A method for rapid detection of water quality of river-inlet based on spectroscopy and intelligent algorithm

Abstract

The application discloses a kind of based on spectroscopy and intelligent algorithm's river mouth water quality rapid detection method for discharging, it is related to water quality detection technical field, comprising the following steps: S001, establish spectrum and time series joint observation layer, obtain water body reflection light signal by the dual polarization scanning and multi-angle calibration in incident path, generate multidimensional spectral baseline matrix;S002, based on multidimensional spectral baseline matrix executes coherent phase analysis, the spectral signal in reflection path is decomposed layer by layer, extract the distribution characteristic map of optical interference order, and generate the shield mask of optical interference wave band.The application is realized to the accurate identification and correction of the spectral distortion caused by oil film interference by multi-angle optical acquisition, coherent phase analysis and multi-source data fusion, and constructs dynamic pollution track and adaptive threshold barrier, effectively improves the identification sensitivity, response speed and system stability of river mouth water quality detection for discharging.

Description

Technical Field

[0001] This invention relates to the field of water quality testing technology, specifically to a rapid water quality testing method for river discharge outlets based on spectroscopy and intelligent algorithms. Background Technology

[0002] "Rapid Detection of Water Quality at River Discharge Outlets Based on Spectroscopy and Intelligent Algorithms" refers to the use of spectral analysis technology to collect optical characteristics of the water body at discharge outlets in the ultraviolet-visible band without reagents. Intelligent algorithms then perform multi-dimensional correction, feature extraction, and modeling of the collected spectral data, enabling minute-level real-time identification and early warning of key pollution indicators such as COD, ammonia nitrogen, total phosphorus, and total nitrogen in complex water quality environments. Based on the material, this method not only achieves high-precision, low-cost, and reagent-free water quality detection through spectral holographic information and electrode interference correction, but also integrates neural networks and AI recognition algorithms to fuse and model multi-source data such as discharge outlet water quality, water quantity, and river remote sensing images. This allows for rapid capture of dynamic changes in pollutant concentrations, establishing an intelligent monitoring system for abnormal discharges, thereby achieving continuous online monitoring of river discharge outlets, abnormal pollution response, and early warning of water environment risks.

[0003] The existing technology has the following shortcomings: In existing technologies, water quality monitoring methods for river discharge outlets based on spectroscopy and intelligent algorithms typically rely on the absorption characteristics of water bodies in the ultraviolet-visible spectral region to reflect changes in pollutant concentrations. However, in complex aquatic environments, when oil films float near discharge outlets, a microscale thin film layer with optical interference properties forms on the surface. This interference layer produces irregular reflection and refraction effects on incident light, causing nonlinear distortion of spectral signals in some key bands. This distortion manifests as peak shifts in the spectral curve, energy distribution distortion, or abnormal enhancement or weakening, easily confused with normal signals from natural water quality fluctuations. Because existing intelligent recognition models often rely on the statistical regularities of training samples to determine fluctuation ranges, they lack the ability to accurately identify these atypical distortion characteristics caused by interference effects. Consequently, they are easily misidentified as general water quality noise fluctuations, leading to a failure to detect and warn of actual hazardous chemical emissions in a timely manner. This problem is particularly serious in sudden discharge scenarios, potentially causing long-term underestimation of pollutant concentrations and rendering the discharge outlet water quality monitoring system insensitive to major risk events, increasing the risk to aquatic environmental safety.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms, comprising the following steps: S001, Establish a joint observation layer of spectral and time series, obtain water body reflected light signals through dual polarization scanning and multi-angle calibration in the incident path, and generate a multi-dimensional spectral baseline matrix; S002 performs coherent phase analysis based on a multidimensional spectral baseline matrix, decomposes the spectral signal in the reflection path layer by layer, extracts the distribution feature map of the optical interference order, and generates a shielding mask for the optical interference band. S003, using the band region marked by the mask, replaces the high-saturation spectral segment with a low-exposure sequence to construct a time series stability score map and marks the nonlinear distortion positioning anchor point; S004 combines nonlinear distortion positioning anchor points with electrode signal data, flow pulse signal data, and remote sensing imaging data as inputs to perform cross-domain data fusion analysis, reconstruct the dynamic evolution trajectory of pollutant concentration, and locate the pollution peak range. S005, based on the evolution trajectory of pollutant concentration, performs dynamic scheduling of the time and frequency structure of the light source, uses double mirror time marker reordering to trigger phase conjugate scanning, suppresses the gain of optical interference signal and restores the true absorption spectral characteristics; S006, within the stable time window after spectral reconstruction is completed, performs time-reversal light field modulation, injects micro-patterns with the opposite phase to the interference, and controls the microbubble array to disturb the oil film thickness on the water surface, generating a detection threshold barrier with time evolution characteristics.

