吸收光谱法快速检测长江水COD的方法  

Method for rapid detection of COD in Yangtze River waterby absorption spectrometry

作　　者：郑培超 吕强[1,2] 王金梅 曾金锐[1,2] 李成林 杨琴[1,2] ZHENG Peichao;LYU Qiang;WANG Jinmei;ZENG Jinrui;LI Chenglin;YANG Qin(School of Optoelectronic Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,P.R.China;Chongqing Municipal Level Key Laboratory of Photoelectronic Information Sensing and Microsystems,Chongqing University of Posts and Telecommunications,Chongqing 400065,P.R.China)

机构地区：[1]重庆邮电大学光电工程学院 [2]重庆邮电大学光电信息感测与微系统重庆市重点实验室,重庆400065

出　　处：《重庆邮电大学学报(自然科学版)》2025年第1期29-36,共8页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家自然科学基金项目(32171627);重庆市教委科学技术研究项目重大项目(KJZD-M202200602)。

摘　　要：为满足地表水化学需氧量的高精度检测,提出了一种基于紫外-可见吸收光谱结合连续投影-支持向量机回归的预测模型。以长江水的吸收光谱为研究对象,采用SPXY划分出训练集和测试集,使用Savitzky-Golay滤波对数据进行预处理,采用连续投影算法筛选出特征波长,利用支持向量机回归拟合特征波长与样本浓度,建立长江水化学需氧量的回归模型。以决定系数R2、均方根误差RMSE、平均相对误差作为模型评价指标,与其他5种混合模型的预测效果进行对比显示,采用连续投影-支持向量机回归所建立的模型结果最优,实现了地表水化学需氧量的快速、准确测量。To meet the high-precision detection requirements of chemical oxygen demand(COD)in surface water,this paper proposes a prediction model based on ultraviolet-visible absorption spectra combined with continuous projection-support vector machine(CP-SVM)regression.Using the absorption spectra of Yangtze River water as the research object,the SPXY method is used to divide the data into training and testing sets.Savitzky-Golay filtering is applied for data preprocessing,and the continuous projection algorithm is used to select characteristic wavelengths.Support vector machine regression is then used to fit the relationship between the selected characteristic wavelengths and sample concentrations,establishing a regression model for the chemical oxygen demand of Yangtze River water.The model’s performance is evaluated using the coefficient of determination(R 2),root mean square error(RMSE),and mean relative error,and compared with the prediction results of five other hybrid models.The model based on continuous projection-support vector machine regression shows the best performance,achieving rapid and accurate measurement of surface water COD.

关 键 词：光谱学 吸收光谱 地表水 水质检测 化学需氧量 

分 类 号：O433.4[机械工程—光学工程] TP73[理学—光学]