COD water pollution analysis system

Abstract

The application discloses a COD water pollution analysis system and belongs to the technical field of water pollution detection. The system solves the problem of false COD detection caused by the use of specific auxiliaries and dyes in printing and dyeing enterprises, which leads to high LAS and high dye composite interference in wastewater, false 48 mg / L masking real 78 mg / L exceeding standard water, and ecological disaster caused by misdischarge into a canal, and the problem of enterprise being punished and suspended production. The system comprises an optical coherence tomography module, a Raman spectrum detection module, and a dielectrophoresis photo-thermal micro-disturbance module. The optical coherence tomography module is used for emitting femtosecond laser and detecting interference signals to obtain depth-resolved scattering distribution of a sample. The Raman spectrum detection module is used for exciting and collecting Raman scattering spectrum of the sample for specific identification of dye and surfactant molecules. The dielectrophoresis photo-thermal micro-disturbance module is used for applying a controllable electric field and photo-thermal excitation to separate different pollutants and measure dielectric and thermal responses thereof. The multi-modal tomography Raman dielectrophoresis photo-thermal imaging combination of the application strips LAS dye interference, dynamically inverses true COD, and titrates a closed loop to avoid false standard reaching.

Description

Technical Field

[0001] This invention relates to the field of water pollution detection technology, and in particular to a COD water pollution analysis system. Background Technology

[0002] COD water pollution analysis systems monitor the degree of organic pollution in water bodies, primarily employing two technical approaches: electrochemical and spectroscopic methods. Each method has its own emphasis in terms of principle, structure, and application scenarios. The electrochemical method relies on the electrochemical oxidation reaction of organic matter in water samples on the surface of a three-electrode sensor. The COD value is calculated by detecting the current generated by the reaction. Equipped with an automatic sampling system, chemical measurement cell, and precision potentiostat, it offers advantages such as rapid measurement and high sensitivity, making it suitable for online, precise monitoring of key wastewater discharge outlets, including industrial wastewater. The spectroscopic method utilizes the absorption characteristics of organic matter to specific wavelengths of ultraviolet light. It measures absorbance using a photodetector and calculates the COD value using Lambert-Beer's law. Composed of an ultraviolet light source, sample flow cell, and turbidity compensation unit, it requires no reagents, produces no secondary pollution, and has a rapid response, making it suitable for routine screening and portable on-site testing of surface water and drinking water sources. Both types of equipment achieve data processing through signal conditioning, analog-to-digital conversion, and a main control unit, providing accurate data support for water pollution control.

[0003] For example, during the summer peak production season, dyeing and printing enterprises used specific auxiliaries and dyes, resulting in linear alkylbenzene sulfonate (LAS) concentrations exceeding 25 mg / L and residual dye concentrations exceeding 55 mg / L in their wastewater. Because LAS lacks a unique ultraviolet absorption peak and is difficult to identify through conventional spectroscopy, and because dye color is considered a normal characteristic of dyeing and printing wastewater, the combined interference of light scattering and background absorption distorts the signal of the online COD analysis system based on the ultraviolet absorption principle. This leads to the instrument outputting a falsely compliant reading of 48 mg / L, while the actual COD concentration is 78 mg / L, severely exceeding the standard. This misleads operators into making incorrect discharge decisions, resulting in a large amount of wastewater containing excessive toxic pollutants being discharged into a tributary of the Grand Canal, directly causing an ecological disaster in the downstream water bodies, and the enterprises involved facing hefty fines and production shutdowns for rectification.

[0004] Therefore, a COD water pollution analysis system is proposed to solve or alleviate the above problems. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a COD water pollution analysis system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A COD water pollution analysis system, including The optical coherence tomography module is used to emit femtosecond lasers and detect interference signals to obtain the depth-resolved scattering distribution of the sample; The Raman spectroscopy detection module is used to excite and acquire the Raman scattering spectrum of the sample for specific identification of dye and surfactant molecules; The dielectric electrophoresis photothermal micro-perturbation module is used to apply a controllable electric field and photothermal excitation to separate different pollutants and measure their dielectric and thermal responses. The acousto-optic tunable filter spectral imaging module is used to quickly scan and image the sample to obtain spatially resolved spectral absorbance data in the ultraviolet-visible band. The titration verification module is used to inject precisely measured amounts of chemical quencher into the sample according to instructions to verify the accuracy of the optical measurement results; The central processing and control module has its output terminals connected to the optical coherence tomography module, Raman spectroscopy detection module, dielectrophoresis photothermal micro-perturbation module, acousto-optic tunable filter spectral imaging module, and titration verification module, respectively. It is used to send synchronization commands, process all sensor data, run inversion algorithms, manage the verification process, and output the final results. The system is integrated with a microfluidic platform, whose signal terminal is connected to the output terminal of the central processing and control module. It is also connected to the sample flow cell of the optical coherence tomography module, the micro-reaction cell of the Raman spectroscopy detection module, the microchannel of the dielectrophoresis photothermal micro-perturbation module, and the micro-reaction cell of the titration verification module through microfluidic channels. This is used to control the delivery, mixing, and distribution of all samples and reagents, as well as the various modules.

[0007] Preferably, the process of sending synchronization commands, processing all sensor data, running the inversion algorithm, managing the verification process, and outputting the final result specifically includes the following steps: S1. Multi-source feature extraction and spatiotemporal registration: The central processing and control module collects and preprocesses the raw measurement data of each module, extracts optical, spectral and physicochemical features related to pollutants, and unifies all features to the same time and space coordinate system. S2. Dynamic joint inversion and concentration solution: Based on the extracted features, the central processing and control module constructs and solves a forward physicochemical model that includes the actual chemical oxygen demand concentration and the concentration parameters of various interfering substances. The actual chemical oxygen demand concentration after removing the composite interference is inverted through a nonlinear optimization algorithm. S3. Credibility assessment and active verification: The central processing and control module assesses the credibility of the inversion results. If the credibility is insufficient, it instructs the titration verification module to perform a chemical verification experiment and uses the verification results to correct the inversion concentration, thereby updating the estimated value of the true chemical oxygen demand concentration. S4. Results Output and System Update: The central processing and control module outputs the final true chemical oxygen demand concentration, the concentration of each interfering substance, and the data quality label, and saves the measurement data to update the parameters of the forward physicochemical model.

