An electronic industry wastewater identification method and system based on pollutant characteristic patterns

Abstract

This invention discloses a method and system for identifying wastewater from the electronics industry based on pollutant characteristic patterns, relating to the field of environmental monitoring and wastewater analysis technology. The method includes: acquiring pollutant response signals and generating original spectra through solid-phase extraction enrichment and liquid chromatography-mass spectrometry detection; extracting concentration and abundance data of organic and inorganic fluorine at multiple time points, constructing ratio evolution trajectories, and identifying transformation nodes; splitting the original spectra by transformation nodes, extracting peak distribution patterns of short-chain perfluorinated compounds, and comparing differences between adjacent spectra segments to generate peak migration characteristics; measuring wastewater salinity, screening pollutant response signals with different degrees of salinity inhibition for source tracing calibration; temporally associating transfer paths with transformation nodes, determining the contribution of transfer paths based on the frequency of transformation node occurrences, and generating source tracing feature spectra; and comparing with a pre-built feature library to output the identification results. This invention achieves high-precision identification and accurate source tracing of wastewater from the electronics industry.

Description

Technical Field

[0001] This invention relates to the field of environmental monitoring and wastewater analysis technology, specifically to a method and system for identifying wastewater from the electronics industry based on pollutant characteristic patterns. Background Technology

[0002] Wastewater from the electronics industry is complex in composition, containing various organic and inorganic fluorides, as well as other pollutants. These pollutants are persistent, bioaccumulative, and potentially toxic in the environment. Traditional wastewater identification methods mainly rely on the concentration measurement of single pollutants or simple chemical fingerprint analysis, which are insufficient to accurately reflect the dynamic characteristics and pollutant transformation patterns of wastewater from the electronics industry.

[0003] While current technologies such as liquid chromatography-mass spectrometry (LC-MS) can detect multiple pollutants in wastewater, they typically focus only on static concentration data and lack in-depth analysis of the pollutant's temporal evolution. Some studies attempt to determine wastewater type by comparing the ratio of organic to inorganic fluorine, but this method ignores the transformation behavior of pollutants under different environmental conditions, leading to insufficient identification accuracy. As emerging pollutants, short-chain perfluorinated compounds have not yet been fully studied in terms of their distribution patterns and migration and transformation characteristics in wastewater, making it difficult for existing identification methods to capture these subtle differences.

[0004] More importantly, environmental factors such as salinity in wastewater can significantly affect the detection response and transformation behavior of certain pollutants. However, existing technologies generally do not consider this interfering factor, leading to significant biases in the analysis of high-salinity wastewater samples. Furthermore, complex transformation relationships exist between pollutants; certain precursors can transform into other pollutants under specific conditions. Identifying this dynamic transformation process is crucial for accurate source tracing, but existing methods lack effective means for identifying transformation nodes and for source tracing calibration.

[0005] Therefore, there is an urgent need to establish a wastewater identification method that can comprehensively consider pollutant concentration evolution, abundance changes, peak distribution pattern shifts, and environmental factor interference, in order to achieve accurate identification and effective source tracing of wastewater from the electronics industry. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for identifying wastewater in the electronics industry based on pollutant characteristic patterns, aiming to solve at least one of the technical problems existing in the prior art.

[0007] The technical solution of this invention is: a method for identifying wastewater from the electronics industry based on pollutant characteristic patterns, comprising the following steps: After solid-phase extraction and enrichment of the wastewater to be tested, the response signals of various pollutants were obtained by liquid chromatography-mass spectrometry to generate the original spectrum; Concentration and abundance data of organic and inorganic fluorine were extracted at multiple time points, and the evolution trajectories of concentration ratio and abundance ratio were constructed. Transformation nodes were identified based on the evolution trajectories of concentration ratio and abundance ratio. The original spectrum is divided into multiple time-series spectrum segments with the conversion node time as the dividing point. The peak distribution patterns of short-chain perfluorinated compounds in each time-series spectrum segment are extracted. The differences in peak distribution patterns of adjacent time-series spectrum segments are compared to identify the combination of spectrum segments in which peak group shift occurs and generate peak group migration features. The salinity of the wastewater to be tested is measured, and pollutant response signals with different degrees of salinity inhibition are screened according to the salinity. Based on the screening results, the peak group transfer path in the peak group migration characteristics is traced and calibrated to generate calibrated peak group migration characteristics. The transfer paths and transformation nodes in the peak group migration characteristics after calibration are temporally correlated, and the contribution of the transfer paths is determined based on the frequency of occurrence of transformation nodes to generate the source feature spectrum. The source traceability feature spectrum is compared with a pre-established feature database of wastewater from the electronics industry to output the identification results.

[0008] After solid-phase extraction enrichment of the wastewater to be tested, the response signals of multiple pollutants were obtained by liquid chromatography-mass spectrometry detection, generating raw spectra including: Conduct conductivity of the wastewater to be tested to obtain the ion intensity curve of the wastewater to be tested. Determine the polarity of the solid phase extraction packing based on the ion intensity curve of the wastewater to be tested and generate extraction flow rate control parameters. Using the selected solid-phase extraction packing material, the sample loading speed is controlled according to the extraction flow rate control parameters for adsorption. The sample loading volume is determined based on the adsorption capacity of the solid-phase extraction packing material and the concentration of pollutants in the wastewater. After the sample loading is completed, rinsing and elution are performed to obtain the separated liquid. The turbidity of the separated liquid is measured, and the centrifugation speed and membrane pore size parameters are selected according to the turbidity value. The separated liquid is then centrifuged and filtered through a membrane to obtain the target component solution. The target component solution was separated by liquid chromatography, and the retention time and peak area data of the pollutant components were recorded to construct the component distribution sequence. Mass spectrometry detection parameters are determined based on the component distribution sequence, and the target component solution is detected by mass spectrometry to obtain the response signals of multiple pollutants; Based on the retention time and peak area of ​​pollutant components, peak shape correction and baseline drift compensation are performed on the response signals of various pollutants to generate the original spectrum.

[0009] Concentration and abundance data of organic and inorganic fluorine were extracted at multiple time points to construct the concentration ratio evolution trajectory and abundance ratio evolution trajectory. Based on the concentration ratio evolution trajectory and abundance ratio evolution trajectory, transformation nodes were identified, including: The response intensities of organic and inorganic fluorine at multiple time points were extracted from the original spectrum and converted into organic fluorine concentration data, inorganic fluorine concentration data, organic fluorine abundance data, and inorganic fluorine abundance data. Calculate the concentration ratio of organic fluorine concentration data to inorganic fluorine concentration data at each time point, and perform curve fitting on the concentration ratio in chronological order to generate the concentration ratio evolution trajectory. The abundance ratio of organic fluorine abundance data to inorganic fluorine abundance data at each time point is calculated, and the abundance ratio is subjected to curve fitting in chronological order to generate the evolution trajectory of the abundance ratio. Calculate the rate of change of concentration ratio between adjacent time points in the evolution trajectory of concentration ratio, identify the time point where the sign of the rate of change of concentration ratio changes as the concentration transition time, calculate the rate of change of abundance ratio between adjacent time points in the evolution trajectory of abundance ratio, and identify the time point where the sign of the rate of change of abundance ratio changes as the abundance transition time. The concentration transition time and the abundance transition time are matched in time, and the concentration transition time and the abundance transition time that overlap in time position are identified as transition nodes.

[0010] The original spectrum is divided into multiple time-series segments using the conversion node as the dividing point. Peak distribution patterns of short-chain perfluorinated compounds within each time-series segment are extracted. Differences in peak distribution patterns between adjacent time-series segments are compared to identify combinations of segments where peak shifts occur, generating peak migration features including: The original graph is divided into multiple time-series graph segments based on the transformation node time, generating a time-series graph sequence; Baseline correction and noise removal are performed on each time series segment in the time series spectrum sequence to obtain the corrected time series spectrum segment; Peak identification was performed on the corrected time series spectrum segments to extract the peak distribution patterns of short-chain perfluorinated compounds, a peak group feature matrix was constructed, and the spatial coordinates and distribution areas of the peak group feature matrix were calculated to obtain the peak group distribution characteristics. The peak group distribution characteristics of adjacent time series spectrum segments are matched and compared, the differences in peak group distribution patterns are calculated, and the peak group change characteristics are obtained. Based on the characteristics of peak group changes, peak group migration trajectories are constructed, the changing trends of peak group migration trajectories are analyzed, and combinations of spectral segments in which peak group shifts occur are identified. Extract the peak group migration direction and migration magnitude from the spectral segment combinations where peak group shifts occur, construct the peak group shift path, and generate peak group migration features containing the peak group shift path.

