Method for predicting ozone dosage and COD (Chemical Oxygen Demand) of leather wastewater based on ultraviolet-visible spectrum

Abstract

The invention relates to the technical field of leather wastewater treatment, and provides an ultraviolet-visible spectrum-based leather wastewater ozone dosage and COD (Chemical Oxygen Demand) prediction method. The method comprises the following steps: firstly, determining the types of organic matters in a tannery wastewater sample, and determining the distribution characteristics of the organic matters in different fluorescent regions through three-dimensional fluorescence spectrum analysis, so as to discriminate the types of organic pollutants; secondly, screening key wavelengths, and determining characteristic absorption wavelengths representing different organic matters through an ultraviolet-visible light spectrum; then carrying out an ozone oxidation experiment, and synchronously measuring a COD value and an ultraviolet absorption spectrum under different ozone addition amounts; based on experimental data, a multiple linear regression and exponential regression prediction model of ozone dosage and a COD prediction model based on partial least squares regression are established. According to the invention, accurate control of ozone dosage and real-time monitoring of COD are realized, the treatment efficiency is improved, the operation cost is reduced, the adaptability to water quality fluctuation is enhanced, and the method has a wide engineering application prospect.

Description

Technical Field

[0001] This invention relates to leather wastewater treatment technology, and provides a method for predicting ozone dosage and COD in leather wastewater based on ultraviolet-visible spectroscopy. Background Technology

[0002] The leather tanning process releases large amounts of organic waste containing chemical auxiliaries, leather residues, and hair degradation products during processes such as liming, deliming, softening, tanning, and retanning and dyeing. Most of this waste is discharged into the aquatic phase, resulting in large volumes of wastewater, high pollution loads, and complex organic matter profiles. Direct discharge without effective treatment can easily lead to eutrophication and increased color in receiving water bodies, posing potential risks to ecosystems and public health.

[0003] For high-load organic pollutants, ozone oxidation technology has been widely used in the advanced treatment of tannery wastewater due to its strong oxidizing properties and the ability to generate hydroxyl radicals (·OH) in situ. On the one hand, ozone can selectively react with aromatic and unsaturated aliphatic functional groups; on the other hand, under catalytic conditions, it can generate (·OH), which efficiently breaks down various stubborn organic compounds through a non-selective chain radical mechanism. However, in engineering practice, ozone dosage often relies on fixed dosage coefficients or operational experience, making it difficult to adaptively adjust to instantaneous fluctuations in influent water quality. When the influent COD fluctuates beyond a certain range (e.g., >30%) within a short period, the dosage deviation can often reach 40%–60%, easily leading to reagent waste or insufficient oxidation, thus affecting effluent stability and resource utilization efficiency.

[0004] Precise control of the ozone process hinges on timely and accurate sensing of influent water quality. However, traditional laboratory testing methods (such as the potassium dichromate method for COD) typically rely on a multi-step process of sampling, pretreatment, colorimetric development, and colorimetric analysis. This process is time-consuming and labor-intensive, making it difficult to meet the needs of online real-time monitoring. More importantly, high salinity and complex matrices significantly amplify methodological errors: high concentrations of chloride ions interfere with COD measurements, requiring additional dechlorination or masking steps. These factors collectively lead to delayed and inaccurate test results, failing to provide a reliable closed-loop feedback signal for ozone dosing.

[0005] Ultraviolet-visible (UV-Vis) absorption spectroscopy can capture the characteristic absorption of organic matter in water within the approximately 200–600 nm wavelength range, enabling rapid acquisition of pollutant fingerprints. This technology requires no chemical reagents, allows for continuous online sampling, and offers advantages such as fast response and rich information dimensions. Through multivariate modeling and algorithmic processing (such as characteristic wavelength extraction, spectral preprocessing, and regression modeling), UV-Vis spectral signals can indirectly characterize conventional COD indicators, providing a feasible path for process monitoring of complex wastewater. However, currently, spectral monitoring devices are mostly independent of ozone dosing systems, lacking data fusion and linkage control mechanisms: spectral data is difficult to convert into executable dosing commands in a timely manner, and the system's response to sudden water quality fluctuations is lagging, making it difficult to achieve adaptive optimal control for target water quality.

[0006] In summary, the large fluctuations in water quality, complex composition, and interference from high-salinity matrices in tannery wastewater make it difficult to achieve stable and economical treatment results using traditional parallel monitoring and empirical dosing methods. Therefore, there is an urgent need for a technical solution that deeply couples online UV-Vis spectral monitoring with ozone dosing control: on the one hand, a robust prediction model for COD indicators is constructed based on spectral data, enabling real-time sensing of changes in influent load and oxidizability; on the other hand, the prediction results are combined with process control strategies to adjust the ozone dosage in real time, achieving rapid response and adaptive adjustment to sudden fluctuations, thereby improving effluent stability and reducing reagent and energy costs. Based on this need, this invention proposes an ozone dosage and COD prediction method based on ultraviolet spectral analysis to solve the problems of monitoring-control disconnect, inaccurate dosing, and high operating costs in existing technologies. Summary of the Invention

[0007] To address the problems existing in the background technology, this invention provides a method for predicting ozone dosage and COD in leather wastewater based on ultraviolet-visible spectroscopy, comprising the following steps:

[0008] S1, Organic matter classification and feature identification: Collect leather tanning wastewater samples, use a three-dimensional fluorescence spectrometer to scan the wastewater samples to obtain three-dimensional fluorescence excitation-emission matrix spectra, and classify and identify organic pollutants according to the five-zone fluorescence zoning method;

[0009] S2, Key wavelength screening: Prepare standard solutions of tyrosine, tryptophan, fulvic acid and humic substances at different concentrations, and use a UV-Vis spectrophotometer to collect UV absorption spectra in the wavelength range of 200-600nm. Extract the characteristic absorption peak positions of each substance and determine the key wavelength as the model input parameters.