[0008] Preferably, the multi-source feature extraction and spatiotemporal registration involves the central processing and control module acquiring and preprocessing the raw measurement data from each module to extract optical, spectral, and physicochemical features related to pollutants, and unifying all features to the same time and space coordinate system. Specifically, this includes the following steps: S1.1 Within a preset measurement time window, the central processing and control module receives the original interference signal from the optical coherence tomography module, the original Raman spectral voltage signal from the Raman spectroscopy detection module, and the original hyperspectral image data from the acousto-optic tunable filter spectral imaging module via a high-speed data bus, and reads the original voltage signal of the microcantilever beam response and the original signal of the dielectric photothermal micro-perturbation module recorded by the control bus. S1.2 The central processing and control module performs Fourier transform and point spread function deconvolution on the original interference signal to calculate the depth-resolved backscattering coefficient distribution, and extracts the average backscattering intensity and the scattering entropy characterizing the uniformity of the scatterer distribution from it. S1.3 The central processing and control module performs dark noise subtraction and instrument response correction on the original Raman spectral voltage signal, and then obtains the standardized Raman spectrum by fluorescence background fitting and subtraction. Then, the characteristic Raman peak intensity index of the target dye and surfactant is calculated by the spectral matching algorithm. S1.4 The central processing and control module performs dark current and flat field correction on the original hyperspectral image data, calculates the absorbance matrix of each point in space, and extracts the average apparent absorbance at a wavelength of 254 nm and the average apparent absorbance in the characteristic absorption band of the dye. S1.5 The central processing and control module performs phase-locked amplification analysis on the original voltage signal of the micro cantilever beam response, extracts its amplitude and phase change characteristics, analyzes the original dielectric response signal, and extracts the dielectric response spectrum characteristics. S1.6 The central processing and control module adds a unified timestamp and spatial location label to all extracted features to complete spatiotemporal registration.

[0009] Preferably, the dynamic joint inversion and concentration solution involves the central processing and control module constructing and solving a forward physicochemical model based on the extracted features. This model includes the actual chemical oxygen demand (COD) concentration and the concentrations of various interfering substances. A nonlinear optimization algorithm is then used to invert the actual COD concentration after removing the composite interferences. Specifically, this includes the following steps: S2.1 The central processing and control module defines the state vector to be solved, which includes the actual chemical oxygen demand concentration, dye concentration, surfactant concentration, average micelle radius, and synergistic enhancement factor of the interaction between dye and surfactant. S2.2 The central processing and control module constructs a forward observation model with a state vector as input. The forward observation model includes an absorbance prediction model, a Raman intensity prediction model, and a backscattering intensity prediction model. S2.3 The central processing and control module uses all the spatiotemporally registered features extracted in S1 as observations to construct a global cost function; S2.4 The central processing and control module uses the Levenberg-Marquardt algorithm to iteratively adjust the state vector and minimize the global cost function. When the iteration meets the convergence condition, it outputs the optimal state vector, which includes the final estimated value of the actual chemical oxygen demand concentration.

[0010] Preferably, the absorbance prediction model is the predicted total absorbance at any detection wavelength, which is equal to the sum of the absorption contribution of the actual chemical oxygen demand component, the dye absorption contribution corrected by the synergistic enhancement factor, and the spurious absorbance contribution caused by surfactant micellar scattering. The dye absorption contribution is the product of its standard molar absorbance and concentration, multiplied by a synergistic enhancement term related to the surfactant concentration. The spurious absorbance contribution is proportional to the surfactant concentration, the square of the micelle size, and the scattering efficiency factor calculated based on Mie theory. The Raman intensity prediction model states that the predicted Raman intensity of the dye characteristic is proportional to its concentration, and the predicted Raman intensity of the surfactant characteristic is also proportional to its concentration. The backscattering intensity prediction model states that the predicted average backscattering coefficient is proportional to the square of the surfactant concentration and the micelle size.

[0011] Preferably, the global cost function is the weighted sum of squares of the differences between each observation and the predicted values ​​of the corresponding absorbance prediction model, Raman intensity prediction model, and backscattering intensity prediction model, with an additional regularization constraint term on the state vector.

[0012] Preferably, the credibility assessment and active verification, wherein the central processing and control module assesses the credibility of the inversion results, and when the credibility is insufficient, instructs the titration verification module to perform a chemical verification experiment, and uses the verification results to correct the inversion concentration, and updates the estimated value of the true chemical oxygen demand concentration accordingly, specifically including the following steps: S3.1 The central processing and control module calculates the comprehensive confidence score of the inversion result based on the minimum value of the global cost function and the estimated variance of each parameter in the optimal state vector. S3.2 If the overall confidence score is lower than the preset low confidence threshold, the central processing and control module determines that the result is questionable and triggers the active verification process. S3.3 In the active verification process, the central processing and control module instructs the system integration and microfluidic platform to transport the water sample to be tested to the micro-reaction cell of the titration verification module, and instructs the titration verification module to inject a calculated dose of specific chemical quencher based on the dye concentration estimate obtained by inversion. S3.4 The central processing and control module synchronously controls the Raman spectroscopy detection module to perform high-frequency monitoring of the micro-reaction cell and record the decay kinetic curve of the Raman characteristic peak intensity of the target dye; S3.5 The central processing and control module compares the measured decay kinetic curve with the theoretical curve predicted based on the inverted dye concentration and the pre-stored reaction kinetic constant. If the deviation exceeds the allowable range, the measured decay data is used as a new constraint, and the module returns to step S2.4 for joint inversion iteration until a state vector that meets the verification conditions is obtained.

[0013] Preferably, the comprehensive confidence score is calculated by taking the negative exponent of the ratio of the minimum value of the global cost function to a preset statistical threshold, and then multiplying it by a normalization factor for the probability density of a multidimensional Gaussian distribution calculated based on the state vector estimation covariance matrix.

[0014] Preferably, the result output and system update involve the central processing and control module outputting the final true chemical oxygen demand (COD) concentration value, the concentrations of each interfering substance, and the data quality label, and saving the measurement data to update the parameters of the forward physicochemical model. This specifically includes the following steps: S4.1 The central processing and control module generates a structured monitoring report, which includes at least the final true chemical oxygen demand concentration value, the estimated concentration values ​​of each interfering substance, the corresponding measurement uncertainty, and the data quality label based on the comprehensive confidence score. S4.2 The central processing and control module stores the complete data packet with a high confidence score in this measurement into the historical knowledge database. The complete data packet includes the original data, extracted features, inverted state vector and verification record. S4.3 The central processing and control module periodically calls the data in the historical knowledge database and updates and optimizes the empirical parameters in the positive observation model through a regression learning algorithm. The empirical parameters include the basic absorption spectrum coefficient of the real chemical oxygen demand and the empirical relationship between the synergistic enhancement factor and the surfactant concentration.

[0015] The present invention has the following beneficial effects: This invention utilizes optical coherence tomography to detect LAS micelle scattering, Raman spectroscopy to specifically identify dyes and LAS molecules, dielectrophoretic photothermal technology to separate pollutant responses, and multi-wavelength spectral imaging combined with dynamic model inversion to accurately remove complex optical interference and calculate the true COD value. This is then verified by titration to close the loop, thereby solving the problem of severely low measured values ​​caused by the synergistic interference of LAS and dyes in the traditional single ultraviolet absorption method, and avoiding false compliance emissions. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a structural block diagram of the present invention.