[0007] Preferably, step S001 includes: An observation base point is set up at the front end of the river discharge outlet area, and multiple observation units are set up in the horizontal direction. An observation arm with a spectral receiver is installed above each observation unit, and a polarization controller with beam modulation function is set at the front end of the receiver to construct a dual-polarization observation path. At the same time, a receiver component with a variable pitch angle is configured at each observation unit to form multiple fixed-angle observation channels. Each channel is connected to the main control device via optical fiber, and spectral data is collected synchronously using a unified timestamp. After acquisition, dark current correction and environmental background normalization were performed on the spectral data, and a reflectance correction factor was constructed by combining the ambient light detection results. Multi-channel difference maps are generated based on the differences in band reflection intensity between channels with different polarization directions and angles. These maps are then stitched together in the time dimension to generate a multi-dimensional spectral baseline matrix that combines spectral and temporal dimensions.

[0008] Preferably, step S002 includes: Based on the multidimensional spectral baseline matrix combining spectral and temporal dimensions, the spectral reflectance data under different incident angles and polarization directions are expanded into curves. By calculating the reflection intensity deviation of each angular channel band by band, a multi-angle fluctuation distribution map in the band direction is obtained, and high fluctuation segments that show dispersion or reversal in the angular dimension are selected as candidate bands for interference response. After locking the candidate bands, phase structure deconstruction is performed on the spectral curves under different angles and polarization conditions. The relative displacement of the peaks and the intensity shift at the same wavelength are measured, the corresponding interference order is calculated, and the two-dimensional interference order distribution map is reconstructed. The interference order distribution maps at different time points are stacked in sequence to generate a three-dimensional interference response map. The main interference band is identified based on the interference response intensity threshold, and an optical interference shielding mask is generated under the condition of continuous characteristics in the spatial and temporal dimensions. The generated masking mask is subjected to consistency correction across channels, and normalized by combining ambient light intensity and angle illumination conditions, outputting a multi-channel joint optical interference mask image that has been comprehensively corrected by time, angle, polarization and environmental parameters.

[0009] Preferably, step S003 includes: Extract the intervals marked as interference bands in the band region calibrated by the mask, and screen the spectral segments that cause signal saturation in the corresponding time series. Call the low-exposure parallel spectral sampling results under the same sampling time node, and back-calculate the equivalent intensity of the main channel based on the light intensity conversion coefficient and replace the high-saturation segments. After the replacement process is completed, the full-channel spectral data are normalized and aligned in their respective wavelength dimensions. The magnitude of the reflection intensity change of each wavelength point in the time series is calculated, and a time series stability score is constructed with wavelength as the horizontal axis and sampling time as the vertical axis. In the stability rating chart, wavebands with continuous non-periodic and drastic fluctuations in reflection values ​​are screened out, and wavebands with nonlinear distortion characteristics are identified by combining the time difference peaks, amplitude abrupt changes and waveform asymmetry features of multiple angle channels. The identified band points are cross-checked with the interference order distribution map and optical interference mask to confirm that they are in the high-order transition region and are covered by the mask, and are finally marked as nonlinear distortion positioning anchor points.

[0010] Preferably, the labeling of nonlinear distortion positioning anchor points is based on the identification criteria of asynchronous peak reflection intensity, abrupt amplitude change, and waveform asymmetry of the band point in multiple angular channels, and the intersection of the band point in the interference order jump region and the mask coverage region is used as the judgment condition.

[0011] Preferably, step S004 includes: The calibrated nonlinear distortion positioning anchor point data is extracted as the starting point for pollution trajectory identification and synchronized with the electrode signal data in time. During the period when the conductivity changes abruptly, the potential difference jumps and the acidity and alkalinity shift synchronously, the anchor points with spectral anomalies and telecommunication response consistency are calibrated as telecommunication coupling confirmation anchor points. Based on the confirmed anchor point of telecommunications coupling, flow pulse data is extracted, and the periods of sudden increase in flow velocity, flow direction offset angle and rapid change in turbulence intensity are analyzed to determine the pollutant transport path and flow direction, and the location of the pollution source is traced back based on the relationship between flow velocity and time. After completing the flow matching, remote sensing images with the same time as the anchor point are retrieved, and layers corresponding to the wavelength of the anchor point are extracted. Gray-scale distribution analysis is used to identify areas with abnormally enhanced reflectance and to determine the consistency between the pollution diffusion boundary and the water flow direction. By fusing and analyzing anchor point data, electrode signals, flow pulse data, and remote sensing images on a unified time axis, the trend of pollution concentration over time is calculated, the dynamic evolution trajectory of pollutant concentration is constructed, and the pollution peak interval is located.

[0012] Preferably, the step of jointly inputting the nonlinear distortion positioning anchor point with electrode signal data, flow pulse signal data and remote sensing imaging data is characterized in that the positioning of the pollution peak interval is based on the joint criteria of synchronous leap of spectral intensity in multiple channels, conductivity continuously higher than the average value by 25%, and the increase of the gray center value of the pollution area in the remote sensing image and the area reaching the maximum.