[0011] The salinity of the wastewater to be tested is measured. Based on the salinity, pollutant response signals with different degrees of salinity inhibition are screened. Based on the screening results, the peak group migration characteristics in the peak group migration features are traced and calibrated to generate calibrated peak group migration characteristics, including: Collect wastewater samples for salinity measurement and obtain salinity values; Based on the salinity measurement values, a dilution ratio sequence is set, and the wastewater sample to be tested is diluted stepwise according to the dilution ratio sequence. The conductivity of the diluted sample is measured, the salinity concentration of the diluted sample is calculated, and a salinity distribution sequence is generated. The pollutant response signals at each salinity concentration in the salinity distribution sequence were measured, the peak height and peak area of ​​the response signals were recorded, the change in response signals between adjacent salinity concentrations was calculated, and a response signal change sequence was generated. Based on the response signal change sequence, a response signal change threshold is set, pollutant peak groups whose response signal change exceeds the response signal change threshold are screened, and the characteristic ion information of the peak groups is recorded. The characteristic ion information of the peak group is matched with the peak group transfer path in the peak group migration feature, the peak group retention time offset is calculated, and the path sequence to be calibrated is generated. Calculate the peak retention time correction value in the path sequence to be calibrated based on the salinity measurement value and the response signal change sequence, perform peak transfer path calibration, and generate the calibration path sequence; The post-calibration peak group retention time in the calibration path sequence is updated to the peak group migration feature to generate the post-calibration peak group migration feature.

[0012] The transfer paths and transformation nodes in the calibrated peak group migration characteristics are temporally correlated, and the contribution of the transfer paths is determined based on the frequency of occurrence of transformation nodes, generating a source tracing feature spectrum including: The peak group transfer path in the calibrated peak group migration characteristics is expanded in time sequence to generate a transfer path time sequence. The transformation nodes are expanded in time sequence, and the transformation nodes are classified according to the changes in the peak groups at the transformation nodes to generate a time sequence of transformation nodes. Calculate the time interval between adjacent transformation nodes in the transformation node time series, determine the frequency of occurrence of transformation nodes based on the time interval, and generate a transformation node frequency sequence; The time series sequence of the transfer path is matched with the time series sequence of the transformation node, and the time series overlap interval of the transfer path and the transformation node is extracted to generate the time series correspondence sequence. Based on the time-series correspondence sequence and the frequency sequence of transformation nodes, the contribution of the transfer path is recursively calculated until the difference between two adjacent calculation results is less than a preset difference threshold, thereby generating a transfer path contribution sequence. Calculate the average contribution value of the transfer path contribution sequence, take the transfer paths with a contribution value higher than the average as the primary feature, and take the transfer paths with a contribution value lower than the average as the secondary feature, and organize the primary and secondary features to generate the source feature spectrum.

[0013] The source tracing feature spectrum is compared with a pre-established feature database of wastewater from the electronics industry to generate the following identification results: Calculate the ratio of response signal intensity between the primary and secondary features in the source traceability feature spectrum, and generate the feature response matrix; The correlation strength of the response signals between the primary and secondary features is calculated based on the feature response matrix. The primary and secondary features are divided into multiple feature groups according to the correlation strength of the response signals. The retention time difference of the peak groups within the feature group and the change ratio of the response signal are calculated to generate the feature group distribution matrix. Calculate the retention time interval and response signal variation amplitude of peak groups in the feature group distribution matrix. Divide the peak group response signal variation into multiple distribution intervals according to the interval and variation amplitude. Extract the peak group combination pattern of the feature group within the distribution interval and generate the evolution feature matrix. Standard feature spectra are extracted from a pre-established database of wastewater characteristics in the electronics industry. The retention time series and response signal sequence of peak groups in the standard feature spectra are calculated to generate a standard feature matrix. The peak combination pattern of the feature group within the distribution interval of the evolution feature matrix is ​​matched with the standard feature matrix. The matching degree between the peak group retention time series and the response signal series is calculated. The type of wastewater in the electronics industry is determined based on the standard feature spectrum corresponding to the highest matching degree, and the identification result is output.

[0014] This invention provides a wastewater identification system for the electronics industry based on pollutant characteristic patterns. The system includes: The enrichment and detection module is used to obtain the response signals of multiple pollutants by solid-phase extraction enrichment of the wastewater to be tested and detection by liquid chromatography-mass spectrometry, and generate the original spectrum. The node identification module is used to extract concentration and abundance data of organic and inorganic fluorine at multiple time points, construct the concentration ratio evolution trajectory and abundance ratio evolution trajectory, and identify transformation nodes based on the concentration ratio evolution trajectory and abundance ratio evolution trajectory. The feature generation module is used to split the original spectrum into multiple time-series spectrum segments with the conversion node time as the dividing point, extract the peak distribution pattern of short-chain perfluorinated compounds in each time-series spectrum segment, compare the differences in peak distribution patterns of adjacent time-series spectrum segments, identify the combination of spectrum segments in which peak group shift occurs, and generate peak group migration features. The source tracing calibration module is used to measure the salinity of the wastewater to be tested, screen pollutant response signals with different degrees of salinity inhibition based on the salinity, perform source tracing calibration on the peak group transfer path in the peak group migration characteristics based on the screening results, and generate calibrated peak group migration characteristics. The feature spectrum generation module is used to temporally correlate the transfer paths and transformation nodes in the migration features of the calibrated peak group, determine the contribution of the transfer paths based on the frequency of occurrence of transformation nodes, and generate the source feature spectrum. The identification output module is used to compare the source traceability feature spectrum with a pre-established feature database of electronic industry wastewater and output the identification results.

[0015] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0016] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps in any of the aforementioned methods.

[0017] This invention overcomes the limitations of traditional static concentration analysis by dynamically identifying transformation nodes based on the evolution trajectory of the ratio of organic to inorganic fluorine, thus improving the temporal resolution of wastewater feature identification. By decomposing the spectrum at transformation nodes and extracting the peak distribution patterns of short-chain perfluorinated compounds, the invention precisely characterizes the pollutant migration and transformation process, revealing the evolutionary patterns at different temporal stages. The introduction of a salinity factor for source tracing calibration eliminates the interference of high-salinity environments on pollutant response signals, solving the problem of insufficient accuracy in complex matrix wastewater analysis. By temporally correlating transfer paths and transformation nodes and quantifying their contribution, a multi-dimensional source tracing feature spectrum is established, enabling precise traceability of wastewater sources. This invention comprehensively considers multiple characteristics such as pollutant concentration evolution, abundance changes, peak distribution pattern shifts, and environmental factor interference, forming a systematic wastewater identification technology system that provides technical support for source determination, pollution control, and environmental supervision of wastewater in the electronics industry. Attached Figure Description

[0018] Figure 1 A flowchart illustrating a method for identifying wastewater in the electronics industry based on pollutant characteristic patterns, provided as an embodiment of the present invention; Figure 2 This embodiment shows a comparison of the conversion node matching success rate under different time window thresholds; Figure 3 This embodiment illustrates a comparison of the feature spectrum matching degree of different wastewater samples. Figure 4 This is a schematic diagram of the structure of an electronic industry wastewater identification system based on pollutant feature patterns, according to an embodiment of the present invention. Detailed Implementation

[0019] like Figure 1 As shown, Figure 1 A flowchart of a method for identifying wastewater in the electronics industry based on pollutant feature patterns, provided as an embodiment of the present invention, is included in the following steps: Step 101: After solid-phase extraction and enrichment of the wastewater to be tested, the response signals of various pollutants are obtained by liquid chromatography-mass spectrometry to generate the original spectrum.

[0020] In some embodiments of the present invention, step 101 may specifically include the following sub-steps: Sub-step 1011: Conduct conductivity of the wastewater to be tested, obtain the ion intensity curve of the wastewater to be tested, determine the polarity of the solid phase extraction packing based on the ion intensity curve of the wastewater to be tested, and generate extraction flow rate control parameters. Sub-step 1012: Using the selected solid-phase extraction packing material, the sample loading speed is controlled according to the extraction flow rate control parameters for adsorption. The sample loading volume is determined based on the adsorption capacity of the solid-phase extraction packing material and the concentration of pollutants in the wastewater. After the sample loading is completed, rinsing and elution are performed to obtain the separated liquid. Sub-step 1013: Measure the turbidity of the separation liquid, select the centrifugation speed and membrane pore size parameters according to the turbidity value, and perform centrifugation and membrane filtration on the separation liquid to obtain the target component solution; Sub-step 1014 involves separating the target component solution by liquid chromatography, recording the retention time and peak area data of the pollutant components, and constructing a component distribution sequence. Sub-step 1015: Determine the mass spectrometry detection parameters based on the component distribution sequence, perform mass spectrometry detection on the target component solution, and obtain the response signals of multiple pollutants; Sub-step 1016: Based on the retention time and peak area of ​​pollutant components, perform peak shape correction and baseline drift compensation on the response signals of multiple pollutants to generate the original spectrum.