[0018] 101. Optical coherence tomography module; 102. Raman spectroscopy detection module; 103. Dielectrophoresis photothermal micro-perturbation module; 104. Acousto-optic tunable filter spectral imaging module; 105. Titration verification module; 201. Central processing and control module; 301. System integration and microfluidic platform. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of this invention, it should be understood that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. They are only used to facilitate the description of this invention and to simplify the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0023] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0024] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0025] A COD water pollution analysis system, such as Figure 1As shown, the system includes an optical coherence tomography module 101, a Raman spectroscopy detection module 102, a dielectrophoresis photothermal micro-perturbation module 103, an acousto-optic tunable filter spectral imaging module 104, a titration verification module 105, a central processing and control module 201, and a system integration and microfluidic platform 301. The optical coherence tomography module 101 emits a femtosecond laser and detects interference signals to obtain the depth-resolved scattering distribution of the sample. The Raman spectroscopy detection module 102 excites and acquires the Raman scattering spectrum of the sample for specific identification of dye and surfactant molecules. The dielectrophoresis photothermal micro-perturbation module 103 applies a controllable electric field and photothermal excitation to separate different contaminants and measure their dielectric and thermal responses. The acousto-optic tunable filter spectral imaging module 104 rapidly scans and images the sample to obtain spatially resolved spectral absorbance data in the ultraviolet-visible band. The titration verification module 105 injects liquid into the sample according to instructions. A precisely measured amount of chemical quencher is injected to verify the accuracy of the optical measurement results. The output of the central processing and control module 201 is connected to the optical coherence tomography module 101, the Raman spectroscopy detection module 102, the dielectrophoresis photothermal micro-perturbation module 103, the acousto-optic tunable filter spectral imaging module 104, and the titration verification module 105, respectively, for sending synchronization commands, processing all sensor data, running inversion algorithms, managing the verification process, and outputting the final results. The signal terminal of the system integration and microfluidic platform 301 is connected to the output of the central processing and control module 201, and it is connected to the sample flow cell of the optical coherence tomography module 101, the micro-reaction cell of the Raman spectroscopy detection module 102, the microchannel of the dielectrophoresis photothermal micro-perturbation module 103, and the micro-reaction cell of the titration verification module 105 through microfluidic channels, for controlling the delivery, mixing, and distribution of all samples and reagents, as well as the individual modules.