[0013] Preferably, step S005 includes: Based on the rapid rise and peak residence periods of pollutant concentration in the pollutant concentration evolution trajectory, the start and end boundaries of the light source response time window are determined, and the excitation time sequence structure of the light source is reconstructed so that the excitation frequency of the light source is higher and the emission duration is longer during periods with higher pollutant concentration, thereby improving the temporal resolution and energy input density at critical moments of pollution development. After completing the time scheduling, based on the distribution characteristics of the pollution interference band and the intrinsic absorption position band of the pollutant, the frequency structure of the light source output is adjusted, the transmission energy of the interference band is reduced, and the transmission power of the absorption band is increased and the bandwidth is reduced, so as to achieve interference suppression and absorption enhancement. After constructing the pollution-adaptive time-frequency light source structure, a dual-mirror time structure is introduced to reverse the pollution intensity curve on the time axis to generate a symmetrical sequence, and simultaneously trigger conjugate beams of opposite phase to form a time-phase conjugate coupling field to eliminate interference shifts caused by microscale optical path differences. Phase conjugate spectral scanning is initiated along the edge of the interference band, triggering forward and conjugate reverse beams and recording the phase synthesis results. Intensity restoration is performed on the spectral curve of the pollution response section to restore the true absorption spectral characteristics.

[0014] Preferably, step S006 includes: After completing the suppression of spectral distortion and the restoration of absorption peaks, the spectral stable period after conjugate scanning correction is selected as the activation window for the inverted light field. A light field time axis with symmetrical time markers is constructed to drive the reverse illumination path and form a reversible interference buffer layer, thereby realizing the time-backward intervention of the historical interference phase trajectory. While the inverted light field is established, a microstructure pattern with the opposite phase shift direction to that in the pollutant interference band is injected into the incident light path through a nanoscale optical interference device. Based on the phase difference quantization data of each main band in the interference enhancement region, an inverse phase shift is introduced at a specific spatial position, and the target incident region is activated through a spatial gating method. Based on pattern injection, microbubbles are released directionally using a high-density micro-nozzle array to dynamically disturb the oil film thickness distribution on the water surface. This causes the bubbles to disrupt the periodic stable structure of the oil film interference path during their ascent, and the disturbance rhythm is guided by the pollution concentration evolution spectrum. By coordinating the time-reversed light field, inverse phase pattern, and microbubble perturbation structure, a dynamic detection threshold barrier based on changes in pollution signal intensity is constructed within the pollution response stability window, thereby achieving rapid suppression of pollution interference signals and closed-loop control of the detection process.

[0015] Preferably, the microbubble disturbance is controlled by setting the rate of change of contaminant concentration, increasing the disturbance density when the contaminant concentration increases and decreasing the disturbance intensity when the contaminant concentration decreases, and is achieved through the asymmetric refraction offset formed by the contact of bubbles with the oil film. The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a joint spectral and temporal observation layer, introducing a dual-polarization, multi-angle optical acquisition mechanism to effectively acquire holographic reflection information of water bodies under complex interference scenarios. Then, through coherent phase analysis and spectral decomposition techniques, it accurately identifies the spectral nonlinear distortion characteristics caused by oil film interference, avoiding misjudging hazardous emissions as natural fluctuations. Simultaneously, by leveraging cross-domain multi-source data fusion, it constructs a dynamic trajectory of pollutant concentrations and further restores the true absorption spectrum through dual-mirror scheduling and phase conjugate intervention of the light source. Finally, by injecting inverse-phase micro-patterns within the spectral stability window and coordinating with bubble perturbation, it achieves physical reduction of interference noise and adaptive construction of the threshold barrier, forming a closed-loop control of the entire pollution detection process. Compared with existing technologies, this scheme not only significantly improves the sensitivity and response speed for identifying atypical pollution signals but also significantly enhances the stability and reliability of the detection system in the face of oil film interference and complex water quality emergencies. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of a rapid water quality detection method for river discharge outlets based on spectroscopy and intelligent algorithms, according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1 The method for rapid detection of water quality at river discharge outlets based on spectroscopy and intelligent algorithms, as shown, includes the following steps: S001, Establish a joint observation layer of spectral and time series, obtain water body reflected light signals through dual polarization scanning and multi-angle calibration in the incident path, and generate a multidimensional spectral baseline matrix containing interference response factors; To achieve stable acquisition of spectral data of water bodies flowing into rivers under complex environments such as oil film interference, it is necessary to first establish a joint observation layer of spectral and time series data to support subsequent interferometry identification and pollution source tracing analysis. This specifically includes the following steps: The river outfall area was selected as the observation target. Observation base points were set within a range of two to ten meters in front of the outfall, and five observation units were evenly distributed horizontally, with a two-meter spacing between each unit, covering a typical area from the direct inflow area to the river channel transition zone. An observation arm equipped with a spectral receiver was installed approximately one meter vertically above each observation unit. A polarization controller with beam modulation function was installed in front of the receiver, with polarization angles set to 0 degrees and 90 degrees corresponding to the horizontal and vertical polarization components, respectively. After passing through the polarization controller, the beam was split by a semi-transparent reflective surface into two independent detection paths, each guided to a separate spectral receiver. The receivers are uncooled photodiode arrays, capable of high-frequency sampling of continuous spectra in the 200 nm to 800 nm wavelength range. The data acquisition frequency for each polarization path was set to 20 times per second, with continuous acquisition for 30 seconds to construct a complete spectral time period. This dual-polarization structure allows for the simultaneous acquisition of the water surface's reflection response to linearly polarized light in both principal directions, thus more comprehensively capturing the anisotropic spectral variation behavior of the water surface under thin-film interference conditions.