[0021] The conductivity of wastewater samples at 20℃ was measured using a conductivity meter. Conductivity data were continuously recorded at different time points, and a conductivity-time curve was plotted. This curve reflects the distribution of ionic strength in the wastewater. Based on the peak range of the curve, a solid-phase extraction packing material of appropriate polarity was selected. If the conductivity curve shows a high peak region, it indicates a high content of polar compounds in the wastewater, and a polar packing material should be selected; if the curve shows a low peak and stable trend, a non-polar packing material should be selected. Simultaneously, by analyzing the relationship between conductivity and flow rate, the extraction flow rate control parameters were determined, which can be expressed by the formula: , where V f For the optimal extraction flow rate, k is the correlation coefficient of the packing type, and C is the optimal flow rate. m R is the maximum conductivity. e This represents the rate of change of electrical conductivity.

[0022] After determining the extraction parameters, the selected solid-phase extraction packing material was used for the adsorption process. The wastewater sample loading rate was controlled by a constant flow pump according to the calculated extraction flow rate control parameters. The sample loading volume was determined based on the adsorption capacity of the solid-phase extraction packing material and the estimated concentration of pollutants in the wastewater, and calculated using the adsorption capacity prediction formula. Among them, V s For the maximum loading volume, m s For the packing mass, C c C represents the unit adsorption capacity of the packing material. p To estimate the concentration of pollutants, S f For safety margin, a value of 0.7-0.8 is typically used. After sample loading, rinse with an appropriate amount of pure water or buffer solution to remove interfering matrix substances, and then elute with a suitable solvent to obtain a separation solution containing the target contaminant.

[0023] The turbidity of the obtained separated liquid was measured using a turbidimeter, with the turbidity value measured in NTU. Appropriate centrifugation speed and membrane pore size parameters were selected based on the turbidity value. When the turbidity value was higher than 10 NTU, high-speed centrifugation was performed first, typically at a speed of 5000-10000 r / min for 15 minutes. When the turbidity value was in the range of 5-10 NTU, a centrifugation speed of 3000-5000 r / min was selected. When the turbidity value was lower than 5 NTU, membrane filtration could be performed directly. For membrane pore size selection, a 0.45 μm pore size membrane was used when the turbidity value was greater than 8 NTU, and a 0.22 μm pore size membrane was used when the turbidity value was less than 8 NTU. Through centrifugation and membrane filtration, suspended particles and flocculent matter in the separated liquid were removed, resulting in a clear solution of the target component.

[0024] The treated target component solution was injected into a liquid chromatography system for separation. A C18 reversed-phase column was used, with the column temperature controlled at 35℃. An acetonitrile-water mobile phase was employed, and a gradient elution program was set. During the chromatographic separation, the retention time and peak area data of each pollutant component were recorded. Based on the relative positions and area ratios of the peaks on the chromatogram, a component distribution sequence was constructed, which included information such as the retention time, relative abundance, and peak width of each component.

[0025] Mass spectrometry detection parameters were determined based on the component distribution sequence. The electrospray ionization source temperature was set to 350℃, capillary voltage to 3.5kV, nebulizing gas pressure to 35psi, and drying gas flow rate to 10L / min. Based on the characteristics of each peak in the component distribution sequence, an appropriate scan range, typically 50-1000 m / z, was selected, using alternating positive and negative ion modes. Mass spectrometry was performed on the target component solution to obtain response signals for various pollutants, including mass-to-charge ratio and ionic intensity data.

[0026] After acquiring the raw response signal, data processing was performed. Based on the retention time and peak area data of pollutant components obtained from liquid chromatography separation, peak shape correction was performed on the mass spectrometry response signal. Gaussian fitting was used to adjust peak asymmetry and eliminate peak tailing. Polynomial fitting was used to compensate for baseline drift, eliminating baseline fluctuations caused by environmental and instrumental factors. Peak resolution algorithms were used to process overlapping peaks, improving the identification accuracy of each component. Finally, a raw spectrum containing retention time, mass-to-charge ratio, and relative abundance information was generated, serving as the basis for subsequent pollutant characteristic pattern analysis.

[0027] This invention achieves efficient enrichment and separation of trace pollutants in wastewater from the electronics industry by guiding the selection of solid-phase extraction conditions through conductivity detection and optimizing sample pretreatment through turbidity measurement. By employing liquid chromatography-mass spectrometry (LC-MS) to acquire pollutant response signals and combining peak shape correction and baseline drift compensation algorithms, the sensitivity and accuracy of wastewater pollutant detection are significantly improved, providing high-quality raw spectra for subsequent wastewater type identification.

[0028] Step 102: Extract concentration and abundance data of organic and inorganic fluorine at multiple time points, construct the concentration ratio evolution trajectory and abundance ratio evolution trajectory, and identify transformation nodes based on the concentration ratio evolution trajectory and abundance ratio evolution trajectory.

[0029] In some embodiments of the present invention, step 102 may specifically include the following sub-steps: Sub-step 1021: Extract the response intensities of organic and inorganic fluorine at multiple time points from the original spectrum and convert them into organic fluorine concentration data, inorganic fluorine concentration data, organic fluorine abundance data, and inorganic fluorine abundance data. Sub-step 1022: Calculate the concentration ratio of organic fluorine concentration data to inorganic fluorine concentration data at each time point, and perform curve fitting on the concentration ratio in chronological order to generate the concentration ratio evolution trajectory. Sub-step 1023: Calculate the abundance ratio of organic fluorine abundance data to inorganic fluorine abundance data at each time point, and perform curve fitting on the abundance ratio in chronological order to generate the evolution trajectory of the abundance ratio. Sub-step 1024: Calculate the concentration change rate of the concentration ratio evolution trajectory between adjacent time nodes, identify the time node where the sign of the concentration change rate changes as the concentration transition time, calculate the abundance change rate of the abundance ratio evolution trajectory between adjacent time nodes, and identify the time node where the sign of the abundance change rate changes as the abundance transition time. Sub-step 1025: Time-matching of concentration conversion time and abundance conversion time, and identifying concentration conversion time and abundance conversion time with overlapping time positions as conversion nodes.

[0030] The response intensities of organic and inorganic fluorine compounds at multiple time points were extracted from the original spectra. The original spectra included retention times, mass-to-charge ratios, and relative abundance information obtained through liquid chromatography-mass spectrometry (LC-MS). Characteristic peaks of organic and inorganic fluorine compounds were identified by analyzing the retention times and mass spectrometric features of the characteristic peaks in the original spectra. Organic fluorine compounds mainly include fluorine-containing organic compounds such as perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), while inorganic fluorine is mainly composed of fluoride ions. For these target compounds, their response intensity data at different time points were extracted from the original spectra.

[0031] Response intensity data needs to be converted into concentration and abundance data. Concentration data conversion is achieved using the standard curve method, where a pre-established correspondence between the response intensity and concentration of a standard compound is established, and the standard curve covers the typical concentration range in wastewater. Based on the measured response intensity and the standard curve, the actual concentrations of organic and inorganic fluorine are calculated. Abundance data conversion employs a normalization method, calculating the percentage of the target compound's peak area to the total peak area to obtain the relative abundance. This process is repeated at multiple time points to obtain time series data for organic and inorganic fluorine concentrations, as well as organic and inorganic fluorine abundance.

[0032] After acquiring the concentration data, the concentration ratio of organic to inorganic fluorine at each time point is calculated. Following this calculation, a curve is fitted to these discrete data points in chronological order to generate the concentration ratio evolution trajectory. The curve fitting employs either polynomial fitting or spline interpolation methods to ensure that the fitted curve is smooth and accurately reflects the data change trend.

[0033] Similarly, the abundance ratio of organic fluorine to inorganic fluorine at each time point is calculated, and the abundance ratio data is then curve-fitted to generate the abundance ratio evolution trajectory.

[0034] Based on the generated concentration ratio evolution trajectory, the concentration change rate between adjacent time points is calculated. The concentration change rate represents the trend of the concentration ratio changing over time; a positive value indicates an increase in the concentration ratio, and a negative value indicates a decrease. By analyzing the change in the sign of the concentration change rate, the time points when the sign of the concentration change rate changes from positive to negative or from negative to positive are identified. These time points are defined as concentration transition moments. Concentration transition moments represent the turning points in the ratio of organic to inorganic fluorine concentrations, reflecting the critical moments in the transformation of pollutants in wastewater.

[0035] Similarly, based on the evolution trajectory of abundance ratios, the abundance change rate between adjacent time points is calculated, the sign change of the abundance change rate is analyzed, and the time points where the sign of the abundance change rate changes are identified are defined as abundance transition moments. Abundance transition moments represent the turning point in the relative abundance relationship between organic and inorganic fluorine in wastewater.