[0026] The optical coherence tomography module 101 includes a femtosecond pulse laser driving unit, a Michelson interferometer optical unit, a balanced photoelectric detection unit, and a high-speed analog-to-digital conversion unit; The femtosecond pulse laser driving unit includes a first high-speed comparator, a gallium nitride field-effect transistor driver, a first gallium nitride field-effect transistor, and a femtosecond pulse laser diode. The differential positive-phase output terminal and differential negative-phase output terminal of the first synchronous trigger signal of the central processing and control module 201 are respectively connected to the positive-phase signal input terminal and the negative-phase signal input terminal of the first high-speed comparator. The single-ended signal output terminal of the first high-speed comparator is connected to the pulse signal input terminal of the gallium nitride field-effect transistor driver. The gate drive signal output terminal of the gallium nitride field-effect transistor driver is connected to the gate terminal of the first gallium nitride field-effect transistor. The drain terminal of the first gallium nitride field-effect transistor is connected to the anode terminal of the femtosecond pulse laser diode, and its source terminal is grounded. The cathode terminal of the femtosecond pulse laser diode is connected to the first positive voltage driving power supply. The optical unit of the Michelson interferometer includes a first fiber coupler, an adjustable optical delay line, a sample arm objective, and a reference arm reflector. The emitted light from the femtosecond pulsed laser diode is connected to the input end of the first fiber coupler via an optical fiber. The first fiber coupler splits the light beam and sends it to the sample arm objective and the reference arm objective respectively. The sample arm optical path is focused by the sample arm objective and illuminates the sample flow cell delivered by the system integration and microfluidic platform 301. The backscattered light returns along the original path. The reference arm optical path is reflected back by the reference arm reflector after passing through the adjustable optical delay line. The two paths of light interfere at the first fiber coupler. The balanced photoelectric detection unit includes a balanced detector chip, whose two differential optical signal input terminals are respectively connected to the two interference signal output terminals of the first optical fiber coupler via optical fibers, for converting the interference optical signal into a differential current signal; The high-speed analog-to-digital converter unit includes a first programmable gain instrumentation amplifier, a first anti-aliasing filter, and a first high-speed analog-to-digital converter. The positive and negative terminals of the differential current output of the balanced detector chip are respectively connected to the positive and negative analog signal input terminals of the first programmable gain instrumentation amplifier. The gain control pin of the first programmable gain instrumentation amplifier is connected to the central processing and control module 201 through the first serial peripheral interface bus. Its amplified differential voltage output terminal is connected to the differential input terminal of the first anti-aliasing filter. The differential output terminal of the first anti-aliasing filter is connected to the positive and negative analog input pins of the first high-speed analog-to-digital converter through the first isolation transformer. The configuration pin of the first high-speed analog-to-digital converter is connected to the central processing and control module 201 through the first serial peripheral interface bus. Its high-speed serial data output pin is connected to the corresponding high-speed serial receiving interface of the central processing and control module 201 through the first protocol data bus. Raman spectroscopy detection module 102 includes a Raman excitation laser driving and temperature control unit, a Raman signal collection optical unit, a weak signal photoelectric conversion and amplification unit, and a Raman spectroscopy digitization unit; The Raman-excited laser driving and temperature control unit includes a first digital-to-analog converter, a first precision operational amplifier, a Raman-excited laser diode driver, a Raman-excited laser diode, a thermistor, and a semiconductor cooler temperature controller. The central processing and control module 201 is connected to the serial data input pin, serial clock pin, and chip select enable pin of the first digital-to-analog converter via a second serial peripheral interface bus. The analog voltage output pin of the first digital-to-analog converter is connected to the non-inverting input of the first precision operational amplifier. The output of the first precision operational amplifier is connected to the analog modulation voltage input of the Raman-excited laser diode driver. The constant current output pin of the Raman-excited laser diode driver is connected to the anode of the Raman-excited laser diode. The thermistor is attached to the heat sink of the Raman-excited laser diode and forms a temperature measurement bridge with a first reference resistor. The output of the temperature measurement bridge is connected to the temperature difference signal detection pin of the semiconductor cooler temperature controller. The pulse width modulation signal output pin of the temperature controller is connected to the gate of the first field-effect transistor to drive the semiconductor cooler. The serial communication pin of the temperature controller is connected to the central processing and control module 201 via the second serial peripheral interface bus. The Raman signal collection optical unit includes an excitation filter, a dichroic mirror, a collection objective, a Raman notch filter, a spectrometer entrance slit, and a SERS detection chip. The laser emitted from the Raman-excited laser diode is reflected sequentially by the excitation filter and the dichroic mirror, and then focused by the collection objective onto the sample point of the SERS detection chip delivered by the system integration and microfluidic platform 301. The generated Raman scattered light is collected by the same collection objective and passes through the dichroic mirror. After the residual excitation light is filtered out by the Raman notch filter, it is converged to the spectrometer entrance slit. The weak signal photoelectric conversion and amplification unit includes a CCD detector, which is a back-illuminated thin backlight charge-coupled device, and a first low-noise transimpedance amplifier. The photosensitive surface of the CCD detector is located on the image plane of the spectrometer entrance slit, and is used to convert the intensity distribution of the Raman scattered light after spectral dispersion into an analog charge signal and output it. Its analog output pin is connected to the high-impedance current signal input terminal of the first low-noise transimpedance amplifier. The Raman spectroscopy digitization unit includes a second high-speed analog-to-digital converter. The voltage output terminal of the first low-noise transimpedance amplifier is connected to the single-ended analog signal input pin of the second high-speed analog-to-digital converter. The configuration pin of the second high-speed analog-to-digital converter is connected to the central processing and control module 201 through the second serial peripheral interface bus. Its high-speed serial data output pin is connected to the corresponding high-speed serial receiving interface of the central processing and control module 201 through the second protocol data bus. The dielectric electrophoresis photothermal micro-perturbation module 103 includes a dielectric electrophoresis radio frequency driving unit, a photothermal excitation and detection unit, and a microfluidic sensing unit. The dielectric electrophoresis RF drive unit includes a direct digital frequency synthesizer, a first RF power amplifier, and an impedance matching network. The central processing and control module 201 is connected to the serial data input pin, serial clock pin, and chip select enable pin of the direct digital frequency synthesizer via a third serial peripheral interface bus. The RF signal output pin of the direct digital frequency synthesizer is connected to the input port of the first RF power amplifier. The output port of the first RF power amplifier is connected to the input terminal of the impedance matching network. The output terminal of the impedance matching network is connected to the first pair of interdigital electrodes located in the microfluidic sensing unit. The photothermal excitation and detection unit includes a modulated laser driver, a modulated laser diode, a microcantilever beam bias and excitation source, and a microcantilever beam signal conditioner. The central processing and control module 201 is connected to the enable control pin of the modulated laser driver through a pulse width modulation controller pin. The current output pin of the modulated laser driver is connected to the anode of the modulated laser diode. The emitted light from the modulated laser diode is focused by a lens and irradiates a specific area of ​​the microcantilever beam in the microfluidic sensing unit. The constant voltage output pin of the microcantilever beam bias and excitation source is connected to the bias electrode of the microcantilever beam. At the same time, the central processing and control module 201 is connected to the configuration pin of the microcantilever beam bias and excitation source through a third serial peripheral interface bus to set the excitation parameters. The piezoresistive signal output terminal of the microcantilever beam is connected to the differential high impedance input terminal of the microcantilever beam signal conditioner. The amplified signal output terminal of the microcantilever beam signal conditioner is connected to the analog input pin of the third high-speed analog-to-digital converter. The microfluidic sensing unit integrates a microchannel including a first pair of interdigital electrodes and a microcantilever beam. Its sample inlet and outlet are connected to the microfluidic platform 301 through pipelines and system integration to receive the sample to be tested delivered by it. The configuration pin of the third high-speed analog-to-digital converter is connected to the central processing and control module 201 through the third serial peripheral interface bus. Its converted data output pin is connected to the corresponding high-speed serial receiving interface of the central processing and control module 201 through the third protocol data bus to upload dielectric electrophoresis response and microcantilever beam deformation data. The acousto-optic tunable filter spectral imaging module 104 includes an acousto-optic tunable filter radio frequency driving unit, a high-speed spectral imaging detection unit, and an imaging optical unit. The acousto-optic tunable filter RF drive unit includes a broadband frequency synthesizer, a second RF power amplifier, and an acousto-optic tunable filter device. The central processing and control module 201 is connected to the serial data input pin, serial clock pin, and chip select enable pin of the broadband frequency synthesizer through a fourth serial peripheral interface bus. The RF signal output pin of the broadband frequency synthesizer is connected to the input port of the second RF power amplifier. The output port of the second RF power amplifier is connected to the piezoelectric transducer RF signal input port of the acousto-optic tunable filter device through an impedance matching circuit. The imaging optical unit includes a light source, which is a broadband halogen tungsten lamp or a light-emitting diode, a first collimating lens, a sample flow cell, a second collimating lens, and an acousto-optic tunable filter device. The broadband light emitted by the light source is collimated by the first collimating lens and then transmitted through the sample flow cell, which is supplied by the system integration and microfluidic platform 301 and filled with the sample to be tested. The emitted light is then collimated by the second collimating lens and then incident on the optical input end of the acousto-optic tunable filter device. The acousto-optic tunable filter device diffracts and outputs monochromatic light of a specific wavelength under the drive of the applied radio frequency signal. The high-speed spectral imaging detection unit includes a CMOS image sensor and its associated driving circuit. Monochromatic light diffracted by an acousto-optic tunable filter device is focused onto the photosensitive surface of the CMOS image sensor by an imaging lens. The exposure trigger pin of the image sensor is connected to the global exposure synchronization trigger signal line provided by the central processing and control module 201. Its image data output pin is connected to the image acquisition port of the central processing and control module 201 through a high-speed low-voltage differential signal bus. Its working mode configuration pin is connected to the central processing and control module 201 through a fourth serial peripheral interface bus. The titration verification module 105 includes a nanoscale precision fluid drive unit, a high-pressure pulse injection unit, a micro-reaction unit with an integrated optical detection window, and an in-situ optical monitoring interface; The nano-scale precision fluid drive unit includes a nano-scale injection pump driver, a piezoelectric actuator, and a pressure sensor. The central processing and control module 201 is connected to the serial data input pin, serial clock pin, and chip select enable pin of the nano-scale injection pump driver via a fifth serial peripheral interface bus. The high-voltage drive output pin of the nano-scale injection pump driver is connected to the electrode of the piezoelectric actuator. The piezoelectric actuator is mechanically installed inside the syringe barrel and drivenly connected to the plunger of the syringe to push a specific chemical quencher stored in the syringe. The pressure sensor is located at the flow path outlet end of the syringe barrel, and its analog signal output terminal is connected to the input pin of a fourth analog-to-digital