[0020] To enhance interferometric recognition sensitivity by incorporating multi-angle reflection information, a variable elevation angle spectral receiver arm assembly is added to each observation unit. This assembly, supported by a mechanical bracket, forms five fixed-angle observation channels at 15°, 30°, 45°, 60°, and 75°. Each channel is equipped with an independent spectral receiver, which is guided via optical fiber to the main control unit for data synchronization. To ensure consistent incident time reference across different angles, all spectral receivers are equipped with GPS clock pulse receiving modules, using a unified timestamp to mark each spectral sample. Each angle group corresponds to a pair of polarization channels, collectively forming a complete ten-channel spectral acquisition array. During acquisition, by configuring fixed exposure time and constant integrated charge acquisition parameters, it is ensured that the received reflected light energy is recorded with a comparable intensity distribution at different incident angles. The acquisition cycle is uniformly set to complete one ten-channel scan every 10 seconds, with the spectral intensity of each scan recorded uniformly at 200 wavelength points. To address background interference caused by changes in ambient light, each observation unit is equipped with an additional pair of ambient light detectors facing the zenith and the sides. These detectors are used to measure the intensity of diffuse scattering from the sky and the angular background lighting conditions in real time, and are then used as reflectance correction factors in subsequent data processing.

[0021] After synchronous acquisition under different polarizations and angles, the acquired spectral data were preprocessed. First, dark current correction was performed on the spectral curve of each channel, and zero-point drift correction was performed based on the baseline values ​​recorded by the sensor under shading. Then, background light intensity normalization was performed. Based on the diffuse scattering intensity acquired by the environmental detector in the previous step, the background contribution coefficient of each channel at the same time was calculated, removing the influence of environmental background from the original signal. After environmental correction, a time gradient map was constructed based on the rate of change of spectral intensity between adjacent time points to preliminarily identify abrupt changes and abnormal shift behaviors. Based on this, a multi-channel difference spectral matrix was constructed by calculating the ratio of reflection intensity at the same wavelength for channels with the same angle under different polarization states, and the intensity difference between adjacent wavelengths for channels with different angles under the same polarization state. This matrix is ​​used to reveal the potential interference wave characteristics in the reflected light signal, which manifest as inconsistent trends at multiple angles in a specific band, and abnormal inverse jumps between polarization directions. This characteristic is usually caused by nonlinear spectral interference resulting from multipath reflection caused by an oil film interference layer with microscopic thickness variations on the water surface.

[0022] Based on processed multi-channel spectral data constructed using difference maps, a matrix stitching operation is performed in the time dimension to generate a complete multidimensional spectral baseline matrix combining spectral and temporal aspects. The three-dimensional coordinate axes of this matrix represent the sampling time point, wavelength range, and channel number (uniquely determined by a combination of polarization direction and angle), with each element representing the reflectance spectral intensity under the corresponding conditions. Through matrix slicing and band difference tracking methods, characteristic regions highly correlated with the interference response can be marked within the matrix; these are spectral segments that exhibit independent fluctuations across multiple channels, not synchronously changing over time. Furthermore, using time consistency verification rules, the fluctuation behavior of these characteristic regions is fitted, eliminating the synchronous disturbances caused by conventional water quality fluctuations, thus identifying these regions as sensitive areas for interference response. Finally, this spectral-temporal multidimensional spectral baseline matrix can serve as a data reference standard for subsequent coherence analysis, mask generation, distortion anchor point annotation, and contamination trajectory deduction, significantly enhancing the stability, sensitivity, and identification capability of the entire detection process under complex interference scenarios.

[0023] S002 performs coherent phase analysis based on a multidimensional spectral baseline matrix, decomposes the spectral signal in the reflection path layer by layer, extracts the distribution feature map of the optical interference order, and generates a shielding mask for the optical interference band. To identify interferometric spectral distortions caused by oil films on the water surface from the constructed joint spectral and time-series observation layer, coherent phase analysis needs to be performed based on the multidimensional spectral baseline matrix. This involves the following steps: Based on the constructed multidimensional spectral baseline matrix combining spectral and temporal coordinates, the spectral reflectance data corresponding to different incident angles and polarization directions at each time point are expanded into curves. Each set of expanded curves uses equally spaced wavelength coordinates as the horizontal axis and reflectance intensity as the vertical axis, forming a two-dimensional spectral reflectance line set. On this basis, the reflectance intensity deviation under all angular channels is calculated band by band, resulting in a multi-angle fluctuation distribution map along the band direction. This map reflects whether the light reflectance performance corresponding to the same wavelength is consistent under different observation angles. If the reflectance value of a certain wavelength point is densely distributed across all angles, it indicates that the spectral response of that band is stable and less susceptible to interference; conversely, if there is severe dispersion or even reversal, it indicates that the region may have multipath interference characteristics caused by surface microstructures (such as oil films). These high-fluctuation segments in the distribution map are initially screened as candidate regions for subsequent coherence analysis. To prevent misjudgment due to random noise, the temporal trend of reflectance intensity changes in each channel is differentially calculated during the screening process, eliminating band points that fluctuate in the same direction over time but have no asynchronous behavior in the angular dimension, thereby accurately identifying the starting region of spectral distortion related to interference.

