人民黄河2024，Vol.46Issue(1)：109-113,5.DOI:10.3969/j.issn.1000-1379.2024.01.019
一种基于多分类器和证据理论融合的水质分类方法
A Water Quality Classification Method Based on the Fusion of Multiple Classifiers and Dempster-Shafer Theory
摘要
Abstract
A water quality classification method based on the fusion of multiple classifiers and evidence theory was proposed to address the is-sues of uneven recognition,low accuracy and poor adaptability of single classifiers for different water quality categories.This method selected three classifiers of deep neural network classifier,improved support vector machine classifier and Bayesian classifier.The reliability function is built by the full probability formula and based on evidence theory,the reliability function was fused to obtain a multi classifier fusion mod-el.It selected 3,558 pieces of water quality data from March 1-22,2022 released by the National Surface Water Quality Automatic Station as the sample set and used DNN water quality classification model,PSO-SVM water quality classification model,Bayesian water quality classifi-cation model and multi-classifier fusion model to test the samples.The results show that the average accuracy,precision,recall and F1 values of the multi classifier fusion model for water quality classification are 94.2%,93.8%,94.2%and 94.0%respectively.Compared to the DNN water quality classification model,PSO-SVM water quality classification model and Bayesian water quality classification model,the accuracy of multi-classifier fusion model has been improved by 5.6%,9.8%and 13.6%respectively,the precision by 5.2%,10.0%and 10.9%re-spectively,the recall by 5.6%,9.8%and 13.6%respectively and the F1 values by 5.4%,10.2%and 12.3%respectively.The multi classi-fier fusion model has better accuracy and adaptability in water quality classification.关键词
水质分类/多分类器/神经网络/证据理论融合Key words
water quality classification/multiple classifiers/neural network/integration of evidence theory分类
资源环境引用本文复制引用
项新建,颜超龙,费正顺,郑永平,李可晗..一种基于多分类器和证据理论融合的水质分类方法[J].人民黄河,2024,46(1):109-113,5.基金项目
浙江省自然科学基金资助项目(LY19F030004,LQ16F030002) （LY19F030004,LQ16F030002）
浙江省重点研发计划项目(2018C01085) （2018C01085）