[0036] The identified concentration and abundance transition times are matched temporally using a time window matching method, with a reasonable time window threshold set, typically half the sampling time interval. When the time difference between the concentration and abundance transition times is less than the set threshold, these two transition times are considered to coincide. These overlapping concentration and abundance transition times are identified as transition nodes. A transition node is a point in time where both the concentration ratio and abundance ratio change significantly, representing a critical point in the important transformation between organic and inorganic fluorine in wastewater.

[0037] like Figure 2 As shown, Figure 2 This paper compares the conversion node matching success rates of the embodiments of the present invention and the existing single-feature extreme value method under different time window thresholds. The horizontal axis represents the time window threshold, and the vertical axis represents the matching success rate. As shown in the figure, the matching success rate of both methods increases with the increase of the time window threshold; however, the embodiments of the present invention achieve a high matching success rate of 91.5% with a smaller time window threshold, and stabilize at 0.5 hours, significantly outperforming the existing technology at the same threshold. This demonstrates that the concentration and abundance dual-time matching mechanism adopted in the present invention has higher accuracy and robustness.

[0038] This invention constructs evolutionary trajectories by extracting concentration and abundance data of organic and inorganic fluorine, enabling dynamic tracking of fluoride pollution characteristics in electronics industry wastewater. The method of identifying transformation nodes based on the evolutionary trajectories of concentration and abundance ratios can accurately capture key time points in the conversion or reversal of organic fluorine to inorganic fluorine in wastewater, effectively reflecting the changing patterns of pollutants during the electronics industry production process.

[0039] Step 103: Using the conversion node time as the dividing point, the original spectrum is divided into multiple time-series spectrum segments. The peak distribution patterns of short-chain perfluorinated compounds in each time-series spectrum segment are extracted. The differences in peak distribution patterns of adjacent time-series spectrum segments are compared to identify the combination of spectrum segments in which peak shift occurs and generate peak migration features.

[0040] In some embodiments of the present invention, step 103 may specifically include the following sub-steps: Sub-step 1031: Divide the original graph into multiple time-series graph segments according to the transformation node time to generate a time-series graph sequence; Sub-step 1032: Perform baseline correction and noise removal on each time series segment in the time series spectrum sequence to obtain the corrected time series spectrum segment; Sub-step 1033 involves peak identification of the corrected time series spectrum segment, extraction of peak distribution patterns of short-chain perfluorinated compounds, construction of peak feature matrix, calculation of spatial coordinates and distribution area of ​​peak feature matrix, and obtaining peak distribution characteristics. Sub-step 1034: Match and compare the peak group distribution characteristics of adjacent time series spectrum segments, calculate the differences in peak group distribution patterns, and obtain peak group change characteristics; Sub-step 1035: Construct peak group migration trajectory based on peak group change characteristics, analyze the changing trend of peak group migration trajectory, and identify the combination of spectral segments in which peak group shift occurs; Sub-step 1036: Extract the peak group migration direction and migration magnitude from the spectral segment combination where peak group shift occurs, construct the peak group shift path, and generate peak group migration features containing the peak group shift path.

[0041] In the process of dividing the original spectrum into multiple time-series segments using the transformation node time as the dividing point, the transformation node times obtained in the previous step need to be arranged in chronological order to form a time point sequence. Assuming that the obtained transformation node times include 5 time points, namely 30 min, 60 min, 90 min, 120 min, and 150 min, the original spectrum from the start time 0 min to the end time 180 min is divided into 6 time-series segments: 0-30 min, 30-60 min, 60-90 min, 90-120 min, 120-150 min, and 150-180 min. For each segment, its complete liquid chromatography-mass spectrometry data, including retention time, mass-to-charge ratio, and response intensity information, are retained to form a time-series spectrum sequence.

[0042] A polynomial fitting method was used to perform baseline correction and noise removal on each time series segment in the time series spectrum sequence. For each time series segment, a third-order polynomial was selected as the baseline fitting function, and the baseline was calculated iteratively. In the first iteration, polynomial fitting was performed on all data points; in the second iteration, the deviation of each data point from the fitted baseline was calculated, and points with deviations greater than twice the standard deviation were marked as possible peaks, and their weights in the fitting were reduced; in the third iteration, the baseline was refitted using the adjusted weights. After 3-5 iterations, a stable baseline function was obtained, and the baseline-corrected signal was obtained by subtracting the baseline from the original signal. Noise removal was performed using wavelet transform, selecting the Daubechies-6 wavelet basis function, and decomposing the signal into five levels. For the high-frequency coefficients after decomposition, a threshold function was applied to remove random noise, with the threshold set to three times the standard deviation of the high-frequency coefficients. After wavelet reconstruction, the denoised signal was obtained, completing the baseline correction and noise removal, and generating the corrected time series segment.

[0043] Peak identification of the corrected time-series spectrum segments and extraction of peak distribution patterns for short-chain perfluorinated compounds requires peak identification using the first derivative method. The first derivative of the corrected signal is calculated, and the point where the derivative changes from positive to negative is identified as the peak vertex. Extending to the left and right from the peak vertex until the absolute value of the first derivative is less than a set threshold, the peak boundary is determined. Based on the characteristics of common short-chain perfluorinated compounds in electronics industry wastewater, including perfluorobutyric acid, perfluorohexanoic acid, and perfluorooctanoic acid, the retention time window for each target compound is determined. The identified peaks are screened, retaining those falling within the target retention time window. Characteristic parameters for each target peak are extracted, including peak height, peak area, peak width, and peak asymmetry. A peak group feature matrix is ​​constructed, where each row represents a peak and each column represents a characteristic parameter. The distribution region of the peak group in the retention time-mass-charge ratio two-dimensional space is calculated, determining the peak group boundary and center point coordinates, thus forming the peak group distribution characteristics.

[0044] The peak distribution characteristics of adjacent time-series spectral segments are matched and compared, and a peak matching algorithm is used to establish the correspondence between peak groups in adjacent spectral segments. For each peak, the similarity between it and its possible corresponding peaks in adjacent spectral segments is calculated. The similarity calculation formula is the ratio of the absolute value of the retention time difference to the peak width. When the similarity is greater than 0.8, it is considered a successful match. For successfully matched peak groups, the difference values ​​of their characteristic parameters are calculated, including the peak height difference rate, peak area difference rate, and peak width difference rate. For peak groups that fail to match, they are recorded as newly added or disappeared peak groups. Combining all the difference parameters, the overall difference value of the peak group distribution pattern is calculated to form the peak group change characteristics.

[0045] Based on the peak group change characteristics, a peak group migration trajectory is constructed. The changing trend of the peak group migration trajectory is analyzed, and the combination of spectral segments where peak group shifts occur is identified. In this process, the feature parameters of matching peak groups in each time series spectral segment are normalized and converted into point coordinates in the feature space. The point coordinates of corresponding peak groups in adjacent time series spectral segments are connected to form peak group migration trajectory segments. The direction vector and length of each trajectory segment are calculated; the direction vector represents the direction of feature change, and the length represents the magnitude of change. A peak group shift judgment threshold is set; when the magnitude of peak group feature change exceeds 50% of the peak width, it is judged as a significant shift. The peak group migration situation in all adjacent spectral segment combinations is analyzed, the spectral segment combinations where peak group shifts occur are identified, and the time interval of the shift and the peak group identifiers involved are recorded.

[0046] The migration direction and amplitude of peak groups in the spectral segment combinations where peak group transfers occur are extracted to construct peak group transfer paths. Peak group transfer features containing these paths are generated, and detailed analysis is performed on the identified spectral segment combinations where peak group transfers have occurred. The changes in peak groups in the retention time and mass-to-charge ratio dimensions are calculated to form two-dimensional migration vectors. The angle of the vector represents the migration direction, and the magnitude represents the migration amplitude. The change sequence of peak group migration vectors in multiple consecutive time-series spectral segments is recorded to construct complete peak group transfer paths. The shape characteristics of the transfer paths are analyzed, such as linear, curved, or oscillating. The migration rate of each segment in the path is calculated and expressed as the characteristic change per unit time. By combining the direction, amplitude, shape, and rate characteristics of the peak group transfer paths, peak group transfer features are generated, including information such as the time interval of the migration, the type of peak groups involved, and the distribution of migration direction and amplitude.

[0047] This invention accurately captures the dynamic changes of short-chain perfluorinated compounds in electronics industry wastewater over time by segmenting the original spectra at transformation nodes and performing peak migration analysis. Peak migration characteristics reflect the microscopic processes of pollutant transformation in wastewater, providing unique identification markers for wastewater from different production processes and treatment stages. It overcomes the limitations of traditional static feature analysis, improving the accuracy and reliability of wastewater type identification through dynamic feature analysis. Combining multidimensional analysis methods of peak distribution patterns and migration characteristics, it can effectively distinguish between electronics industry wastewater with similar components but different sources, providing a scientific basis for precise wastewater monitoring and targeted treatment.