[0024] After identifying candidate interference response bands, phase structure deconstruction is performed on the spectral curves of these bands under different angles and polarization conditions. Specifically, a reference channel (e.g., 45-degree incident angle, horizontal polarization) is selected, and its spectral curve in the target band is used as a baseline. Then, the peak and trough positions of the curves in the same band from other channels are compared with the baseline curve, and the relative displacement of peaks and intensity shifts between different curves at the same wavelength are measured. According to interference theory, this displacement corresponds to the optical path difference in the incident path, allowing for further calculation of the interference order at that point. The interference order values ​​of all channels in the same band are spatially reconstructed to form a two-dimensional interference order distribution map. This map uses angle as the vertical axis and wavelength as the horizontal axis, with each cell representing the interference order corresponding to a specific wavelength observed at that angle. By analyzing the jump trends and order continuity of this map in the wavelength dimension, it is possible to further determine whether there are non-uniform interference regions caused by uneven oil film thickness or refractive index variations. The graph is stacked in consecutive time frames, that is, the distribution graphs at different time points are superimposed in sequence to obtain a three-dimensional interferometric response spectrum, which can comprehensively reflect the trend and spatial distribution characteristics of a certain interferometric disturbance over time.

[0025] After obtaining the three-dimensional interferometric response spectrum, the main interfering bands are identified based on their spatial consistency and temporal persistence, and an optical interference shielding mask is generated. During mask generation, an interferometric response intensity threshold is first set. Based on the maximum interferometric order jump amplitude and average offset in each band of the interferometric order distribution obtained in the previous step, band points exceeding the set threshold are marked as interference points. Spatially, the interference point must exhibit consistent interference characteristics in at least three angular channels; temporally, it must maintain its interference state for at least three consecutive time points. Band points meeting these two conditions are identified as the main interfering response points. Using these main response points as the core, several wavelength points before and after them are enclosed and extended to form a complete band of interference shielding intervals. After integrating all shielding intervals on the wavelength axis, the final set of main interfering bands is obtained, and an optical interference shielding mask is constructed in Boolean form. The wavelength distribution accuracy of the mask is maintained at a resolution of no more than 2 nanometers to ensure fine spectral signal stripping in subsequent processing.

[0026] To further improve the adaptability and accuracy of the mask, consistency correction is needed for the generated mask across different channels. Specifically, the spectral curve of the portion overlapping with the mask band is extracted from each channel, and its continuity is analyzed with the curve segments in the non-masked region. If a significant morphological jump is found between the curve within the mask area and the curve outside the mask, and this jump exhibits consistent displacement characteristics across different angle channels, then this mask segment is confirmed as a genuine interference disturbance region and is retained. If the jump within the mask area is consistent with the trend outside the mask and is not verifiable in other channels, then this segment is marked as an uncertain mask area and is not used for the stripping operation. Furthermore, based on the ambient light intensity and irradiation conditions obtained in the first step, a background light normalization factor is added to the mask band for correction to prevent mask mislabeling under conditions of drastic changes in sunlight intensity. Finally, a multi-channel joint optical interference mask image, comprehensively corrected for time, angle, polarization, and environmental parameters, is output. This image exhibits high matching and recognition accuracy, providing reliable support for the stable reconstruction of subsequent spectral data and the extraction of contamination features.

[0027] S003, using the band region marked by the mask, replaces the high-saturation spectral segment with a low-exposure sequence to construct a time series stability score map and marks the nonlinear distortion positioning anchor point; To repair highly saturated spectral segments severely affected by interference disturbances and to accurately identify abnormal water quality signals, low-exposure replacement, time-series scoring, and distortion anchor point annotation were carried out based on the generated optical interference mask. The specific implementation process is as follows: In the established masking system, all intervals designated as interference bands are extracted, and spectral segments exhibiting signal saturation are selected from the corresponding time series. The criterion for saturation is that, at a specific sampling time, the spectral reflectance intensity at a certain wavelength continuously approaches the upper limit of the photodetector, accompanied by characteristics such as flattening of reflectance values, disappearance of peak structures, or a sudden drop in curve slope. After identifying the highly saturated segments, pre-set low-exposure parallel spectral sampling results at the same sampling time node are retrieved. Low-exposure sampling is achieved by shortening the exposure time and reducing the received light flux, maintaining sufficient dynamic range even under strong reflection conditions, thus avoiding signal saturation. These low-exposure data are recorded independently in each channel, covering the same wavelength range and time node as the main sampling data. These low-exposure segments are used as the basis for replacement data, and their corresponding exposure parameters, light flux settings, and actual received signal strength are recorded, providing a physical conversion basis for the next replacement processing step.

