自然科学基金项目(41972304)资助。
