基于近红外光谱技术的COD快速无损定量预测模型研究

扫码查看

原文链接

NETL
NSTL
万方数据

中文摘要：本研究旨在开发一种基于近红外光谱技术的快速无损定量预测模型,以提高水体中化学需氧量(COD)的监测效率.实验采用傅里叶变换近红外光谱仪,处理和分析了 1 719条光谱数据,构建了 1D-3Conv-CNN模型.经过数据预处理和模型优化,该模型在全谱波段下表现优异,训练集和测试集的决定系数(R-)分别为0.992 1和0.990 6,均方根误差(RMSE)分别为27.47 mg/L和29.93 mg/L,平均绝对误差(MAE)分别为22.36 mg/L和24.58 mg/L.结果表明,该模型能够准确反映COD浓度,具有较高的预测精度和实际应用潜力.

外文标题：Research on the Fast and Non Destructive Quantitative Prediction Model of COD Based on Near Infrared Spectroscopy Technology

外文摘要：This study aims to develop a rapid and non-destructive quantitative prediction model based on near-infrared spectroscopy technology to improve the monitoring efficiency of chemical oxygen demand(COD)in water bodies.The experiment used a Fourier transform near-infrared spectrometer to process and analyze 1719 spectral data,and constructed a 1D-3Conv-CNN model.After data preprocessing and model optimization,the model performed excellently in the full spectrum band,with determination coefficients(R2)of 0.992 1 and 0.990 6 for the training and testing sets,root mean square errors(RMSE)of 27.47 mg/L and 29.93 mg/L,and mean absolute errors(MAE)of 22.36 mg/L and 24.58 mg/L,respectively.The results indicate that the model can accurately reflect COD concentration,with high prediction accuracy and practical application potential.

外文关键词：

near infrared spectroscopy technologyCOD concentration1D-3Conv-CNN modelnon destructive quantification

作者：

祝凯良

展开 >

作者单位：

华熙生物科技股份有限公司,山东济南 250000

关键词：

近红外光谱技术 COD浓度 1D-3Conv-CNN模型无损定量

出版年：

2024

DOI：