[0010] S3, Ozone oxidation experiment; different ozone dosages were set up for gradient experiments, and samples were taken at regular intervals during the ozone oxidation reaction to simultaneously measure the COD value at each time point and collect ultraviolet absorption spectrum data.

[0011] S4, Construction of ozone dosage prediction model: The ultraviolet absorption spectrum data obtained from the ozone oxidation experiment are processed to extract the absorbance value at the key wavelength and calculate the integral area within a certain wavelength range. A prediction model is established with the absorbance at the key wavelength and the integral area of ​​the fixed band as independent variables and the ozone dosage as the dependent variable.

[0012] S5, COD prediction model construction: COD of collected influent and effluent samples is measured and ultraviolet spectra are obtained by scanning. Characteristic wavelengths are determined, and a COD prediction model is established with absorbance at the characteristic wavelength as the independent variable and COD as the dependent variable.

[0013] In the preferred embodiment, the organic matter classification and feature identification in step S1 specifically includes:

[0014] S11, a three-dimensional fluorescence spectrometer was used to scan the leather tanning wastewater sample to obtain a three-dimensional fluorescence excitation-emission matrix spectrum;

[0015] S12, the fluorescence signal is divided into five regions according to the internationally accepted five-region fluorescence zoning method: Region I, Region II, Region III, Region IV, and Region V. Region I contains tyrosine proteins; Region II contains tryptophan proteins; Region IV contains a mixture of tyrosine and tryptophan proteins. Regions III and V contain humic organic matter, with Region III containing fulvic acids and Region V containing humic substances.

[0016] In a preferred embodiment, the five-zone fluorescence partitioning method specifies that: Zone I has an excitation wavelength Ex of 200-250 nm and an emission wavelength Em of 250-330 nm; Zone II has an excitation wavelength Ex of 200-250 nm and an emission wavelength Em of 330-380 nm; Zone III has an excitation wavelength Ex of 200-250 nm and an emission wavelength Em of 380-550 nm; Zone IV has an excitation wavelength Ex of 250-400 nm and an emission wavelength Em of 330-380 nm; and Zone V has an excitation wavelength Ex of 250-400 nm and an emission wavelength Em of 380-550 nm.

[0017] In a preferred embodiment, the key wavelength screening in step S2 specifically includes:

[0018] S21, prepare standard solutions of tyrosine, tryptophan and fulvic acid at different concentrations respectively, and collect ultraviolet absorption spectra in the wavelength range of 200-600nm using a UV-Vis spectrophotometer.

[0019] S22, extract the characteristic absorption peaks of tyrosine at 225nm and 272nm, extract the characteristic absorption peaks of tryptophan at 225nm and 275nm, and extract the characteristic absorption peak of fulvic acid at 280nm.

[0020] S23, 254nm was selected as the ultraviolet absorption wavelength to represent humic substances;

[0021] S24, comprehensively determined five key wavelengths of 225nm, 254nm, 272nm, 275nm and 280nm as model input parameters.

[0022] In the preferred embodiment, the ozone oxidation experiment in step S3 specifically includes:

[0023] S31, ozone dosage set at 15-40mg / min, gradient 5mg The ozone was continuously introduced into a 1L tannery wastewater sample at a constant gas flow rate, with the rate increasing by 1min, and the reaction time was set to 30 minutes.

[0024] S32, during the ozone introduction reaction process, samples were taken at regular intervals, and the COD value at each sampling time point was determined by the potassium dichromate method;

[0025] S33, simultaneously acquires ultraviolet spectra in the wavelength range of 200-600nm with a wavelength interval of 0.5nm, and repeats the measurement 3 times for each group of samples and takes the average value.

[0026] In the preferred embodiment, the ozone dosage prediction model construction in step S4 specifically includes:

[0027] S41. Import the UV absorption spectrum data obtained within the first 16 minutes of the reaction into Origin software, and use the five-point extremum method to extract the absorbance values ​​at five wavelengths: 225nm, 254nm, 272nm, 275nm and 280nm.

[0028] S42, the absorbance data and the corresponding ozone dosage are labeled to form a dataset, and a multiple linear regression model is used for fitting to establish a multiple linear regression prediction model;

[0029] S43, area integration is performed on the 230-300nm band region, and an exponential regression model is fitted with the integrated area as the independent variable and the ozone dosage as the dependent variable to establish an exponential regression prediction model.

[0030] In the preferred embodiment, the mathematical formula of the multiple linear regression prediction model is:

[0031] ;

[0032] in: The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value is at a wavelength of 280 nm; -3868.7 is a constant term, and 7884.7, -1973.9, 6869.0, -1648.6, and -5403.3 are regression coefficients.

[0033] In the preferred embodiment, the mathematical formula of the exponential regression prediction model is:

[0034] ;

[0035] in, Ozone dosage (mgO3 / min); It is a natural constant; The integral area in the 230-300nm band; is the coefficient of exponential regression.

[0036] In the preferred embodiment, the COD prediction model construction in step S5 specifically includes:

[0037] S51, COD was measured on 28 influent and 56 effluent water samples, and ultraviolet spectra in the range of 200-600nm were obtained by scanning.

