Research on the chemical oxygen demand spectral inversion model in water based on IPLS-GAN-SVM hybrid algorithm DOI Creative Commons
Qirong Lu, Jian Zou,

Yingya Ye

et al.

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(4), P. e0301902 - e0301902

Published: April 11, 2024

Spectral collinearity and limited spectral datasets are the problems influencing Chemical Oxygen Demand (COD) modeling. To address first problem obtain optimal modeling range, spectra preprocessed using six methods including Standard Normal Variate, Savitzky-Golay Smoothing Filtering (SG) etc. Subsequently, 190–350 nm range is divided into 10 subintervals, Interval Partial Least Squares (IPLS) used to perform PLS on each interval. The results indicate that it best modeled in 7th (238~253 nm). values of Mean Square Error (MSE), Absolute (MAE) R2score model without pretreatment 1.6489, 1.0661, 0.9942. After pretreatment, SG better than others, with MSE MAE decreasing 1.4727, 1.0318 improving 0.9944. Using model, predicted COD for three samples 10.87 mg/L, 14.88 19.29 mg/L. small dataset, Generative Adversarial Networks data augmentation, obtained Support Vector Machine (SVM) that, compared original SVM’s have decreased, while its accuracy has improved by 2.88%, 11.53%, 18.07%, 17.40%, 18.74%.

Language: Английский

Substantial increase of organic carbon storage in Chinese lakes DOI Creative Commons

Dong Liu,

Kun� Shi, Peng Chen

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: Sept. 14, 2024

Language: Английский

Citations

11
Eutrophication exacerbated organic pollution in lakes across China during the 1980s–2010s DOI

Dong Liu,

Chenxue Zhang,

Nuoxiao Yan

et al.

Water Research, Journal Year: 2024, Volume and Issue: 268, P. 122782 - 122782

Published: Nov. 10, 2024

Language: Английский

Citations

6
Health of plateau soil environment: Corresponding relationship of heavy metals in different land use/cover types (LULCC) DOI

Zhenghui Fu,

Yong Liu,

Xia Jiang

et al.

The Science of The Total Environment, Journal Year: 2025, Volume and Issue: 973, P. 179162 - 179162

Published: March 23, 2025

Language: Английский

Citations

0
Methods for molecular characterization of dissolved organic matter in the alpine water environment: an overview DOI Creative Commons
Yongbao Zhang, Jianqing Du, Kang Xiao

et al.

Frontiers in Environmental Chemistry, Journal Year: 2024, Volume and Issue: 5

Published: Jan. 18, 2024

The alpine area has become a sensitive indicator and amplifier of global climate change human activities because its unique geographical climatic conditions. Being an essential part biochemical cycling, dissolved organic matter (DOM) could effectively help understand the process, structure, function aquatic ecosystems. Due to low content sampling difficulties, analysis DOM in water demands high sensitivity with sample volume, which not been comprehensively reviewed. This review summarizes sampling, pretreatment, methods involving characterization concentration, spectroscopy, molecular structure. Overall, conventional parameters are basis advanced methods. Spectroscopic tests can reveal optical properties response lights from ultraviolet infrared wavelengths, distinguish chemical composition. Molecular structure characterizations provide microscopic information such as functional groups, element ratios, weights. combination multiple depict composition perspectives. In sum, optimized high-sensitivity characterization, method integration crucial for analyzing components waters. These perspectives standardize process correlation between properties, well migration transformation DOM.

Language: Английский

Citations

1
Research on the chemical oxygen demand spectral inversion model in water based on IPLS-GAN-SVM hybrid algorithm DOI Creative Commons
Qirong Lu, Jian Zou,

Yingya Ye

et al.

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(4), P. e0301902 - e0301902

Published: April 11, 2024

Spectral collinearity and limited spectral datasets are the problems influencing Chemical Oxygen Demand (COD) modeling. To address first problem obtain optimal modeling range, spectra preprocessed using six methods including Standard Normal Variate, Savitzky-Golay Smoothing Filtering (SG) etc. Subsequently, 190–350 nm range is divided into 10 subintervals, Interval Partial Least Squares (IPLS) used to perform PLS on each interval. The results indicate that it best modeled in 7th (238~253 nm). values of Mean Square Error (MSE), Absolute (MAE) R2score model without pretreatment 1.6489, 1.0661, 0.9942. After pretreatment, SG better than others, with MSE MAE decreasing 1.4727, 1.0318 improving 0.9944. Using model, predicted COD for three samples 10.87 mg/L, 14.88 19.29 mg/L. small dataset, Generative Adversarial Networks data augmentation, obtained Support Vector Machine (SVM) that, compared original SVM’s have decreased, while its accuracy has improved by 2.88%, 11.53%, 18.07%, 17.40%, 18.74%.

Language: Английский

Citations

1