converter. The data output pin of the fourth analog-to-digital converter is connected to the central processing and control module 201 via the fifth serial peripheral interface bus, forming a closed-loop flow control. The high-voltage pulse injection unit includes a high-voltage pulse generator and an injection electrode. The central processing and control module 201 is connected to the configuration pin of the high-voltage pulse generator through the fifth serial peripheral interface bus. The high-voltage pulse output pin of the high-voltage pulse generator is connected to the injection electrode located inside the micro-reaction unit for applying electrochemical-assisted injection or stirring pulses when needed. The microreaction unit with integrated optical detection window includes a microreaction cell with a transparent optical window, a stirrer integrated within the cell, and an injection electrode. The inlet of the microreaction cell is connected to the sample / reagent selection valve outlet of the system integration and microfluidic platform 301 via a microtube, receiving the water sample to be tested and the quenching agent from the syringe barrel delivered by the system integration and microfluidic platform 301. Its outlet is connected to the waste liquid pipeline, and the drive pin of the stirrer is connected to the general input / output pin of the central processing and control module 201. The in-situ optical monitoring interface is the optical window of the micro-reaction cell. The position of this window is aligned with the optical path of the collecting objective in the Raman spectroscopy detection module 102, so that the Raman spectroscopy detection module 102 can directly monitor the quenching reaction occurring in the micro-reaction cell in situ and in real time without moving the sample, and send the collected dynamic Raman spectral data to the central processing and control module 201 through its data bus. The central processing and control module 201 includes a heterogeneous computing core unit, a high-speed data interface unit, a precision clock and synchronization unit, a multi-channel control bus interface unit, and a system communication unit; The heterogeneous computing core unit includes a system-on-a-chip that integrates a programmable logic unit and a multi-core processor system, as well as a first dynamic random access memory connected to the multi-core processor system. The programmable logic unit is internally configured with multiple high-speed serial transceiver physical layer interfaces, a direct digital frequency synthesizer soft core, and hardware acceleration logic for implementing data preprocessing and model inversion algorithms. The high-speed data interface unit includes multiple interface connectors that are physically connected to the high-speed serial transceiver in the programmable logic unit. The first interface connector is connected to the first high-speed analog-to-digital converter in the optical coherence tomography module 101 via a first protocol data bus. The second interface connector is connected to the second high-speed analog-to-digital converter in the Raman spectroscopy detection module 102 via a second protocol data bus. The third interface connector is connected to the third high-speed analog-to-digital converter in the dielectric electrophoresis photothermal micro-perturbation module 103 via a third protocol data bus. The fourth interface connector is connected to the CMOS image sensor in the acousto-optic tunable filter spectral imaging module 104. The precision clock and synchronization unit includes a low phase noise clock generator, whose external reference clock pin is connected to the output of a temperature-controlled crystal oscillator. The multiple low-jitter clock output pins generated by the generator are respectively connected to the global clock input pin of the programmable logic unit of the system-on-a-chip and to the interface connectors of the high-speed data interface unit through differential clock lines. This is used to provide synchronous sampling clocks for all high-speed analog-to-digital converters and image sensors. At the same time, the multi-core processor system of the system-on-a-chip configures the clock generator through the first internal integrated circuit bus. The multi-channel control bus interface unit includes a multi-channel serial peripheral interface level conversion buffer array and an isolated controller LAN transceiver. The input of the buffer array is connected to the master device pins of multiple serial peripheral interfaces of the multi-core processor system of the system-on-a-chip. Its output serves as the first to fifth serial peripheral interface buses, respectively, and is connected to the configuration interfaces of the slave chips in the optical coherence tomography module 101, Raman spectroscopy detection module 102, dielectric electrophoresis photothermal micro-perturbation module 103, acousto-optic tunable filter spectral imaging module 104, and titration verification module 105. The controller-side pins of the isolated controller LAN transceiver are connected to the controller LAN controller pins of the system-on-a-chip. Its bus-side differential pins are connected to the system management microcontroller in the system integration and microfluidic platform 301 through the isolated controller LAN bus. The system communication unit includes an industrial Ethernet physical layer chip, whose media-independent interface pins are connected to the Ethernet media access controller of the multi-core processor system of the system-on-a-chip, and whose network transformer is connected to a standard Ethernet interface to enable network communication with an external monitoring system. The system integration and microfluidic platform 301 includes a system management and communication unit, a multi-channel fluid drive and control unit, and a sample flow path and temperature control unit; The system management and communication unit includes a system management microcontroller and a second dynamic random access memory and non-volatile memory connected thereto. The microcontroller's controller LAN controller pin is connected to the controller-side interface of the isolated controller LAN transceiver. The bus-side differential pin of the transceiver is connected to the isolated controller LAN transceiver in the central processing and control module 201 through the isolated controller LAN bus to receive control commands and report platform status. The multi-channel fluid drive and control unit includes a precision injection pump, a multi-channel injection pump driver, a multi-position multi-way solenoid valve array, and corresponding valve drive circuits. The serial peripheral interface pins of the multi-channel injection pump driver are connected to the system management microcontroller through the sixth serial peripheral interface bus. Its multi-channel motor drive output pins are respectively connected to the stepper motors of multiple precision injection pumps. Each solenoid valve coil in the multi-position multi-way solenoid valve array is driven by an independent low-side driver channel. The serial input pins of the low-side driver are connected to the system management microcontroller through the sixth serial peripheral interface bus. The sample flow path and temperature control unit includes a sample / reagent container, a flow path network consisting of polytetrafluoroethylene tubing and a multi-position multi-port solenoid valve array, an integrated Peltier temperature control module, and multiple flow and pressure sensors. The flow path network has a common inlet and multiple outlets. The multiple outlets of the flow path network are connected to the sample flow cell of the optical coherence chromatography module 101, the SERS detection chip of the Raman spectroscopy detection module 102, the microchannel of the dielectrophoresis photothermal micro-perturbation module 103, and the inlet of the micro-reaction cell of the titration verification module 105, respectively, through tubing. The driver control pin of the Peltier temperature control module is connected to the pulse width modulation output pin of the system management microcontroller, and its temperature feedback thermistor is connected to the analog-to-digital converter input pin of the microcontroller. The analog output terminal of the flow and pressure sensors is connected to the input channel of the fifth analog-to-digital converter, and the data output terminal of the fifth analog-to-digital converter is connected to the system management microcontroller through the sixth serial peripheral interface bus. The central processing and control module 201 is used to send synchronization commands, process all sensor data, run inversion algorithms, manage the verification process, and output the final results. Specifically, it includes the following steps: S1. Multi-source feature extraction and spatiotemporal registration: The central processing and control module 201 collects and preprocesses the raw measurement data of each module, extracts optical, spectral and physicochemical features related to pollutants, and unifies all features to the same time and space coordinate system. S1.1 Within a preset measurement time window, the central processing and control module 201 receives the original interference signal from the optical coherence tomography module 101, the original Raman spectral voltage signal from the Raman spectroscopy detection module 102, and the original hyperspectral image data from the acousto-optic tunable filter spectral imaging module 104 via a high-speed data bus, and reads the original voltage signal of the microcantilever beam response and the original signal of the dielectric photothermal micro-perturbation module 103 recorded by the control bus. S1.2 The central processing and control module 201 performs Fourier transform and point spread function deconvolution on the original interference signal to calculate the depth-resolved backscattering coefficient distribution. It extracts the average backscattering intensity and the scattering entropy that characterizes the uniformity of the scatterer distribution from the backscattering coefficient distribution. Based on the depth-resolved backscattering coefficient distribution, it calculates the probability distribution in the depth dimension and the information entropy of the probability distribution. The information entropy is used to quantify the stability of the colloidal dispersion system. S1.3 The central processing and control module 201 performs dark noise subtraction and instrument response correction on the original Raman spectral voltage signal, and then obtains the standardized Raman spectrum through fluorescence background fitting and subtraction. Then, the characteristic Raman peak intensity index of the target dye and surfactant is calculated through the spectral matching algorithm. S1.4 The central processing and control module 201 performs dark current and flat field correction on the original hyperspectral image data, calculates the absorbance matrix of each point in space, and extracts the average apparent absorbance at a wavelength of 254 nm and the average apparent absorbance in the characteristic absorption band of the dye. S1.5 The central processing and control module 201 performs phase-locked amplification analysis on the original voltage signal of the micro cantilever beam response, extracts its amplitude and phase change characteristics, analyzes the original dielectric response signal, and extracts the dielectric response spectrum characteristics. S1.6 The central processing and control module 201 adds a unified timestamp and spatial location label to all extracted features to complete spatiotemporal registration; S2. Dynamic joint inversion and concentration solution: Based on the extracted features, the central processing and control module 201 constructs and solves a forward physicochemical model that includes the actual chemical oxygen demand concentration and the concentration parameters of various interfering substances. The actual chemical oxygen demand concentration after stripping the composite interference is inverted through a nonlinear optimization algorithm. S2.1, Central Processing and Control Module 201 defines the state vector to be solved. The state vector includes the actual chemical oxygen demand concentration, dye concentration, surfactant concentration, average micelle radius, and synergistic enhancement factor of the interaction between dye and surfactant. S2.2, Central Processing and Control Module 201 constructs a forward observation model with the state vector as input. The forward observation model includes an absorbance prediction model, a Raman intensity prediction model, and a backscattering intensity prediction model. S2.3 The central processing and control module 201 uses all the spatiotemporally registered features extracted by S1 as observations to construct a global cost function; S2.4 The central processing and control module 201 uses the Levenberg-Marquardt algorithm to iteratively adjust the state vector and minimize the global cost function. When the iteration meets the convergence condition, it outputs the optimal state vector, which includes the final estimated value of the actual chemical oxygen demand concentration. The absorbance prediction model is the predicted total absorbance at any detection wavelength, which is equal to the sum of the absorption contribution of the actual chemical oxygen demand component, the dye absorption contribution corrected by the synergistic enhancement factor, and the spurious absorbance contribution caused by surfactant micellar scattering. The dye absorption contribution is the product of its standard molar absorbance and concentration, multiplied by a synergistic enhancement term related to the surfactant concentration. The spurious absorbance contribution is proportional to the surfactant concentration, the square of the micelle size, and the scattering efficiency factor calculated based on Mie theory. Among them, the Raman intensity prediction model predicts that the Raman intensity of the dye characteristic is proportional to its concentration, and the Raman intensity of the surfactant characteristic is proportional to its concentration. The backscattering intensity prediction model states that the predicted average backscattering coefficient is proportional to the square of the surfactant concentration and the micelle size.