[0038] S52, the five-point extreme value method was used to identify the peak positions and extract features of the absorption spectrum curves of the samples. The frequency of the absorption peaks of all samples was statistically analyzed, and 10 characteristic wavelengths were identified: 225nm, 227nm, 242nm, 249nm, 254nm, 267nm, 272nm, 275nm, 279nm and 280nm. During ozone treatment, the absorbance of these wavelengths showed a significant decrease or tended to disappear, indicating that it has a good response to the degradation of organic matter.

[0039] S53, the absorbance values ​​at 10 characteristic wavelengths are labeled with the actual measured COD data to construct a dataset, and a COD prediction model is established using partial least squares regression.

[0040] In the preferred embodiment, the mathematical formula for the COD prediction model is:

[0041] ;

[0042] in, Chemical oxygen demand (COD) ); The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 227 nm; The absorbance value at a wavelength of 242 nm; The absorbance value at a wavelength of 249 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 267 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value at a wavelength of 279 nm; is the absorbance value at a wavelength of 280 nm; -32.34733 is a constant term, and 133.8412, -96.835106, 72.941815, 4.9676696, 45.647989, -286.48653, -320.1749, 360.6204, 202.139048, and 53.598849 are partial least squares regression coefficients.

[0043] The beneficial effects achieved by this invention are as follows:

[0044] This invention collects wastewater samples from different treatment stages of ozone oxidation pilot experiments, obtains their absorption spectra using ultraviolet-visible spectroscopy, and constructs a mapping model between ultraviolet absorption intensity and ozone demand. During operation, the system can monitor the spectral characteristics of the wastewater in real time, predict the ozone dosage based on the model, and dynamically adjust the dosage strategy to achieve dual optimization of oxidation efficiency and economy. Furthermore, this invention introduces a simultaneous COD prediction mechanism. By collecting wastewater samples from multiple time periods and combining measured COD values ​​with their corresponding absorbance data, a quantitative correlation model between ultraviolet spectra and COD concentration is established, forming a real-time perception of changes in influent load and oxidizability. This invention improves ozone utilization, reduces operating costs, and enhances the system's adaptability to complex influent conditions, demonstrating good engineering promotion value and practical prospects; specifically:

[0045] First, this invention establishes an organic matter classification and identification system combining three-dimensional fluorescence spectroscopy and ultraviolet-visible spectroscopy, achieving precise classification and feature extraction of complex organic pollutants in tanning wastewater. This technical solution employs a five-zone fluorescence partitioning method to classify organic pollutants into five major categories: tyrosine-like proteins, tryptophan-like proteins, fulvic acid-like substances, mixed tyrosine-like and tryptophan-like proteins, and humic substances. It then combines this with ultraviolet absorption spectroscopy to extract characteristic wavelength information for each type of substance. This classification and identification method comprehensively reflects the composition, structure, and concentration distribution of organic pollutants in tanning wastewater, providing a scientific theoretical basis for subsequent precise treatment and effectively solving the technical challenge of accurately identifying complex organic components using traditional methods.

[0046] Secondly, this invention establishes a dual ozone dosage prediction model based on multiple linear regression and exponential regression. By screening five key characteristic wavelengths—225nm, 254nm, 272nm, 275nm, and 280nm—a quantitative relationship between ultraviolet absorption spectral characteristics and ozone demand is constructed. The multiple linear regression model uses the absorbance values ​​of multiple characteristic wavelengths for comprehensive prediction, while the exponential regression model reflects the overall organic matter load based on the integral area in the 230-300nm band. This parallel dual-model prediction mechanism adapts to the treatment needs of different types and concentrations of tannery wastewater, significantly improving the accuracy and adaptability of ozone dosage and avoiding the problems of reagent waste or insufficient oxidation caused by traditional empirical dosage methods.

[0047] Third, this invention establishes a COD prediction model based on partial least squares regression. By extracting absorbance information at 10 characteristic wavelengths, it achieves rapid and accurate prediction of chemical oxygen demand. This model fully considers the correlation and complementarity between different wavelengths, effectively handles multicollinearity issues in spectral data, and improves the stability and generalization ability of the prediction model. This online COD prediction technology provides a real-time means of assessing pollution load in wastewater treatment processes, enabling operators to promptly grasp the treatment effects and providing important technical support for optimizing process parameters and ensuring stable compliance of effluent quality.

[0048] Fourth, this invention establishes a predictive formula system with clear physical meaning and high accuracy through statistical analysis and mathematical modeling of a large amount of systematic experimental data. The multiple linear regression formula, through weighted calculation of absorbance values ​​at five key wavelengths, directly reflects the contribution of different types of organic matter to ozone demand through the positive or negative sign of its regression coefficients, providing operators with a clear physical interpretation basis. The exponential regression formula, based on the nonlinear relationship between the integral area and ozone dosage, accurately describes the exponential growth characteristics of ozone demand as organic matter concentration increases, avoiding prediction bias in linear models under high concentration conditions. The COD prediction formula effectively handles the correlation problem between multiple characteristic wavelengths through partial least squares regression, improving the model's stability and prediction accuracy, and providing a reliable mathematical tool for quality control in wastewater treatment processes.

[0049] Fifth, the technical solution combining spectral analysis and prediction models provided by this invention enables intelligent control and real-time monitoring of the ozone oxidation treatment process for tannery wastewater. This method features fast response, simple operation, and low cost. It dynamically adjusts the ozone dosing strategy based on real-time changes in influent water quality, thereby improving treatment efficiency while reducing operating costs. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the three-dimensional fluorescence partitioning of the influent water sample from the ozone catalytic oxidation tank of the tannery wastewater treatment plant obtained in Example 1;

[0051] Figure 2 The images show the UV-Vis spectra of different concentrations of organic compounds collected in Example 1.