[0048] Step 104: Measure the salinity of the wastewater to be tested, screen pollutant response signals with different degrees of salinity inhibition based on the salinity, perform source tracing calibration on the peak group migration characteristics based on the screening results, and generate calibrated peak group migration characteristics.

[0049] In some embodiments of the present invention, step 104 may specifically include the following sub-steps: Sub-step 1041: Collect the wastewater sample to be tested for salinity measurement and obtain the salinity measurement value; Sub-step 1042: Set the dilution ratio sequence according to the salinity measurement value, dilute the wastewater sample to be tested step by step according to the dilution ratio sequence, measure the conductivity of the diluted sample, calculate the salinity concentration of the diluted sample, and generate a salinity distribution sequence. Sub-step 1043: Measure the pollutant response signal at each salinity concentration in the salinity distribution sequence, record the peak height and peak area of ​​the response signal, calculate the change in response signal between adjacent salinity concentrations, and generate a response signal change sequence. Sub-step 1044: Set a response signal change threshold based on the response signal change sequence, filter out pollutant peak groups whose response signal change exceeds the response signal change threshold, and record the characteristic ion information of the peak group; Sub-step 1045: Match the peak group characteristic ion information with the peak group transfer path in the peak group migration characteristics, calculate the peak group retention time offset, and generate the path sequence to be calibrated. Sub-step 1046: Calculate the peak retention time correction value in the path sequence to be calibrated based on the salinity measurement value and the response signal change sequence, perform peak transfer path calibration, and generate the calibration path sequence; Sub-step 1047 updates the peak group retention time after calibration in the calibration path sequence to the peak group migration feature, generating the peak group migration feature after calibration.

[0050] The salinity of the wastewater was determined using the conductivity method. A 20 mL sample of the wastewater was taken, and its conductivity was measured using a conductivity meter. The conductivity meter needed to be calibrated before measurement. The conductivity probe was inserted into the wastewater sample, and the conductivity value was recorded after the reading stabilized. The salinity value was calculated using the conversion formula between conductivity and salinity.

[0051] Based on the measured salinity values, a dilution ratio sequence should be established, covering the range from the original sample to highly diluted samples. For wastewater with a salinity of 25‰, the dilution ratio sequence can be set as 1x, 2x, 4x, 8x, 16x, and 32x. The original wastewater samples are diluted according to the above dilution ratios, preparing six 50mL volumetric flasks. 50, 25, 12.5, 6.25, 3.125, and 1.5625mL of the original wastewater sample are added respectively, and then the volume is adjusted to 50mL with ultrapure water. The conductivity of each diluted sample is measured, and the salinity concentration is calculated based on the conductivity. The formula for calculating salinity concentration is: , of which S i S0 represents the salinity concentration after dilution; D represents the original wastewater salinity. i This indicates the dilution factor. The calculated salinity distribution sequence is 25, 12.5, 6.25, 3.125, 1.5625, and 0.78125‰.

[0052] The response signals of pollutants at various salinity concentrations in the salinity distribution sequence were determined using liquid chromatography-mass spectrometry (LC-MS). Samples at each dilution factor were measured sequentially, and the response signals of the target pollutant in each sample were recorded. Taking perfluorooctanoic acid (PFOA) as an example, the peak height and peak area of ​​its characteristic ion mass-to-charge ratio (M / C ratio) of 413.0 were recorded. For each target pollutant, a table corresponding to the response signal and salinity was established. The change in pollutant response signal between adjacent salinity concentrations was calculated using the following formula: Where, ΔR i，j R represents the change in the response signal from salinity j to salinity i; i R represents the peak area of ​​the response signal at salinity i. jThis represents the peak area of ​​the response signal at salinity j. For each pollutant, five sets of response signal change data were calculated, forming a response signal change sequence.

[0053] Based on the response signal change sequence, a response signal change threshold of 20% was set. When the change in response signal between adjacent salinity concentrations exceeded 20%, the pollutant was considered to be highly inhibited by salinity. Pollutant peak groups with response signal changes exceeding the threshold were screened, and the characteristic ion information of these peak groups, including mass-to-charge ratio, retention time, and peak shape characteristics, was recorded. Perfluorooctanoic acid (PFOA) compounds, such as PFOA and PFOA sulfonic acid, typically exhibit a high salinity inhibition effect and require calibration. However, some short-chain perfluorinated compounds, such as PFOA, may exhibit a smaller salinity inhibition effect, and their response signal changes may be below the threshold, requiring no calibration.

[0054] The selected peak group characteristic ion information is matched with the peak group migration characteristics obtained in the previous steps. For each recorded characteristic ion, the corresponding peak group is searched in the peak group migration characteristics to confirm the peak group transfer path information. The retention time shift of the peak group under different salinity conditions is calculated. The retention time shift refers to the difference between the retention time of the same peak group in the sample to be tested and the retention time under standard conditions. The peak groups that need to be calibrated and their retention time shifts are recorded to generate a calibration path sequence.

[0055] The retention time correction values ​​for peak groups in the calibration path sequence are calculated based on the salinity measurements and the response signal change sequence. The calculation of retention time correction values ​​is based on the quantitative relationship of salinity inhibition effect. For each peak group to be calibrated, the retention time correction value is calculated based on its response signal change and the salinity value of the sample. The retention time correction value generally increases with increasing salinity, but the correction coefficient varies for different pollutants. Peak group transfer path calibration is performed, applying the calculated retention time correction values ​​to the transfer path of the corresponding peak group, adjusting the time coordinates in the path, and generating the calibration path sequence.

[0056] The calibrated peak group retention times from the calibration path sequence are updated in the peak group migration features. For each calibrated peak group transfer path, its time coordinate information is updated while maintaining the peak group transfer direction and amplitude characteristics. The updated peak group migration features contain the calibrated peak group transfer paths, correcting for retention time shifts caused by salinity inhibition effects and more accurately reflecting the true migration characteristics of pollutants. The generated calibrated peak group migration features will be used for subsequent wastewater identification and classification.

[0057] This invention effectively solves the problem of interference from high-salinity wastewater on pollutant analysis based on this salinity calibration method, improving the accuracy and reliability of wastewater identification in the electronics industry. By establishing a quantitative relationship between pollutant response signals and salinity, precise correction of peak migration characteristics is achieved, eliminating systematic biases caused by salinity changes. The calibrated peak migration characteristics can more realistically reflect the characteristic patterns of pollutants in electronics industry wastewater, providing a reliable basis for the accurate identification of wastewater from different process sources.

[0058] Step 105: The transfer paths and transformation nodes in the migration features of the calibrated peak group are temporally correlated, and the contribution of the transfer paths is determined based on the frequency of occurrence of the transformation nodes to generate the source feature spectrum.

[0059] In some embodiments of the present invention, step 105 may specifically include the following sub-steps: Sub-step 1051: Expand the peak group transfer path in the calibrated peak group migration characteristics according to the time sequence to generate a transfer path time sequence. Sub-step 1052: Expand the transformation nodes in time sequence, classify the transformation nodes according to the changes in peak groups at the transformation nodes, and generate a time sequence of transformation nodes. Sub-step 1053: Calculate the time interval between adjacent conversion nodes in the conversion node time sequence, determine the frequency of occurrence of conversion nodes based on the time interval, and generate a conversion node frequency sequence. Sub-step 1054: Perform time-series mapping between the transfer path time series and the transformation node time series, extract the time-series overlap intervals between the transfer path and the transformation node, and generate a time-series mapping sequence; Sub-step 1055: Based on the time-series corresponding sequence and the frequency sequence of transformation nodes, the contribution of the transfer path is recursively calculated until the difference between two adjacent calculation results is less than a preset difference threshold, thereby generating a transfer path contribution sequence. Sub-step 1056: Calculate the average contribution value of the transfer path contribution sequence, take the transfer paths with a contribution value higher than the average value as the primary feature, take the transfer paths with a contribution value lower than the average value as the secondary feature, and organize the primary and secondary features to generate the source feature spectrum.

[0060] The peak migration characteristics of the calibrated peak groups are expanded chronologically. These peak migration paths record the changing trends of pollutant peak groups in different process flows, including peak appearance, splitting, merging, and disappearance. Each migration path consists of an initial peak group, an intermediate conversion process, and a final peak group, expanded with time as the horizontal axis to form a temporal sequence of migration paths. For wastewater from the electronics industry, a typical migration path is the gradual degradation of tetrachloroethylene into trichloroethylene, dichloroethylene, and vinyl chloride under specific process conditions. The time range of the migration path temporal sequence is typically 0-48 hours, with a temporal resolution of 0.5 hours, thus comprehensively capturing the changing characteristics of pollutants during wastewater treatment.