[0027] The global cost function is the weighted sum of squares of the differences between each observation and the predicted values ​​of the corresponding absorbance prediction model, Raman intensity prediction model, and backscattering intensity prediction model, with an additional regularization constraint term on the state vector. S3. Credibility assessment and active verification: The central processing and control module 201 assesses the credibility of the inversion results. If the credibility is insufficient, it instructs the titration verification module 105 to perform a chemical verification experiment and uses the verification results to correct the inversion concentration, thereby updating the estimated value of the true chemical oxygen demand concentration. S3.1 The central processing and control module 201 calculates the comprehensive confidence score of this inversion result based on the minimum value of the global cost function and the estimated variance of each parameter in the optimal state vector. S3.2 If the overall confidence score is lower than the preset low confidence threshold, the central processing and control module 201 will determine the result as questionable and trigger the active verification process. S3.3 In the active verification process, the central processing and control module 201 instructs the system integration and microfluidic platform 301 to transport the water sample to be tested to the micro-reaction cell of the titration verification module 105, and instructs the titration verification module 105 to inject a calculated dose of specific chemical quencher based on the dye concentration estimate obtained by inversion. S3.4 The central processing and control module 201 synchronously controls the Raman spectroscopy detection module 102 to perform high-frequency monitoring of the micro-reaction cell and record the decay kinetic curve of the Raman characteristic peak intensity of the target dye. S3.5 The central processing and control module 201 compares the measured decay kinetic curve with the theoretical curve predicted based on the inverted dye concentration and the pre-stored reaction kinetic constant. If the deviation exceeds the allowable range, the measured decay data is used as a new constraint, and the process returns to step S2.4 for joint inversion iteration until a state vector that meets the verification conditions is obtained. The overall confidence score is calculated by taking the negative exponent of the ratio of the minimum value of the global cost function to a preset statistical threshold, and then multiplying it by a normalization factor for the probability density of a multidimensional Gaussian distribution calculated based on the state vector estimation covariance matrix. S4. Results Output and System Update: The central processing and control module 201 outputs the final true chemical oxygen demand concentration, the concentration of each interfering substance, and the data quality label, and saves the measurement data to update the parameters of the forward physicochemical model. S4.1 The central processing and control module 201 generates a structured monitoring report, which includes at least the final true chemical oxygen demand concentration value, the estimated concentration values ​​of each interfering substance, the corresponding measurement uncertainty, and the data quality label based on the comprehensive confidence score. S4.2 The central processing and control module 201 stores the complete data packet with a high confidence score in this measurement into the historical knowledge database. The complete data packet includes the original data, extracted features, inverted state vector and verification record. S4.3 The central processing and control module 201 periodically calls the data in the historical knowledge database and updates and optimizes the empirical parameters in the positive observation model through a regression learning algorithm. The empirical parameters include the basic absorption spectrum coefficient of the real chemical oxygen demand and the empirical relationship between the synergistic enhancement factor and the surfactant concentration.