[0028] Low-exposure sampling data is applied to replace high-saturation segments to ensure that the replacement values ​​are consistent with the main sequence in intensity, avoiding discontinuities or abrupt changes in the time series. To achieve intensity alignment, an intensity conversion coefficient is established based on the overlapping data of the main sampling channel and the low-exposure channel at adjacent time points in the same wavelength band. This conversion coefficient is the measured ratio between the main sampling exposure parameters and the low-exposure parameters, used to back-calculate the equivalent intensity of the main channel from the low-exposure data. After conversion, the adjusted low-exposure value replaces the original high-saturation segment, and interpolation is performed at the boundaries of continuous wavelength segments to ensure a natural transition of the replacement segment in terms of curve shape with the normal segments before and after it, eliminating abrupt jumps or sawtooth errors in reflection values ​​caused by the replacement boundary. After completing this replacement process in each channel and each sampling time period, all channels are normalized and aligned in their respective wavelength dimensions to maintain comparability of reflection intensities between different angles and polarization directions, avoiding misjudgments of local anomalies caused by gain differences between channels. This replacement process is significantly superior to existing methods that directly discard high saturation values ​​or use fixed interpolation for filling. It is supported by real physical measurements and preserves key reflection structure details, providing more reliable basic data for subsequent distortion identification.

[0029] Based on the complete spectral data after substitution processing, a stability scoring map covering the entire time series was constructed. In this map construction process, wavelength was used as the horizontal axis and sampling time as the vertical axis, with the variation in reflection intensity of each band at different time points serving as the scoring criterion. Specifically, the maximum difference range, rate of change, and pattern of reflection values ​​for each wavelength point throughout the entire time series were calculated, and a single-point stability score was formed. A lower score indicates greater stability of the band over time, and is more likely to reflect the actual water quality background; a higher score indicates significant fluctuations in the band's reflection behavior, and is more likely to be affected by interference or pollution events. The scores of all bands were mapped into a two-dimensional stability scoring map, visualized using color depth, with darker colors representing high stability areas and lighter colors representing high volatility areas. By weighted fusion of the scoring maps from different polarization directions and angle channels, anomalous band segments exhibiting unstable trends in most observation paths can be further identified. This scoring map not only provides a means of quantifying band stability but also forms the spatiotemporal basis for identifying nonlinear distortions.

[0030] By combining the unstable regions in the scoring spectrum, nonlinear distortion positioning anchor points are marked. The marking of nonlinear distortion positioning anchor points is based on the identification criteria of asynchronous reflection intensity peaks, abrupt amplitude changes, and waveform asymmetry of the band point in multiple angular channels, and the intersection of the band point in the interference order jump region and the mask coverage region is used as the judgment condition.

[0031] Asynchronous peak reflection intensity, abrupt amplitude change, and waveform asymmetry refer to the occurrence of time-difference peaks, drastic jumps in reflection values ​​between adjacent time nodes, and asymmetrical changes in the peak and trough shapes of the spectral curves, respectively, when the selected band points in the scoring chart appear in different angle channels. Among them, the time-difference peaks reflect that the reflection intensity reaches its maximum value at different angles at different angles. Abrupt amplitude change is manifested as continuous and drastic fluctuations in reflection values ​​or a sudden drop in the slope of the curve. Waveform asymmetry is reflected as different rates of change of the curve at the rising and falling edges. Such fluctuations are not caused by uniform changes in solute concentration, but are triggered by nonlinear processes such as surface micro-interference, thin film reflection, or eddy current disturbance.

[0032] The high-order jump region refers to the location in the interference order distribution map where the interference order changes abruptly in the wavelength dimension during coherent phase analysis. This is achieved by calculating the relative displacement and intensity shift of the wave peaks under different angles and polarization channels. This location reflects the non-uniform interference region caused by uneven oil film thickness or refractive index variations, and represents the most significant spatial range of the interference response. The mask coverage region, on the other hand, refers to the band points marked according to the interference response intensity threshold in the interference order distribution map and the three-dimensional interference response spectrum. These band points are then selected as continuous wavelength intervals of the main interference band after spatial consistency and temporal persistence screening, and a spectral band mask is constructed in Boolean form. The high-order jump region reflects the location of drastic changes in interference characteristics, while the mask coverage region corresponds to the band where the interference effect persists. The intersection of the two is used to confirm the validity of the nonlinear distortion band points.

[0033] First, bands exhibiting continuous, non-periodic, and drastic fluctuations in reflectance values ​​are selected from the scoring map. These bands are then compared across multiple channels. If a point shows temporally variable peaks, abrupt amplitude changes, or waveform asymmetry at different angles, it indicates that the fluctuations are not caused by uniform solute concentration changes, but rather by nonlinear processes such as surface micro-interference, thin-film reflection, and eddy current disturbances. Based on this, the interference order distribution map and mask map generated in the second step are cross-validated to confirm that the point is indeed in a high-order transition region and covered by an interference mask. Combined with the light intensity change rate at the current time point, it is finally determined as an effective nonlinear distortion localization anchor point. For each effective anchor point, its wavelength value, time series position, angle channel number, stability score, and anchor point strength value are recorded and marked on the final anchor point distribution map, forming a nonlinear distortion identification map with clear physical basis and spatiotemporal coordinates. This map will serve as a reference starting point for the evolution of pollution trajectories in subsequent steps, providing spatial and temporal support for pollution peak locking, thereby establishing a high-resolution, high-sensitivity foundation for pollution source tracing.