[0052] Figure 3 The graph shows the COD change under different ozone dosages and the ultraviolet spectrum of the reaction endpoint in the small-scale ozone oxidation experiment of Example 1.

[0053] Figure 4 This is a comparison chart of ozone prediction and actual values ​​based on a multiple linear regression model in Example 1.

[0054] Figure 5 This is a fitting graph of the ozone dose index regression model based on the integral area in Example 1;

[0055] Figure 6 This is a comparison chart of the predicted and actual COD concentration values ​​based on the PLSR model in Example 1. Detailed Implementation

[0056] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Reference Figures 1-6 This invention provides a method for predicting ozone dosage and COD in leather wastewater based on ultraviolet-visible spectroscopy. By establishing a quantitative relationship between ultraviolet spectral characteristics and ozone oxidation demand, it achieves accurate prediction of ozone dosage and real-time monitoring of chemical oxygen demand during the treatment of leather wastewater. This method fully utilizes the characteristic absorption differences of different organic pollutants in the ultraviolet region, combined with advanced spectral analysis techniques and mathematical modeling methods, providing scientific and technical support for the efficient treatment of leather wastewater.

[0058] Step S1: Organic matter classification and feature identification;

[0059] First, representative samples of tanning wastewater were collected, encompassing typical water quality characteristics observed during the tanning production process. Tanning wastewater contains a large amount of protein degradation products, tanning agent residues, dye auxiliaries, and various organic additives. These substances exhibit different absorption characteristics in the ultraviolet-visible region, providing a rich source of information for spectral analysis.

[0060] Three-dimensional fluorescence spectroscopy was used to scan tanning wastewater samples to obtain three-dimensional fluorescence excitation-emission matrix spectra. The excitation wavelength scanning range was set to 200 nm to 400 nm, and the emission wavelength scanning range was set to 250 nm to 550 nm, with a preferred scanning interval of 5 nm, to obtain complete three-dimensional fluorescence excitation-emission matrix spectra. Three-dimensional fluorescence spectroscopy provides information in both excitation and emission wavelength dimensions, offering higher selectivity and resolution compared to traditional two-dimensional spectroscopy, effectively distinguishing organic compounds with similar structures but different fluorescence properties.

[0061] Fluorescence signals were partitioned into five regions according to the internationally accepted five-region fluorescence zoning method: Region I, Region II, Region III, Region IV, and Region V, corresponding to tyrosine-like proteins, tryptophan-like proteins, fulvic acid-like substances, a mixture of tyrosine and tryptophan-like proteins, and humic substances, respectively. Among these:

[0062] The excitation wavelength Ex in region I is 200-250nm and the emission wavelength Em is 250-330nm;

[0063] The excitation wavelength Ex in region II is 200-250 nm and the emission wavelength Em is 330-380 nm.

[0064] The excitation wavelength Ex in region III is 200-250 nm and the emission wavelength Em is 380-550 nm.

[0065] The excitation wavelength Ex in region IV is 250-400 nm and the emission wavelength Em is 330-380 nm;

[0066] The excitation wavelength Ex in the V region is 250-400nm and the emission wavelength Em is 380-550nm.

[0067] This partitioning method effectively categorizes organic pollutants in complex tanning wastewater, providing a clear theoretical basis for subsequent spectral feature extraction and model construction. The fluorescence intensity of each partition reflects the concentration level of the corresponding type of organic matter. By comparing the fluorescence intensity distribution of different partitions, the compositional characteristics and relative content of organic pollutants in the wastewater can be preliminarily determined.

[0068] Step S2: Key wavelength screening;

[0069] To establish an accurate and reliable prediction model, it is necessary to determine the key characteristic wavelengths through systematic standard solution experiments. Standard solutions of tyrosine, tryptophan, fulvic acid, and humic substances at different concentrations were prepared, and ultraviolet absorption spectra in the wavelength range of 200-600 nm were acquired using a UV-Vis spectrophotometer. The recommended concentration range is 1 mg / L to 7 mg / L.

[0070] Analysis using UV-Vis spectrophotometry revealed characteristic absorption peaks for tyrosine at 225 nm and 272 nm. The absorption at 225 nm was mainly caused by the π→π transition of the benzene ring, while the absorption at 272 nm was related to the n→π transition of the phenolic hydroxyl group. Similarly, characteristic absorption peaks for tryptophan were observed at 225 nm and 275 nm. The absorption at 225 nm, similar to that of tyrosine, was mainly caused by aromatic transitions in the indole ring system, while the absorption at 275 nm reflected the characteristic spectral properties of the indole ring. A characteristic absorption peak for fulvic acid at 280 nm was found, which was primarily related to the phenolic and quinone functional groups in its molecular structure.

[0071] 254 nm was selected as the ultraviolet absorption wavelength to represent humic substances. Due to their complex molecular composition and diverse chemical structures, humic substances are difficult to accurately characterize using a single wavelength. Based on common practices in water quality analysis both domestically and internationally, and statistical analysis of a large amount of experimental data, 254 nm is widely considered an important wavelength for characterizing the ultraviolet absorption of humic substances. The absorbance at 254 nm not only reflects the presence of humic substances but also comprehensively characterizes the overall level of other recalcitrant organic matter in wastewater with similar spectral characteristics.