[0061] Transition nodes are points in time where a peak group undergoes a significant change. They need to be analyzed chronologically and categorized. Based on the changes in peak groups at these nodes, transition nodes can be classified into four types: generation, disappearance, splitting, and merging. Generation transition nodes indicate the emergence of a new peak group, such as intermediate products generated after adding new chemical reagents to wastewater. Disappearance transition nodes indicate the complete degradation or removal of a peak group. Splitting transition nodes indicate the decomposition of a peak group into multiple peak groups, such as the breakdown of large molecular pollutants into smaller molecular fragments. Merging transition nodes indicate the combination of multiple peak groups into one, such as the polymerization of small molecular pollutants into large molecules. For each transition node, its occurrence time, type, and the peak group information involved are recorded to form a time series sequence of transition nodes.

[0062] The time interval between adjacent conversion nodes in the conversion node time series is calculated. The frequency of occurrence of a conversion node is determined based on the calculated time interval, defined as the number of times a conversion node appears per unit time. For consecutive conversion nodes with a time interval of less than 2 hours, they are considered high-frequency conversion regions with a relatively high frequency of occurrence; for conversion nodes with a time interval of more than 8 hours, they are considered low-frequency conversion regions with a relatively low frequency of occurrence. A conversion node frequency sequence is generated by statistically analyzing the number of conversion nodes in different time periods.

[0063] The time series sequences of transfer paths and transformation nodes are matched temporally, and overlapping time intervals between the transfer paths and transformation nodes are extracted. These overlapping intervals refer to the regions where the time periods of change in the transfer path and the time periods of occurrence of the transformation node overlap. The correlation between them is determined by comparing the time points of peak group changes in the transfer path with the occurrence times of the transformation nodes. When the peak group changes in the transfer path and the transformation nodes are close in time, with a time difference of no more than 1 hour, they are considered to be correlated. For each transfer path, the number and type of transformation nodes associated with it are counted, forming a time series correspondence sequence.

[0064] Based on the time-series correspondence sequence and the frequency sequence of transformation nodes, the contribution of the transfer path is calculated recursively. The formula for calculating the contribution of the transfer path is: Among them, C n (i) represents the contribution of the i-th transfer path calculated in the n-th iteration; α represents the iteration weight coefficient, ranging from 0 to 1, usually set to 0.7; m represents the number of transformation nodes associated with the i-th transfer path; F j W(i,j) represents the frequency of occurrence of the j-th transformation node; W(i,j) represents the association weight between the i-th transformation path and the j-th transformation node, which is determined according to the type of transformation node: 0.8 for generation and disappearance types, and 1.2 for splitting and merging types. Indicates the first The contribution of the i-th transfer path is calculated in the next iteration. The initial condition C0(i)=1 indicates that the initial contributions of all transfer paths are equal. The recursive calculation process continues until the difference between two adjacent calculation results is less than the preset difference threshold of 0.01, i.e. The sequence of transfer path contribution values ​​obtained through recursive calculation reflects the degree of contribution of each transfer path to the wastewater characteristics.

[0065] The average contribution value of the transfer path contribution sequence is calculated by summing the contribution values ​​of all transfer paths and dividing by the total number of transfer paths. Transfer paths with contribution values ​​above the average are marked as primary features, as these paths have a significant impact on wastewater characteristics and can effectively distinguish wastewater from different sources. Transfer paths with contribution values ​​below the average are marked as secondary features, as these paths have a smaller impact on wastewater characteristics but may provide supplementary information under certain conditions. Primary and secondary features are sorted from highest to lowest contribution value and organized into a structured source tracing feature spectrum. The source tracing feature spectrum includes information such as: transfer path identifiers, initial peak group information, intermediate transformation processes, termination peak group information, associated transformation nodes, and contribution values.

[0066] This invention effectively solves the technical challenge of complexity identification in the source tracing of wastewater in the electronics industry based on a time-series correlation method of transfer paths and transformation nodes. By analyzing the time-series correlation of peak transfer paths and transformation nodes, the contribution of different pollutant characteristics to wastewater properties is accurately quantified, establishing a hierarchical source tracing feature spectrum. This feature spectrum can accurately reflect the process source characteristics of wastewater in the electronics industry, distinguish subtle process differences, and improve the accuracy and reliability of wastewater source identification. The contribution-level classification simplifies the wastewater identification process, reduces computational complexity, and makes the method more suitable for industrial applications, providing strong technical support for the precise management of wastewater in the electronics industry.

[0067] Step 106: Compare the source traceability feature spectrum with the pre-established feature database of electronic industry wastewater and output the recognition results.

[0068] In some embodiments of the present invention, step 106 may specifically include the following sub-steps: Sub-step 1061: Calculate the ratio of response signal intensity between the primary and secondary features in the source traceability feature spectrum, and generate the feature response matrix; Sub-step 1062: Calculate the correlation strength of the response signals between the main features and the secondary features based on the feature response matrix; divide the main features and the secondary features into multiple feature groups according to the correlation strength of the response signals; calculate the retention time difference of the peak groups within the feature groups and the change ratio of the response signals; and generate the feature group distribution matrix. Sub-step 1063: Calculate the retention time interval and response signal change amplitude of the peak group in the feature group distribution matrix. Divide the peak group response signal change into multiple distribution intervals according to the interval and change amplitude. Extract the peak group combination pattern of the feature group within the distribution interval and generate the evolution feature matrix. Sub-step 1064: Extract standard feature spectrum from the pre-established electronic industry wastewater feature library, calculate the retention time series and response signal sequence of peak groups in the standard feature spectrum, and generate a standard feature matrix; Sub-step 1065 involves matching the peak group combination pattern of the feature group within the distribution interval in the evolution feature matrix with the standard feature matrix, calculating the matching degree between the peak group retention time series and the response signal sequence, determining the type of wastewater from the electronics industry based on the standard feature spectrum corresponding to the highest matching degree, and outputting the identification result.

[0069] The distinction between primary and secondary features is based on the transfer path contribution calculated in the preceding steps. Features with a contribution higher than the average are considered primary features, while those lower are considered secondary features. For each primary and secondary feature, the response signal intensity value of its corresponding peak group is extracted, and the response signal intensity ratio between the primary and secondary features is calculated. The calculation of the response signal intensity ratio is based on mass spectrometry peak area or peak height data, reflecting the relative content relationship between different pollutants. For multiple groups of primary and secondary features, a two-dimensional matrix of response signal intensity ratios is constructed to form a feature response matrix. The rows and columns of the feature response matrix correspond to primary and secondary features, respectively, and the matrix elements are the response signal intensity ratio values ​​of the corresponding features. For typical organic pollutants in electronics industry wastewater, such as perfluorooctanoic acid (PFOA) and perfluorodecanoic acid (PFDC), the response signal intensity ratio between the two is calculated and recorded in the feature response matrix.

[0070] The correlation strength of the response signals between primary and secondary features is calculated based on the feature response matrix. The correlation strength reflects the degree of mutual influence between different features, and the calculation formula is as follows: , where C ij R represents the correlation strength of the response signal between primary feature i and secondary feature j; ij R represents the ratio of the response signal intensity corresponding to the primary feature i to the secondary feature j in the feature response matrix; n represents the total number of secondary features; ikI represents the ratio of the response signal intensity corresponding to the primary feature i to the secondary feature k in the feature response matrix. i I represents the response signal strength of the main feature i; j This represents the response signal strength of secondary feature j. The correlation strength calculation considers both the absolute signal strength and relative ratio of the features, comprehensively reflecting the relationships between them. Based on the calculated correlation strength of the response signals, primary and secondary features are divided into multiple feature groups. The correlation strength threshold is set to 0.5; features with a correlation strength greater than 0.5 are grouped into the same feature group. For each feature group, the retention time difference and response signal change ratio of the peak groups within the group are calculated. The retention time difference is the maximum value minus the minimum retention time of different peak groups within the group, and the response signal change ratio is the ratio of the maximum to the minimum response signal within the group. The retention time differences and response signal change ratios of each feature group are organized into a matrix to generate the feature group distribution matrix.