[0034] S004 combines nonlinear distortion positioning anchor points with electrode signal data, flow pulse signal data, and remote sensing imaging data as inputs to perform cross-domain data fusion analysis, reconstruct the dynamic evolution trajectory of pollutant concentration, and locate the pollution peak range. Based on the completion of nonlinear distortion localization anchor point annotation, in order to identify the evolution trend and peak concentration location of pollution events, it is necessary to fuse and analyze the spectral anchor points with electrode signals, flow data, and remote sensing images. The specific implementation process is as follows: Data from calibrated nonlinear distortion positioning anchor points are extracted as the starting point for pollution trajectory identification. Each anchor point contains characteristic information including wavelength position, corresponding time node, angle observation channel number, polarization direction, score value, and reflection intensity abrupt change amplitude. In the pollution data processing workflow, all anchor points are first sorted chronologically and corresponding to water quality electrode measurement data at the same time point. Water quality electrodes are set at three locations: upstream of the drainage outlet, at the outlet edge, and downstream extension, collecting information on water conductivity, potential difference, and pH changes at a sampling frequency of once per second. Within the time window of each anchor point, the signal gradient changes between the three measurement points are analyzed. If, within the anchor point's time period, there is a jump in potential difference greater than three times the normal fluctuation value, a short-term abrupt change in conductivity from low to high or high to low, and a synchronous shift in pH value, it indicates that the time period of the anchor point not only exhibits nonlinear anomalies in the spectrum but also has pollution interference characteristics in its physical and electrical properties. These anchor points are selected as "telecommunication coupling confirmation anchor points" and proceed to subsequent processing. This process can effectively eliminate isolated spectral noise, ensuring that the trajectory construction is based on authentic sources and responds synchronously with changes in water quality.

[0035] Synchronous analysis of flow pulse data was performed on distorted anchor points confirmed by telecommunications coupling to further clarify the pollutant transport path and concentration propagation direction. Specifically, flow records were extracted for five minutes before and after each anchor point. These records were acquired by bidirectional ultrasonic flowmeters deployed in the cross-section inside the discharge outlet, covering instantaneous flow velocity, flow pulse frequency, flow direction offset angle, and turbulence intensity changes. Sampling was performed once per second to form a high-temporal-resolution flow data sequence. During data comparison, if the flow velocity at a certain anchor point showed a short-term surge, the pulse frequency changed from uniform to non-periodic fluctuation, or the flow direction offset angle exceeded ten degrees, it indicated that there might be external disturbance in the water body at this time, i.e., pollutant discharge was accompanied by instantaneous hydrodynamic changes. Based on these characteristics, the anchor point was located as the concentration front occurrence point on the flow axis. Subsequently, by combining the flow velocity value and pollutant flow direction, possible pollutant source areas in the forward time period were traced back. Using the pollutant flow velocity and distance calculation formulas, the transmission time from the pollution source location to the anchor point was estimated, and the anchor point forward path was constructed. This step enables the extensional mapping of anchor points from temporal information to spatial information, providing a basis for azimuth coordinates for subsequent remote sensing data comparison and trajectory stitching.

[0036] Based on the initial fusion of anchor points and hydraulic responses, remote sensing image information was incorporated for verifying the spatial expansion of pollution. The remote sensing image acquisition device was set up on a three-axis gimbal platform 8 meters vertically above the discharge outlet, equipped with a high-resolution multispectral imaging device. The sampling frequency was once every 30 seconds, and the imaging range covered a 50-meter radius area around the discharge outlet, with a spectral coverage of 400 to 1000 nanometers. For each time point corresponding to the coupled anchor point, remote sensing image slices were retrieved, and the remote sensing channel layer closest to the reflection wavelength of the anchor point was extracted and processed. During image processing, the pollution area boundary was extracted through difference enhancement, regional grayscale distribution analysis, and edge contrast expansion. If the image exhibited asymmetric enhancement of water reflectance, a strip-like diffusion structure, or blurred edge reflection boundaries, and showed a significant area expansion trend compared to previous and subsequent time slices, it indicated that the pollutants had entered a diffusion state at the time point corresponding to that anchor point. The pollution boundary in the remote sensing image was extrapolated outward to form a pollutant diffusion area map, which was then overlaid and analyzed with the aforementioned flow tracking direction. If the image diffusion direction closely matches the flow path, confirming that the pollutant movement is consistent with the hydrodynamic direction, then the boundary of the polluted area in the remote sensing image can be used as the basis for the spatial extension of the pollution trajectory. This step achieves four-dimensional data integration, from anchor point temporal positioning, electrode response property verification, hydrodynamic supplementation of the flow structure, to the identification of polluted sections in remote sensing space, thus constructing a panoramic propagation structure of the pollution event.

[0037] All data sources were matched and stitched together on a unified timeline to form a dynamic evolution trajectory of pollutant concentration. The stitching process was executed according to the following logic: anchor points were used as time reference points, and the flow path was traced back to the possible initial point of pollution release. The time point and location of the first abrupt change in conductivity in the electrode data were used as the pollution initiation marker. Subsequently, based on the trend of changes in the area and central gray value of the pollution diffusion region in the remote sensing image, the concentration growth rate at each anchor point was calculated. A sequence map of the change in pollution concentration over time was established, with time as the horizontal axis and the estimated concentration value as the vertical axis, marking key change points, peak points, and decreasing trend points. When identifying pollution peak intervals, the following conditions must be met: first, the spectral intensity shows a synchronous jump in multiple channels and lasts for more than 3 time points; second, the conductivity increases and remains above 25% of the average value; and third, the pollution area in the remote sensing image reaches its maximum and the central gray value is more than 15% higher than that of the adjacent time period. The time period that meets the above three conditions is identified as the pollutant concentration peak interval and marked as a "high-risk pollution area" in the map. This trajectory map will be used in subsequent light source control and precise spectral reconstruction processes, providing key parameter inputs for time windows, location segments, and pollution intensity.