[0072] Five key wavelengths—225 nm, 254 nm, 272 nm, 275 nm, and 280 nm—were selected as model input parameters. The selection of these five wavelengths considered both the characteristic absorption of various organic compounds and the independence and complementarity between wavelengths, thus comprehensively reflecting the spectral characteristics of organic pollutants in tannery wastewater.

[0073] Step S3: Ozone oxidation experiment;

[0074] Ozone oxidation experiments are a crucial step in establishing predictive models and require the construction of a complete small-scale ozone oxidation experimental setup. This setup mainly includes several core components such as a pure oxygen supply system, an ozone generation system, a concentration monitoring system, a flow control system, a reactor system, and a tail gas treatment system.

[0075] The ozone dosage was set at 15-40 mg O3 / min, increasing in increments of 5 mg O3 / min. Ozone was continuously introduced into a 1 L tannery wastewater sample at a constant gas flow rate for a reaction time of 30 minutes. This dosage range was chosen based on the typical pollution load of tannery wastewater and the economics of ozone oxidation, achieving effective treatment while avoiding resource waste due to excessive ozone use.

[0076] During the ozone introduction reaction, samples were taken at regular intervals, and the COD values ​​at each sampling time point were determined using the potassium dichromate method. The potassium dichromate method is a national standard method for COD determination, offering advantages such as high accuracy and good reproducibility. Mercuric sulfate was added during the reaction as a chloride masking agent to eliminate chloride ion interference. The reaction temperature was controlled at 150℃±2℃, and the reaction time was 2 hours.

[0077] Ultraviolet spectra were simultaneously acquired in the wavelength range of 200-600 nm with wavelength intervals of 0.5 nm. Each sample was measured three times, and the average value was taken. This spectral scanning range covers the characteristic absorption regions of most organic pollutants, and the 0.5 nm wavelength interval provides sufficient spectral resolution. Repeated measurements eliminate the influence of factors such as instrument drift and changes in ambient temperature on the measurement results, improving the reliability and reproducibility of the data.

[0078] Step S4: Construction of ozone dosage prediction model;

[0079] The ozone dosage prediction model was constructed based on a large amount of spectral and water quality data obtained from ozone oxidation experiments. A quantitative relationship between ultraviolet absorption characteristics and ozone demand was established through mathematical modeling.

[0080] The UV absorption spectral data acquired during the initial 16 minutes of the reaction were imported into Origin software, and the absorbance values ​​at five wavelengths (225 nm, 254 nm, 272 nm, 275 nm, and 280 nm) were extracted using the five-point extremum method. The data from the first 16 minutes of the reaction were chosen based on ozone oxidation kinetics; during this period, the ozone oxidation reaction is most active, the degradation rate of organic pollutants is highest, and the changes in spectral characteristics are most significant.

[0081] Absorbance data and corresponding ozone dosages were labeled to form a dataset. A multiple linear regression model was then used to fit the dataset and establish a multiple linear regression prediction model. Multiple linear regression is a classic statistical analysis method that establishes a linear relationship between multiple independent variables and a dependent variable.

[0082] The mathematical formula for the multiple linear regression prediction model is:

[0083] ;

[0084] in: The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value is at a wavelength of 280 nm; -3868.7 is a constant term, and 7884.7, -1973.9, 6869.0, -1648.6, and -5403.3 are regression coefficients.

[0085] An area integral was performed on the 230-300 nm wavelength region, and an exponential regression model was fitted with the integrated area as the independent variable and the ozone dosage as the dependent variable to establish an exponential regression prediction model. The integrated area comprehensively reflects the overall content and structural complexity of organic matter. During ozone oxidation, the required ozone dosage often exhibits a non-linear growth trend with the increase of organic matter concentration, and the exponential model describes this relationship well.

[0086] The mathematical formula for the exponential regression prediction model is:

[0087] ;

[0088] in:

[0089] Ozone dosage (mgO3 / min); It is a natural constant; The integral area in the 230-300nm band; is the coefficient of exponential regression.

[0090] Step S5: COD prediction model construction;

[0091] The purpose of constructing the chemical oxygen demand (COD) prediction model is to establish a quantitative relationship between ultraviolet absorption spectral characteristics and wastewater pollution load, so as to achieve rapid assessment and online monitoring of wastewater treatment effects.

[0092] COD was measured on 28 influent and 56 effluent samples, and ultraviolet spectra in the 200-600 nm range were obtained. The influent samples represent the original pollution status of the tannery wastewater, while the effluent samples reflect the water quality improvement effects under different treatment conditions.

[0093] The five-point extreme value method was used to identify peak positions and extract features from the absorption spectra of the samples. Frequency statistics were performed on the absorption peaks of all samples, identifying 10 characteristic wavelengths: 225 nm, 227 nm, 242 nm, 249 nm, 254 nm, 267 nm, 272 nm, 275 nm, 279 nm, and 280 nm. During ozone treatment, the absorbance at these wavelengths showed a significant decrease or near disappearance, indicating that the organic matter corresponding to these wavelengths has high representativeness and stability in tannery wastewater, and also exhibits good response characteristics during ozone oxidation.

[0094] A dataset was constructed by labeling absorbance values ​​at 10 characteristic wavelengths with actual measured COD data, and a COD prediction model was established using partial least squares regression. Partial least squares regression is a multivariate statistical analysis method that combines principal component analysis, canonical correlation analysis, and multiple linear regression, and is particularly suitable for situations where there is multicollinearity among independent variables and the sample size is relatively small.