[0071] The retention time interval and response signal variation amplitude of peak groups in the feature group distribution matrix are calculated. The retention time interval refers to the difference in retention time between adjacent peak groups, and the response signal variation amplitude refers to the percentage change in response signal between adjacent peak groups. For each feature group in the feature group distribution matrix, the retention time interval and response signal variation amplitude of all adjacent peak pairs within the group are calculated. Based on the calculation results, the peak group response signal variation is divided into multiple distribution intervals. The retention time interval is divided into small interval intervals, medium interval intervals, and large interval intervals, corresponding to intervals less than 1 minute, intervals between 1 and 3 minutes, and intervals greater than 3 minutes, respectively. The response signal variation amplitude is divided into small amplitude variation intervals, medium amplitude variation intervals, and large amplitude variation intervals, corresponding to variations less than 20%, variations between 20 and 50%, and variations greater than 50%, respectively. For each distribution interval, the peak group combination pattern within the feature group is extracted. The peak group combination pattern includes information such as the number of peak groups, the distribution range of retention time, and the distribution characteristics of the peak group response signal. The peak group combination patterns in different distribution intervals are organized into a matrix form to generate an evolution feature matrix.

[0072] Standard feature spectra are extracted from a pre-established wastewater feature library for the electronics industry. This library contains standard feature spectra of various typical electronic process wastewaters, such as photolithography wastewater, developing wastewater, and etching wastewater. For each type of wastewater, the feature library records the standard feature spectra of its characteristic pollutants. Retention time series and response signal sequences of peak groups are extracted from the standard feature spectra. The retention time series is a sequence formed by arranging the retention times of each peak group in the standard feature spectrum from smallest to largest, and the response signal sequence is a sequence formed by arranging the response signal intensities of the corresponding peak groups in order of retention time. The retention time series and response signal sequences are combined into a matrix to generate a standard feature matrix. The elements in the standard feature matrix include information such as peak group number, retention time, response signal intensity, and peak group feature description.

[0073] The peak combination pattern of feature groups within the distribution intervals of the evolution feature matrix is ​​matched with the standard feature matrix. Distribution interval matching refers to comparing the distribution intervals of the wastewater to be identified with the distribution intervals in the standard feature matrix. For each distribution interval, the matching degree between the wastewater to be identified and the standard feature matrix is ​​calculated. The matching degree is calculated based on the similarity between the retained time series and the response signal sequence, and the formula is: M = w t ×M t +w r ×M r Where M represents the overall matching degree; M t Indicates the degree of matching of the retained time series; M r Indicates the degree of matching of the response signal sequence; w t This represents the weighting coefficient for retention time matching, typically set to 0.6; w r The weighting coefficient representing the matching of the response signal is typically set to 0.4. The retention time series matching degree is determined by calculating the deviation between the retention time of the wastewater to be identified and the corresponding peak group in the standard feature matrix; the smaller the deviation, the higher the matching degree. The response signal series matching degree is determined by calculating the relative difference between the response signal of the wastewater to be identified and the corresponding peak group in the standard feature matrix; the smaller the difference, the higher the matching degree. For typical pollutants in electronics industry wastewater, such as perfluorinated compounds and phthalates, the matching degree calculation considers their unique response characteristics. Based on the calculated matching degree, the standard feature spectrum corresponding to the highest matching degree is determined. The standard feature spectrum with the highest matching degree represents the most likely type of wastewater to be identified. The identification results are output, including wastewater type, matching degree value, and information on the main characteristic pollutants.

[0074] like Figure 3 As shown, Figure 3This paper demonstrates that our technical solution calculates the correlation strength of response signals between primary and secondary features, divides features into multiple feature groups, generates a feature group distribution matrix and an evolution feature matrix, and then performs distribution interval matching with a standard feature matrix. This enables us to more accurately capture subtle changes in the wastewater feature spectrum. In contrast, the traditional DTW algorithm cannot effectively distinguish the response signal intensity ratio between primary and secondary features when dealing with complex matrix interference, resulting in a low overall matching degree. Our solution retains the reasonable allocation of time matching weight (0.6) and response signal matching weight (0.4), further improving the robustness of the matching and significantly outperforming existing technologies.

[0075] This invention presents a pattern comparison identification method based on source tracing feature spectra and a feature database of wastewater from the electronics industry, enabling accurate identification and classification of wastewater from different process sources within the electronics industry. By establishing a correlation between the response signals of primary and secondary features, and combining peak retention time and response signal variation characteristics, a multi-dimensional feature description system is formed, effectively overcoming the insufficient accuracy problem caused by traditional methods relying on only a single pollutant or limited indicators. It exhibits extremely high detection sensitivity and identification specificity for trace pollutants in electronics industry wastewater, accurately distinguishing wastewater from different sources even under complex matrix interference and with numerous pollutant types. This provides reliable technical support for precise treatment and targeted emission reduction, improving the targeting and efficiency of wastewater treatment.

[0076] like Figure 4 As shown, Figure 4 A schematic diagram of a wastewater identification system for the electronics industry based on pollutant feature patterns, provided in an embodiment of the present invention, is shown. The system includes: The enrichment and detection module 401 is used to obtain the response signals of multiple pollutants by liquid chromatography-mass spectrometry after solid-phase extraction enrichment of the wastewater to be tested, and generate the original spectrum. The node identification module 402 is used to extract the concentration and abundance data of organic and inorganic fluorine at multiple time points, construct the concentration ratio evolution trajectory and the abundance ratio evolution trajectory, and identify the transformation node based on the concentration ratio evolution trajectory and the abundance ratio evolution trajectory. The feature generation module 403 is used to split the original spectrum into multiple time-series spectrum segments with the conversion node time as the dividing point, extract the peak distribution pattern of short-chain perfluorinated compounds in each time-series spectrum segment, compare the differences in peak distribution patterns of adjacent time-series spectrum segments, identify the combination of spectrum segments in which peak group shift occurs, and generate peak group migration features. The source tracing calibration module 404 is used to measure the salinity of the wastewater to be tested, screen pollutant response signals with different degrees of salinity inhibition based on the salinity, perform source tracing calibration on the peak group transfer path in the peak group migration feature based on the screening results, and generate calibrated peak group migration feature. The feature spectrum generation module 405 is used to temporally correlate the transfer paths and transformation nodes in the migration features of the calibrated peak group, determine the contribution of the transfer paths based on the frequency of occurrence of the transformation nodes, and generate the source feature spectrum. The identification output module 406 is used to compare the source traceability feature spectrum with the pre-established feature database of electronic industry wastewater and output the identification results.

[0077] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0078] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing a computer program, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0079] The specific embodiments described above are preferred embodiments of the present invention and are not intended to limit the specific scope of the present invention. The scope of the present invention includes, but is not limited to, these specific embodiments. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for identifying wastewater from the electronics industry based on pollutant characteristic patterns, characterized in that, Includes the following steps: After solid-phase extraction and enrichment of the wastewater to be tested, the response signals of various pollutants were obtained by liquid chromatography-mass spectrometry to generate the original spectrum; Concentration and abundance data of organic and inorganic fluorine were extracted at multiple time points, and the evolution trajectories of concentration ratio and abundance ratio were constructed. Transformation nodes were identified based on the evolution trajectories of concentration ratio and abundance ratio. The original spectrum is divided into multiple time-series spectrum segments with the conversion node time as the dividing point. The peak distribution patterns of short-chain perfluorinated compounds in each time-series spectrum segment are extracted. The differences in peak distribution patterns of adjacent time-series spectrum segments are compared to identify the combination of spectrum segments in which peak group shift occurs and generate peak group migration features. The salinity of the wastewater to be tested is measured, and pollutant response signals with different degrees of salinity inhibition are screened according to the salinity. Based on the screening results, the peak group transfer path in the peak group migration characteristics is traced and calibrated to generate calibrated peak group migration characteristics. The transfer paths and transformation nodes in the peak group migration characteristics after calibration are temporally correlated, and the contribution of the transfer paths is determined based on the frequency of occurrence of transformation nodes to generate the source feature spectrum. The source traceability feature spectrum is compared with a pre-established feature database of wastewater from the electronics industry to output the identification results.

2. The method according to claim 1, characterized in that, After solid-phase extraction enrichment of the wastewater to be tested, the response signals of multiple pollutants were obtained by liquid chromatography-mass spectrometry detection, generating raw spectra including: Conduct conductivity of the wastewater to be tested to obtain the ion intensity curve of the wastewater to be tested. Determine the polarity of the solid phase extraction packing based on the ion intensity curve of the wastewater to be tested and generate extraction flow rate control parameters. Using the selected solid-phase extraction packing material, the sample loading speed is controlled according to the extraction flow rate control parameters for adsorption. The sample loading volume is determined based on the adsorption capacity of the solid-phase extraction packing material and the concentration of pollutants in the wastewater. After the sample loading is completed, rinsing and elution are performed to obtain the separated liquid. The turbidity of the separated liquid is measured, and the centrifugation speed and membrane pore size parameters are selected according to the turbidity value. The separated liquid is then centrifuged and filtered through a membrane to obtain the target component solution. The target component solution was separated by liquid chromatography, and the retention time and peak area data of the pollutant components were recorded to construct the component distribution sequence. Mass spectrometry detection parameters are determined based on the component distribution sequence, and the target component solution is detected by mass spectrometry to obtain the response signals of multiple pollutants; Based on the retention time and peak area of ​​pollutant components, peak shape correction and baseline drift compensation are performed on the response signals of various pollutants to generate the original spectrum.