[0028] To address the issue of inaccurate measurements using traditional ultraviolet absorption methods caused by the combined effects of high concentrations of linear alkylbenzene sulfonates and residual dyes in dyeing and printing wastewater, this study departs from the approach of directly estimating chemical oxygen demand (COD) based on absorbance at a single wavelength. Instead, multiple sensors with different physical principles are employed to work together. By measuring various properties of the water sample and processing the data using a computational program, the amount of oxidizable organic matter is ultimately obtained.

[0029] When the system starts working, multiple sensors simultaneously measure the water sample. The optical coherence tomography section emits a laser with an extremely short duration, receives and analyzes the scattered light signals returned from different depths inside the water sample. This process obtains the distribution information of tiny particles in the water sample. When the content of linear alkylbenzene sulfonates in the water is high, the micelles formed by them will significantly enhance the scattered light signal and show a specific distribution, thereby revealing the presence and state of linear alkylbenzene sulfonates.

[0030] The Raman spectroscopy measurement section utilizes the surface enhancement effect. A water sample flows through a chip with a special nanostructure that amplifies the signals of molecules adsorbed on its surface. This section emits another laser beam to illuminate the water sample on the chip, collecting the generated Raman scattered light and analyzing its composition. Different molecules produce scattered light of different wavelengths, forming unique spectra. By comparing these spectra, characteristic signals of dye molecules and linear alkylbenzene sulfonate molecules can be identified, thus confirming their types and estimating their relative abundance.

[0031] The photothermal perturbation component of dielectrophoresis applies an external influence to the water sample. It generates an alternating electric field in the microchannel, causing micelles and particles in the water to move. At the same time, another modulated laser beam locally heats the water sample, and the resulting minute deformation is detected by a microcantilever beam. The dye particles and linear alkylbenzene sulfonate micelles respond differently to the electric field and heat. By analyzing the differences in these responses, the influence of the two on the overall measurement results can be distinguished.

[0032] The acousto-optic tunable filter spectral imaging section acquires the absorption of light at multiple wavelengths by a water sample. It allows broadband light to pass through the water sample and then uses a filter controlled by a radio frequency signal to sequentially select monochromatic light of different wavelengths. The light intensity is recorded by a sensor, thereby obtaining the transmission or absorption images of the water sample at a series of wavelengths, forming complete spectral data, which includes comprehensive information on the contribution of dye color and the influence of particle scattering.

[0033] The central processing unit receives measurement data from all the above components. The program first aligns this data temporally and spatially. Then, it expresses the total absorbance as the output of a mathematical model. The model's input parameters include the actual chemical oxygen demand (COD) concentration, dye concentration, linear alkylbenzene sulfonate (LMSB) concentration, micelle size, and a factor describing their interaction. The model also specifies that the Raman signal intensity is proportional to the corresponding substance concentration, and the scattered signal intensity is related to the LSB concentration and micelle size.

[0034] The program employs an optimization calculation method, repeatedly adjusting the aforementioned input parameters to minimize the overall difference between the various signal values ​​calculated by the model (such as absorbance, Raman intensity, and scattering intensity) and the actual measured values. When the calculation process stops, a set of parameter values ​​is obtained, which includes the estimated chemical oxygen demand concentration after removing the dye absorption and linear alkylbenzene sulfonate scattering effects.

[0035] The system includes a verification step. When the program is unsure about the calculation results, it controls a fluid platform to deliver a small amount of water sample into a micro-reaction cell, along with precisely measured amounts of reagent that reacts with the dye. Simultaneously, the Raman spectroscopy measurement section monitors the changes in the dye signal during the reaction in real time. The observed signal attenuation process is compared with the predicted process based on the calculation results. If they match, the reliability of the calculation results is confirmed; if they do not match, the monitored attenuation data is used as new conditions to rerun the optimization calculation to obtain the corrected parameters.

[0036] Finally, the system generates a report providing the chemical oxygen demand (COD) value, information on potential interfering substances, and an explanation of the reliability of that value. Operators use this report to assess water quality and make appropriate treatment decisions, avoiding discharges based on erroneous readings.

[0037] The entire process involved measuring water sample properties from multiple angles, establishing a mathematical model to integrate data, using a calculation program to solve for the actual organic matter concentration, and setting up physical and chemical experiments for cross-validation, thus achieving the measurement of chemical oxygen demand under complex interference.

[0038] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A COD water pollution analysis system, characterized in that, include The optical coherence tomography module (101) is used to emit a femtosecond laser and detect interference signals to obtain the depth-resolved scattering distribution of the sample; The Raman spectroscopy detection module (102) is used to excite and acquire the Raman scattering spectrum of the sample for specific identification of dye and surfactant molecules; The dielectric electrophoresis photothermal micro-perturbation module (103) is used to apply a controllable electric field and photothermal excitation to separate different pollutants and measure their dielectric and thermal responses. The acousto-optic tunable filter spectral imaging module (104) is used to quickly scan and image the sample to obtain spatially resolved spectral absorbance data in the ultraviolet-visible band. The titration verification module (105) is used to inject precisely measured chemical quencher into the sample according to instructions to verify the accuracy of the optical measurement results. The central processing and control module (201) has its output terminals connected to the optical coherence tomography module (101), the Raman spectroscopy detection module (102), the dielectrophoresis photothermal micro-perturbation module (103), the acousto-optic tunable filter spectral imaging module (104), and the titration verification module (105), respectively. It is used to send synchronization commands, process all sensor data, run inversion algorithms, manage the verification process, and output the final results. The system integration and microfluidic platform (301) has its signal terminal connected to the output terminal of the central processing and control module (201), and it is connected to the sample flow cell of the optical coherence chromatography module (101), the micro-reaction cell of the Raman spectroscopy detection module (102), the microchannel of the dielectrophoresis photothermal micro-perturbation module (103), and the micro-reaction cell of the titration verification module (105) through microfluidic channels, so as to precisely control the delivery, mixing and distribution of all samples and reagents and even each module.

2. The COD water pollution analysis system according to claim 1, characterized in that, The process for sending synchronization commands, processing all sensor data, running the inversion algorithm, managing the verification process, and outputting the final result specifically includes the following steps: S1. Multi-source feature extraction and spatiotemporal registration: The central processing and control module (201) collects and preprocesses the raw measurement data of each module, extracts optical, spectral and physicochemical features related to pollutants, and unifies all features to the same time and space coordinate system. S2. Dynamic joint inversion and concentration solution: The central processing and control module (201) constructs and solves a forward physicochemical model based on the extracted features, which includes the actual chemical oxygen demand concentration and the concentration parameters of various interfering substances. The actual chemical oxygen demand concentration after stripping the composite interference is inverted through a nonlinear optimization algorithm. S3. Credibility assessment and active verification: The central processing and control module (201) assesses the credibility of the inversion results. When the credibility is insufficient, it instructs the titration verification module (105) to perform a chemical verification experiment and uses the verification results to correct the inversion concentration and update the estimated value of the real chemical oxygen demand concentration accordingly. S4. Results Output and System Update: The central processing and control module (201) outputs the final true chemical oxygen demand concentration, the concentration of each interfering substance, and the data quality label, and saves the measurement data to update the parameters of the forward physicochemical model.