[0095] The mathematical formula for the COD prediction model is:

[0096] ;

[0097] in: Chemical oxygen demand (mg / L); The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 227 nm; The absorbance value at a wavelength of 242 nm; The absorbance value at a wavelength of 249 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 267 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value at a wavelength of 279 nm; is the absorbance value at a wavelength of 280 nm; -32.34733 is a constant term, and 133.8412, -96.835106, 72.941815, 4.9676696, 45.647989, -286.48653, -320.1749, 360.6204, 202.139048, and 53.598849 are partial least squares regression coefficients.

[0098] The present invention provides a method for predicting ozone dosage and COD in leather wastewater based on ultraviolet-visible spectroscopy. By establishing a quantitative relationship between spectral characteristics and treatment parameters, it achieves precise control and real-time monitoring of the leather wastewater treatment process. This method fully utilizes the advantages of rapid response and rich information in ultraviolet spectroscopy technology, combined with advanced mathematical modeling methods, providing scientific and technical support for the efficient treatment of leather wastewater. Compared with traditional empirical dosing methods, this method dynamically adjusts the ozone dosage according to real-time changes in influent water quality, significantly improving treatment efficiency, reducing operating costs, and minimizing adverse environmental impacts.

[0099] Example 1: In this example, influent water samples were collected from the ozone catalytic oxidation tank of a tannery wastewater treatment plant. A three-dimensional fluorescence spectrometer was used to scan the water samples, obtaining a three-dimensional excitation-emission matrix (EEM). Based on the collected EEM, the fluorescence signals were divided into five regions according to the internationally accepted five-zone fluorescence zoning method: Region I (excitation wavelength Ex 200-250 nm, emission wavelength 250-330 nm), Region II (200-250 nm, 330-380 nm), Region III (200-250 nm, 380-550 nm), Region IV (Ex 250-400 nm, Em 330-380 nm), and Region V (Ex 250-400 nm, Em 380-550 nm). These five zones correspond to five typical types of organic pollutants: tyrosine-like proteins, tryptophan-like proteins, fulvic acid-like substances, mixtures of tyrosine and tryptophan-like proteins, and humic substances (see attached). Figure 1 (As shown).

[0100] To clarify the ultraviolet absorption characteristics of typical organic compounds in tannery wastewater and provide absorption peak position information for subsequent model construction, this invention uses ultraviolet-visible spectroscopy to analyze representative pollutants. The specific method is as follows: Standard solutions of tyrosine, tryptophan, fulvic acid, and humic substances at different concentrations were prepared, and ultraviolet absorption spectra in the wavelength range of 200–600 nm were acquired using an ultraviolet-visible spectrometer (e.g., Figure 2 (as shown in the figure), and the positions of the characteristic absorption peaks of each substance in the ultraviolet region were extracted.

[0101] Experimental results showed that tyrosine solution exhibited significant absorption peaks at 225 nm and 272 nm, tryptophan solution had characteristic absorption wavelengths of 225 nm and 275 nm, while fulvic acid showed significant absorption at 280 nm. Since humic substances are complex mixtures, their UV absorption peaks cannot be clearly identified by measuring a single component, making it difficult to determine a single characteristic wavelength. Therefore, referring to conventional practices in water quality research both domestically and internationally, 254 nm was selected as an important wavelength representing the UV absorption of humic substances.

[0102] Based on the above results, the absorbance values ​​at five key wavelengths—225 nm, 254 nm, 272 nm, 275 nm, and 280 nm—were ultimately selected as the model input parameters for establishing the subsequent ozone dosage prediction model. This wavelength selection strategy combines experimental data with literature verification, demonstrating strong representativeness and practicality, and effectively reflecting the ultraviolet spectral characteristics of typical recalcitrant organic compounds in tannery wastewater.

[0103] To investigate the oxidative degradation effect of organic pollutants in tannery wastewater under different ozone addition conditions and to obtain the response relationship between ultraviolet spectral characteristics and chemical oxygen demand (COD), this invention constructs a small-scale ozone oxidation experimental device and sets representative ozone addition concentration parameters for systematic experiments.

[0104] In the experiment, ozone gas of different concentrations was prepared using an ozone generator. The ozone dosage was set at 15-40 mg O3 / min, increasing in a gradient of 5 mg O3 / min. Ozone was continuously introduced into a 1 L tannery wastewater sample at a constant gas flow rate for a reaction time of 30 minutes. During the ozone reaction, samples were taken at regular intervals for multi-index detection: the COD value at each sampling time point during the reaction was determined using the national standard method—potassium dichromate method (GB11914-89); and ultraviolet spectra were collected in the wavelength range of 200–600 nm with wavelength intervals of 0.5 nm. Each sample was measured three times, and the average value was taken to eliminate the influence of instrument drift and random errors on the absorbance results.

[0105] The COD values ​​obtained from the experiment and their corresponding UV absorption spectra were simultaneously imported into Origin software for processing, and COD degradation maps and UV absorption spectra under different ozone dosages were plotted (as attached). Figure 3 (As shown in the figure). Experimental results show that with the gradual increase of ozone dosage, the degradation effect of organic matter in wastewater is significantly enhanced. When the ozone dosage increases to 40 mg O3 / min, the COD value at the reaction endpoint decreases significantly, reaching a minimum of 28 mg / L, which meets the Class IV water quality standard in the "Surface Water Environmental Quality Standard" (GB3838-2002).