3. The method according to claim 1, characterized in that, Concentration and abundance data of organic and inorganic fluorine were extracted at multiple time points to construct the concentration ratio evolution trajectory and abundance ratio evolution trajectory. Based on the concentration ratio evolution trajectory and abundance ratio evolution trajectory, transformation nodes were identified, including: The response intensities of organic and inorganic fluorine at multiple time points were extracted from the original spectrum and converted into organic fluorine concentration data, inorganic fluorine concentration data, organic fluorine abundance data, and inorganic fluorine abundance data. Calculate the concentration ratio of organic fluorine concentration data to inorganic fluorine concentration data at each time point, and perform curve fitting on the concentration ratio in chronological order to generate the concentration ratio evolution trajectory. The abundance ratio of organic fluorine abundance data to inorganic fluorine abundance data at each time point is calculated, and the abundance ratio is subjected to curve fitting in chronological order to generate the evolution trajectory of the abundance ratio. Calculate the rate of change of concentration ratio between adjacent time points in the evolution trajectory of concentration ratio, identify the time point where the sign of the rate of change of concentration ratio changes as the concentration transition time, calculate the rate of change of abundance ratio between adjacent time points in the evolution trajectory of abundance ratio, and identify the time point where the sign of the rate of change of abundance ratio changes as the abundance transition time. The concentration transition time and the abundance transition time are matched in time, and the concentration transition time and the abundance transition time that overlap in time position are identified as transition nodes.

4. The method according to claim 1, characterized in that, The original spectrum is divided into multiple time-series segments using the conversion node as the dividing point. Peak distribution patterns of short-chain perfluorinated compounds within each time-series segment are extracted. Differences in peak distribution patterns between adjacent time-series segments are compared to identify combinations of segments where peak shifts occur, generating peak migration features including: The original graph is divided into multiple time-series graph segments based on the transformation node time, generating a time-series graph sequence; Baseline correction and noise removal are performed on each time series segment in the time series spectrum sequence to obtain the corrected time series spectrum segment; Peak identification was performed on the corrected time series spectrum segments to extract the peak distribution patterns of short-chain perfluorinated compounds, a peak group feature matrix was constructed, and the spatial coordinates and distribution areas of the peak group feature matrix were calculated to obtain the peak group distribution characteristics. The peak group distribution characteristics of adjacent time series spectrum segments are matched and compared, the differences in peak group distribution patterns are calculated, and the peak group change characteristics are obtained. Based on the characteristics of peak group changes, peak group migration trajectories are constructed, the changing trends of peak group migration trajectories are analyzed, and combinations of spectral segments in which peak group shifts occur are identified. Extract the peak group migration direction and migration magnitude from the spectral segment combinations where peak group shifts occur, construct the peak group shift path, and generate peak group migration features containing the peak group shift path.

5. The method according to claim 1, characterized in that, The salinity of the wastewater to be tested is measured. Based on the salinity, pollutant response signals with different degrees of salinity inhibition are screened. Based on the screening results, the peak group migration characteristics in the peak group migration features are traced and calibrated to generate calibrated peak group migration characteristics, including: Collect wastewater samples for salinity measurement and obtain salinity values; Based on the salinity measurement values, a dilution ratio sequence is set, and the wastewater sample to be tested is diluted stepwise according to the dilution ratio sequence. The conductivity of the diluted sample is measured, the salinity concentration of the diluted sample is calculated, and a salinity distribution sequence is generated. The pollutant response signals at each salinity concentration in the salinity distribution sequence were measured, the peak height and peak area of ​​the response signals were recorded, the change in response signals between adjacent salinity concentrations was calculated, and a response signal change sequence was generated. Based on the response signal change sequence, a response signal change threshold is set, pollutant peak groups whose response signal change exceeds the response signal change threshold are screened, and the characteristic ion information of the peak groups is recorded. The characteristic ion information of the peak group is matched with the peak group transfer path in the peak group migration feature, the peak group retention time offset is calculated, and the path sequence to be calibrated is generated. Calculate the peak retention time correction value in the path sequence to be calibrated based on the salinity measurement value and the response signal change sequence, perform peak transfer path calibration, and generate the calibration path sequence; The post-calibration peak group retention time in the calibration path sequence is updated to the peak group migration feature to generate the post-calibration peak group migration feature.

6. The method according to claim 1, characterized in that, The transfer paths and transformation nodes in the calibrated peak group migration characteristics are temporally correlated, and the contribution of the transfer paths is determined based on the frequency of occurrence of transformation nodes, generating a source tracing feature spectrum including: The peak group transfer path in the calibrated peak group migration characteristics is expanded in time sequence to generate a transfer path time sequence. The transformation nodes are expanded in time sequence, and the transformation nodes are classified according to the changes in the peak groups at the transformation nodes to generate a time sequence of transformation nodes. Calculate the time interval between adjacent transformation nodes in the transformation node time series, determine the frequency of occurrence of transformation nodes based on the time interval, and generate a transformation node frequency sequence; The time series sequence of the transfer path is matched with the time series sequence of the transformation node, and the time series overlap interval of the transfer path and the transformation node is extracted to generate the time series correspondence sequence. Based on the time-series correspondence sequence and the frequency sequence of transformation nodes, the contribution of the transfer path is recursively calculated until the difference between two adjacent calculation results is less than a preset difference threshold, thereby generating a transfer path contribution sequence. Calculate the average contribution value of the transfer path contribution sequence, take the transfer paths with a contribution value higher than the average as the primary feature, and take the transfer paths with a contribution value lower than the average as the secondary feature, and organize the primary and secondary features to generate the source feature spectrum.

7. The method according to claim 1, characterized in that, The source tracing feature spectrum is compared with a pre-established feature database of wastewater from the electronics industry to generate the following identification results: Calculate the ratio of response signal intensity between the primary and secondary features in the source traceability feature spectrum, and generate the feature response matrix; The correlation strength of the response signals between the primary and secondary features is calculated based on the feature response matrix. The primary and secondary features are divided into multiple feature groups according to the correlation strength of the response signals. The retention time difference of the peak groups within the feature group and the change ratio of the response signal are calculated to generate the feature group distribution matrix. Calculate the retention time interval and response signal variation amplitude of peak groups in the feature group distribution matrix. Divide the peak group response signal variation into multiple distribution intervals according to the interval and variation amplitude. Extract the peak group combination pattern of the feature group within the distribution interval and generate the evolution feature matrix. Standard feature spectra are extracted from a pre-established database of wastewater characteristics in the electronics industry. The retention time series and response signal sequence of peak groups in the standard feature spectra are calculated to generate a standard feature matrix. The peak combination pattern of the feature group within the distribution interval of the evolution feature matrix is ​​matched with the standard feature matrix. The matching degree between the peak group retention time series and the response signal series is calculated. The type of wastewater in the electronics industry is determined based on the standard feature spectrum corresponding to the highest matching degree, and the identification result is output.

8. A wastewater identification system for the electronics industry based on pollutant characteristic patterns, used to implement the method described in any one of claims 1-7, characterized in that, The system includes: The enrichment and detection module is used to obtain the response signals of multiple pollutants by solid-phase extraction enrichment of the wastewater to be tested and detection by liquid chromatography-mass spectrometry, and generate the original spectrum. The node identification module is used to extract concentration and abundance data of organic and inorganic fluorine at multiple time points, construct the concentration ratio evolution trajectory and abundance ratio evolution trajectory, and identify transformation nodes based on the concentration ratio evolution trajectory and abundance ratio evolution trajectory. The feature generation module is used to split the original spectrum into multiple time-series spectrum segments with the conversion node time as the dividing point, extract the peak distribution pattern of short-chain perfluorinated compounds in each time-series spectrum segment, compare the differences in peak distribution patterns of adjacent time-series spectrum segments, identify the combination of spectrum segments in which peak group shift occurs, and generate peak group migration features. The source tracing calibration module is used to measure the salinity of the wastewater to be tested, screen pollutant response signals with different degrees of salinity inhibition based on the salinity, perform source tracing calibration on the peak group transfer path in the peak group migration characteristics based on the screening results, and generate calibrated peak group migration characteristics. The feature spectrum generation module is used to temporally correlate the transfer paths and transformation nodes in the migration features of the calibrated peak group, determine the contribution of the transfer paths based on the frequency of occurrence of transformation nodes, and generate the source feature spectrum. The identification output module is used to compare the source traceability feature spectrum with a pre-established feature database of electronic industry wastewater and output the identification results.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 7.

Images