基于GA改进FSVM的耕地土壤重金属含量反演  

Inversion of Heavy Metal Content in Cultivated Soil Based on GA Improved Fuzzy Support Vector Machine

作　　者：蒋明 黄栋良 马炎生 雷帆 Jiang Ming;Huang Dongliang;Ma Yansheng;Lei Fan(Hu'nan Vocational College of Engineering,Hu'nan,410151;Hu'nan Engineering Research Center of 3D Real Scene Construction and Application Technology,Hu'nan,410000;Key Laboratory of Natural Resources Monitoring and Supervision in Southern Hilly Region,Ministry of Natural Resources,Hu'nan,410118)

机构地区：[1]湖南工程职业技术学院,湖南410151 [2]实景三维建设与应用技术湖南省工程研究中心,湖南410000 [3]自然资源部南方丘陵区自然资源监测监管重点实验室,湖南410118

出　　处：《当代化工研究》2024年第22期126-128,共3页Modern Chemical Research

基　　金：国家自然科学基金“高温差地区光谱解混反演路域植被空间格局的辐射传输模型研究”(项目编号:41671498);自然资源部南方丘陵区自然资源监测监管重点实验室开放基金课题“近30年湖南省土地利用方式变化的驱动因素及对陆地生态系统碳循环的影响研究”(项目编号:NRMSSHR2022Z02)。

摘　　要：耕地是粮食生产的命根子,而耕地土壤重金属污染问题日益严重,保护耕地,监测耕地土壤重金属污染势在必行。以此提出了一种融合遗传算法的模糊支持向量机回归方法,有效提高了耕地中土壤重金属的预测精度。采用遗传算法优化模糊支持向量机中的参数,利用优化模型预测耕地土壤中重金属铁、锌和铜的含量,并与一般模糊支持向量机预测模型进行对比。结果表明:与一般模糊支持向量机相比,采用遗传算法优化的模糊支持向量机获得了较好的预测结果,模型均方根误差和平均相对误差得到了很大的降低,其中,对重金属锌的预测中,均方根误差总的降低了5.04,平均相对误差总的降低了7.86%。研究表明,遗传算法参数寻优,可以显著提高模糊支持向量机回归模型的预测精度,模型预测效果更好。Farmland is the lifeblood of grain production,and the problem of heavy metal pollution in farmland soil is becoming increasingly serious.It is imperative to protect farmland and monitor heavy metal pollution in farmland soil.This article proposes a fuzzy support vector machine regression method that integrates genetic algorithm,effectively improving the prediction accuracy of soil heavy metals in cultivated land.Using genetic algorithm to optimize the parameters in fuzzy support vector machine,and using the optimization model to predict the content of heavy metals iron,zinc,and copper in cultivated soil,and comparing it with the general fuzzy support vector machine prediction model.The results showed that compared with the general fuzzy support vector machine,the fuzzy support vector machine optimized by genetic algorithm obtained better prediction results,and the root mean square error and average relative error of the model were greatly reduced.Among them,in the prediction of heavy metal zinc,the total root mean square error was reduced by 5.04,and the total average relative error was reduced by 7.86%.Research has shown that optimizing genetic algorithm parameters can significantly improve the prediction accuracy of fuzzy support vector machine regression models,resulting in better model prediction performance.

关 键 词：模糊支持向量机 遗传算法 土壤重金属 参数寻优 定量反演 

分 类 号：X[环境科学与工程]