[0106] At an ozone dosage of 40 mg O3 / min, the oxidation reaction essentially reached its endpoint at 16 minutes, and the reaction system tended to stabilize. Therefore, this invention selects the ultraviolet spectra of samples taken within the first 16 minutes of the ozone reaction as the basic dataset for constructing the ozone prediction model, providing key spectral information support for establishing a high-precision intelligent ozone dosage prediction model.

[0107] An ozone dosage prediction model was established. First, the UV absorption spectrum data acquired during the initial 16 minutes of the reaction were imported into Origin software. The five-point extremum method in Origin was used to identify peak positions and extract features from the absorption spectrum curves of the samples in the 200–600 nm range, determining the absorbance values ​​at five wavelengths: 225 nm, 254 nm, 272 nm, 275 nm, and 280 nm. These absorbance data were then labeled with their corresponding ozone requirements to form a dataset.

[0108] The dataset was imported into MATLAB software, and a multiple linear regression model was used to fit the data, with absorbance as the independent variable and ozone demand as the dependent variable. The predictive performance of the model was verified by the coefficient of determination (R²) and root mean square error (RMSE). The results are attached. Figure 4 As shown, R² is 0.9472 and RMSE is 47.1734, indicating that the model has a high degree of fit to the data and can predict ozone dose well.

[0109] The mathematical formula for the multiple linear regression model is:

[0110] ;

[0111] In addition to single-point absorbance characteristics, to further extract overall absorption intensity information, the 230–280 nm range was selected as the integration region, and the area of ​​the absorbance curve within this range was calculated. This integration method, based on the numerical integration of the original absorbance with respect to wavelength, comprehensively reflects the overall presence level of typical recalcitrant organic compounds such as aromatic compounds and conjugated systems, and is one of the commonly used reaction indicators in advanced oxidation reactions.

[0112] The UV absorption spectra obtained during the initial 16 minutes of the reaction were imported into Origin software. Area integration was performed on the 230-300 nm wavelength region to determine the integrated area under different ozone dosages. The area data and corresponding ozone dosages were labeled to form a dataset. This dataset was then imported into MATLAB software, and an exponential regression model was fitted with the integrated area as the independent variable and the ozone dosage as the dependent variable. The results are attached. Figure 5 As shown, the fitting results are R²=0.9821, RMSE=28.4932. The mathematical formula for the exponential regression model is:

[0113] ;

[0114] Where A is the integral area in the 230–300 nm band; a COD concentration prediction model is established;

[0115] COD was measured in 28 influent and 56 effluent water samples, and ultraviolet spectra in the 200-600 nm range were obtained. The data were imported into Origin software, and the five-point extreme value method was used to identify peak positions and extract features from the absorption spectrum curves of the samples. Based on this, the frequency of absorption peaks in all samples was statistically analyzed, and wavelengths with high frequency in different influent samples were counted, ultimately identifying 10 characteristic wavelengths: 225, 227, 242, 249, 254, 267, 272, 275, 279, and 280 nm. These wavelengths were considered to correspond to organic matter effectively degraded during ozone oxidation, and were considered representative and stable. The absorbance values ​​at these 10 characteristic wavelengths were labeled with the actual measured COD data to construct a dataset. The data were imported into MATLAB software, and a COD prediction model was established using partial least squares regression (PLSR) with absorbance as the independent variable and COD as the dependent variable. The results are attached. Figure 6 As shown, the model's coefficient of determination (R²) is 0.9170, indicating that the model can explain approximately 91.70% of the COD variation. The root mean square error (RMSE) is 3.0885, demonstrating that the model has high accuracy and stability in predicting COD values.

[0116] The mathematical formula for the PLSR model is:

[0117] .

[0118] Where x1-x10 correspond to absorbance values ​​at wavelengths of 225, 227, 242, 249, 254, 267, 272, 275, 279, and 280 nm, respectively.

[0119] This invention introduces a simultaneous COD prediction mechanism. By collecting wastewater samples from multiple time periods and combining measured COD values ​​with their corresponding absorbance data, a quantitative correlation model between ultraviolet spectra and COD concentration is established, enabling online assessment of pollution load. This invention improves ozone utilization, reduces operating costs, and enhances the system's adaptability to complex influent conditions, demonstrating significant engineering application value and promising prospects.

[0120] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting ozone dosage and COD in leather wastewater based on ultraviolet-visible spectroscopy, characterized in that, Includes the following steps: S1, Organic matter classification; collect leather tanning wastewater samples, use a three-dimensional fluorescence spectrometer to scan the wastewater samples to obtain a three-dimensional fluorescence excitation-emission matrix spectrum, and classify and identify organic pollutants according to the five-zone fluorescence zoning method; S2, Key wavelength screening: Prepare standard solutions of tyrosine, tryptophan, fulvic acid and humic substances at different concentrations, and use a UV-Vis spectrophotometer to collect UV absorption spectra in the wavelength range of 200-600nm. Extract the characteristic absorption peak positions of each substance and determine the key wavelength as the model input parameters. S3, Ozone oxidation experiment; different ozone dosages were set up for gradient experiments, and samples were taken at regular intervals during the ozone oxidation reaction to simultaneously measure the COD value at each time point and collect ultraviolet absorption spectrum data. S4, Construction of ozone dosage prediction model: The ultraviolet absorption spectrum data obtained from the ozone oxidation experiment are processed to extract the absorbance value at the key wavelength and calculate the integral area within a certain wavelength range. A prediction model is established with wavelength absorbance and integral area as independent variables and ozone dosage as dependent variable. S5, COD prediction model construction: COD of collected influent and effluent samples is measured and ultraviolet spectra are obtained by scanning. Characteristic wavelengths are determined, and a COD prediction model is established with absorbance at the characteristic wavelength as the independent variable and COD as the dependent variable.