3. The COD water pollution analysis system according to claim 2, characterized in that, The multi-source feature extraction and spatiotemporal registration, the central processing and control module (201) collects and preprocesses the raw measurement data of each module, extracts optical, spectral and physicochemical features related to pollutants, and unifies all features to the same time and space coordinate system, specifically including the following steps: S1.1 Within a preset measurement time window, the central processing and control module (201) receives the original interference signal from the optical coherence tomography module (101), the original Raman spectral voltage signal from the Raman spectroscopy detection module (102), and the original hyperspectral image data from the acousto-optic tunable filter spectral imaging module (104) via the high-speed data bus, and reads the original voltage signal of the microcantilever beam response and the original signal of the dielectric photothermal micro-perturbation module (103) recorded by the control bus. S1.2 The central processing and control module (201) performs Fourier transform and point spread function deconvolution on the original interference signal to calculate the depth-resolved backscattering coefficient distribution, and extracts the average backscattering intensity and the scattering entropy characterizing the uniformity of the scatterer distribution from it. S1.3 The central processing and control module (201) performs dark noise subtraction and instrument response correction on the original Raman spectral voltage signal, and then obtains the standardized Raman spectrum by fluorescence background fitting and subtraction, and then calculates the characteristic Raman peak intensity index of the target dye and surfactant by spectral matching algorithm. S1.4 The central processing and control module (201) performs dark current and flat field correction on the original hyperspectral image data, calculates the absorbance matrix of each point in space, and extracts the average apparent absorbance at a wavelength of 254 nm and the average apparent absorbance in the characteristic absorption band of the dye. S1.5 The central processing and control module (201) performs phase-locked amplification analysis on the original voltage signal of the micro cantilever beam response, extracts its amplitude and phase change characteristics, analyzes the original dielectric response signal, and extracts the dielectric response spectrum characteristics. S1.6 The central processing and control module (201) adds a unified timestamp and spatial location label to all extracted features to complete spatiotemporal registration.

4. The COD water pollution analysis system according to claim 2, characterized in that, The dynamic joint inversion and concentration solution, the central processing and control module (201) constructs and solves a forward physicochemical model based on the extracted features, including the actual chemical oxygen demand (COD) concentration and the concentration parameters of various interfering substances. The actual COD concentration after removing the composite interference is inverted through a nonlinear optimization algorithm, specifically including the following steps: S2.1 The central processing and control module (201) defines the state vector to be solved, which includes the actual chemical oxygen demand concentration, dye concentration, surfactant concentration, average micelle radius and synergistic enhancement factor of dye-surfactant interaction; S2.2 The central processing and control module (201) constructs a forward observation model with the state vector as input. The forward observation model includes an absorbance prediction model, a Raman intensity prediction model, and a backscattering intensity prediction model. S2.3 The central processing and control module (201) uses all the spatiotemporally registered features extracted in S1 as observation values ​​to construct a global cost function; S2.4 The central processing and control module (201) uses the Levenburg Marquardt algorithm to iteratively adjust the state vector and minimize the global cost function. When the iteration meets the convergence condition, it outputs the optimal state vector, which includes the final estimated value of the actual chemical oxygen demand concentration.

5. A COD water pollution analysis system according to claim 4, characterized in that, The absorbance prediction model is the predicted total absorbance at any detection wavelength, which is equal to the sum of the absorption contribution of the actual chemical oxygen demand component, the dye absorption contribution corrected by the synergistic enhancement factor, and the spurious absorbance contribution caused by surfactant micellar scattering. The dye absorption contribution is the product of its standard molar absorbance and concentration, multiplied by a synergistic enhancement term related to the surfactant concentration. The spurious absorbance contribution is proportional to the surfactant concentration, the square of the micelle size, and the scattering efficiency factor calculated based on Mie theory. The Raman intensity prediction model states that the predicted Raman intensity of the dye characteristic is proportional to its concentration, and the predicted Raman intensity of the surfactant characteristic is also proportional to its concentration. The backscattering intensity prediction model states that the predicted average backscattering coefficient is proportional to the square of the surfactant concentration and the micelle size.

6. The COD water pollution analysis system according to claim 4, characterized in that, The global cost function is the weighted sum of squares of the differences between each observation and the predicted values ​​of the corresponding absorbance prediction model, Raman intensity prediction model, and backscattering intensity prediction model, with an additional regularization constraint term on the state vector.

7. A COD water pollution analysis system according to claim 2, characterized in that, The credibility assessment and active verification, the central processing and control module (201) assesses the credibility of the inversion results. When the credibility is insufficient, it instructs the titration verification module (105) to perform a chemical verification experiment, and uses the verification results to correct the inversion concentration, and updates the estimated value of the true chemical oxygen demand concentration accordingly. The specific steps include the following: S3.1 The central processing and control module (201) calculates the comprehensive confidence score of the inversion result based on the minimum value of the global cost function and the estimated variance of each parameter in the optimal state vector; S3.2 If the overall confidence score is lower than the preset low confidence threshold, the central processing and control module (201) determines the result to be questionable and triggers the active verification process; S3.3 In the active verification process, the central processing and control module (201) instructs the system integration and microfluidic platform (301) to transport the water sample to be tested to the micro-reaction cell of the titration verification module (105), and instructs the titration verification module (105) to inject a calculated dose of specific chemical quencher according to the dye concentration estimate obtained by inversion. S3.4 The central processing and control module (201) synchronously controls the Raman spectroscopy detection module (102) to perform high-frequency monitoring on the micro-reaction cell therein and record the decay kinetic curve of the Raman characteristic peak intensity of the target dye; S3.5 The central processing and control module (201) compares the measured decay kinetic curve with the theoretical curve predicted based on the inverted dye concentration and the pre-stored reaction kinetic constant. If the deviation exceeds the allowable range, the measured decay data is used as a new constraint, and the process returns to step S2.4 for joint inversion iteration until a state vector that meets the verification conditions is obtained.

8. A COD water pollution analysis system according to claim 7, characterized in that, The comprehensive confidence score is calculated by taking the negative exponent of the ratio of the minimum value of the global cost function to a preset statistical threshold, and then multiplying it by a normalization factor for the probability density of a multidimensional Gaussian distribution calculated based on the state vector estimation covariance matrix.

9. A COD water pollution analysis system according to claim 2, characterized in that, The results output and system update involve the central processing and control module (201) outputting the final true chemical oxygen demand concentration value, the concentration of each interfering substance, and the data quality label, and saving the measurement data to update the parameters of the forward physicochemical model. The specific steps include: S4.1 The central processing and control module (201) generates a structured monitoring report, which includes at least the final true chemical oxygen demand concentration value, the estimated concentration values ​​of each interfering substance, the corresponding measurement uncertainty, and the data quality label based on the comprehensive confidence score. S4.2 The central processing and control module (201) stores the complete data packet with a high confidence score in this measurement into the historical knowledge database. The complete data packet includes the original data, extracted features, inverted state vector and verification record. S4.3 The central processing and control module (201) periodically calls the data in the historical knowledge database and updates and optimizes the empirical parameters in the positive observation model through the regression learning algorithm. The empirical parameters include the basic absorption spectrum coefficient of the real chemical oxygen demand and the empirical relationship between the synergistic enhancement factor and the surfactant concentration.

Images