2. The method according to claim 1, characterized in that, The organic matter classification and feature identification in step S1 specifically includes: S11, a three-dimensional fluorescence spectrometer was used to scan the leather tanning wastewater sample to obtain a three-dimensional fluorescence excitation-emission matrix spectrum; S12, the fluorescence signal is divided into five regions according to the internationally accepted five-region fluorescence zoning method. Region I is tyrosine proteins; Region II is tryptophan proteins; Region IV is a mixture of tyrosine and tryptophan proteins; Regions III and V are humic organic matter, with Region III being fulvic acids and Region V being humic substances.

3. The method according to claim 2, characterized in that, In the five-zone fluorescence partitioning method, the excitation wavelength Ex of zone I is 200-250 nm and the emission wavelength Em is 250-330 nm; the excitation wavelength of zone II is 200-250 nm and the emission wavelength Em is 330-380 nm; the excitation wavelength of zone III is 200-250 nm and the emission wavelength Em is 380-550 nm; the excitation wavelength Ex of zone IV is 250-400 nm and the emission wavelength Em is 330-380 nm; and the excitation wavelength of zone V is 250-400 nm and the emission wavelength Em is 380-550 nm.

4. The method according to claim 1, characterized in that, The key wavelength screening in step S2 specifically includes: S21, prepare standard solutions of tyrosine, tryptophan and fulvic acid at different concentrations respectively, and collect ultraviolet absorption spectra in the wavelength range of 200-600nm using a UV-Vis spectrophotometer. S22, extract the characteristic absorption peaks of tyrosine at 225nm and 272nm, extract the characteristic absorption peaks of tryptophan at 225nm and 275nm, and extract the characteristic absorption peak of fulvic acid at 280nm. S23, 254nm was selected as the ultraviolet absorption wavelength to represent humic substances; S24, comprehensively determined five key wavelengths of 225nm, 254nm, 272nm, 275nm and 280nm as model input parameters.

5. The method according to claim 1, characterized in that, The ozone oxidation experiment in step S3 specifically includes: S31, ozone dosage set at 15-40mg / min, gradient 5mg The ozone was continuously introduced into a 1L tannery wastewater sample at a constant gas flow rate, with the rate increasing by 1min, and the reaction time was set to 30 minutes. S32, during the ozone introduction reaction process, samples were taken at regular intervals, and the COD value at each sampling time point was determined by the potassium dichromate method; S33, simultaneously acquires ultraviolet spectra in the wavelength range of 200-600nm with a wavelength interval of 0.5nm, and repeats the measurement 3 times for each group of samples and takes the average value.

6. The method according to claim 1, characterized in that, The ozone dosage prediction model construction in step S4 specifically includes: S41. Import the UV absorption spectrum data obtained within the first 16 minutes of the reaction into Origin software, and use the five-point extremum method to extract the absorbance values ​​at five wavelengths: 225nm, 254nm, 272nm, 275nm and 280nm. S42, the absorbance data and the corresponding ozone dosage are labeled to form a dataset, and a multiple linear regression model is used for fitting to establish a multiple linear regression prediction model; S43, area integration is performed on the 230-300nm band region, and an exponential regression model is fitted with the integrated area as the independent variable and the ozone dosage as the dependent variable to establish an exponential regression prediction model.

7. The method according to claim 6, characterized in that, The mathematical formula for the multiple linear regression prediction model is as follows: ； in: The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value is at a wavelength of 280 nm; -3868.7 is a constant term, and 7884.7, -1973.9, 6869.0, -1648.6, and -5403.3 are regression coefficients.

8. The method according to claim 6, characterized in that, The mathematical formula for the exponential regression prediction model is: ； in, This refers to the amount of ozone added. It is a natural constant; The integral area in the 230-300nm band; For exponential regression coefficients.

9. The method according to claim 1, characterized in that, The COD prediction model construction in step S5 specifically includes: S51, COD was measured on 28 influent and 56 effluent water samples, and ultraviolet spectra in the range of 200-600nm were obtained by scanning. S52, the five-point extreme value method was used to identify the peak position and extract features of the absorption spectrum curve of the sample. The frequency statistics of the absorption peaks of all samples were performed to determine 10 characteristic wavelengths: 225nm, 227nm, 242nm, 249nm, 254nm, 267nm, 272nm, 275nm, 279nm and 280nm. S53, the absorbance values ​​at 10 characteristic wavelengths are labeled with the actual measured COD data to construct a dataset, and a COD prediction model is established using partial least squares regression.

10. The method according to claim 9, characterized in that, The mathematical formula for the COD prediction model is: ； in, Chemical oxygen demand (COD) ); The absorbance value at a wavelength of 225 nm; The absorbance value at a wavelength of 227 nm; The absorbance value at a wavelength of 242 nm; The absorbance value at a wavelength of 249 nm; The absorbance value at a wavelength of 254 nm; The absorbance value at a wavelength of 267 nm; The absorbance value at a wavelength of 272 nm; The absorbance value at a wavelength of 275 nm; The absorbance value at a wavelength of 279 nm; is the absorbance value at a wavelength of 280 nm; -32.34733 is a constant term, and 133.8412, -96.835106, 72.941815, 4.9676696, 45.647989, -286.48653, -320.1749, 360.6204, 202.139048, and 53.598849 are partial least squares regression coefficients.

Images