[0038] S005, based on the evolution trajectory of pollutant concentration, performs dynamic scheduling of the time and frequency structure of the light source, uses double mirror time marker reordering to trigger phase conjugate scanning, suppresses the gain of optical interference signal and restores the true absorption spectral characteristics; Based on the obtained pollutant concentration evolution trajectory and peak concentration segment identification results, in order to avoid misjudgment of spectral distortion caused by oil film interference, it is necessary to perform joint scheduling of the time and frequency structure of the light source, and introduce temporal rearrangement and phase conjugate scanning to suppress the interference signal enhancement phenomenon and restore the true absorption characteristics. The specific process includes the following steps: Based on the rapid increase and peak residence periods of pollution concentration marked in the pollution evolution trajectory, the start and end boundaries of the light source response window are determined. This response window starts at the time when the pollution intensity changes significantly and extends to the end of the pollution concentration peak, covering the entire process of rapid pollutant accumulation. For each sampling time point within this response window, the light source excitation time sequence structure is reconstructed. The core principle of this reconstruction is: the higher the pollution concentration, the higher the light source excitation frequency and the longer the emission duration; the faster the pollution concentration changes, the shorter the time interval between light source activations, forming a dynamic pulse-like excitation rhythm. Through this scheduling method, the light source has higher temporal resolution and energy input density at critical moments in pollution development, significantly improving the response speed of spectral detection to changes in pollution signals. This type of time structure reconstruction differs from the traditional equal-step light source sampling logic; instead, it achieves real-time rhythm adjustment based on the pollution evolution rate and intensity, constituting a light source-level coordinated response to the dynamic state of pollution events.

[0039] After time scheduling is completed, the frequency structure of the light source output is precisely controlled based on the specific distribution characteristics of the pollution interference bands. The goal of frequency structure control is to actively suppress the pollutant interference bands and enhance the pollution identification bands. Specifically, band clusters highly correlated with the oil film interference response in the previous stage of pollution evolution trajectory are extracted and designated as "suppression bands"; simultaneously, the intrinsic absorption position bands of the pollutants are extracted and designated as "enhancement bands." For the suppression bands, emission energy is reduced, i.e., within the pollution response time window, the light source radiation intensity of these bands is decreased to reduce multiple reflections and superimposed resonances caused by them in the oil film interference layer; for the enhancement bands, emission power is increased and their bandwidth is reduced, concentrating energy near the characteristic absorption peaks of the pollutants, thereby enhancing the resolution of the spectral absorption structure. This control strategy forms a pollution-adaptive spectral distribution in both spatiotemporal dimensions, balancing interference suppression and pollution identification, and possessing a clear response hierarchy and spectral reconstruction capability. Compared with traditional light sources with fixed spectral width and constant energy output, this scheme can dynamically switch spectral forms according to the pollution state, greatly improving the effective identification capability under interference conditions.

[0040] Based on the completed pollution-adaptive time-frequency light source structure, a time-mirror structure is introduced to achieve active conjugate intervention on the interference superposition path. This structure consists of two sets of time sequences: one is a forward sampling time sequence, i.e., the actual occurrence time sequence of pollution events; the other is a mirror sequence, which is generated by reversing the pollution intensity curve on the time axis to generate a symmetrical sequence and set as the mirror time reference. In actual operation, at each forward time point, when the light source is triggered, a corresponding conjugate trigger at the reverse time point is simultaneously activated, so that the two beams of light form an interference symmetry point in the middle of the actual spatial path. These two beams of light carry opposite phase information, and through spatial superposition within the same wavelength band, phase cancellation is achieved, thereby forming a time-phase conjugate coupling field. This field has a precise correction effect on the phase shift caused by the microscale optical path difference formed by the oil film. After each forward light source scan is completed, the corresponding mirror sequence immediately triggers a reverse scan, so that multipath interference within the pollution interference section is dynamically suppressed. This mirror rearrangement mechanism does not rely on complex numerical modeling. Instead, it starts from the intrinsic characteristics of light propagation to construct an active interference neutralization mechanism, thereby achieving structural-level elimination of nonlinear reflection signals. This is a photophysical control behavior that traditional spectrometers cannot achieve.

[0041] After completing the temporal structure rearrangement and conjugate path establishment, a phase conjugate spectral scan is performed to restore the spectral curves and accurately reproduce the absorption peaks. The phase conjugate scan starts at the edge of the interference band and progresses gradually along the wavelength direction, with each scan step set to 2 nanometers to ensure that each typical interference peak and absorption groove is covered. At each band point, a forward beam and a conjugate reverse beam are triggered, and the difference in reflection response and phase synthesis result of the two beams in that band are recorded. If a sudden drop in reflection intensity and a near-zero phase synthesis result occur at a specific band point, it is equivalent to the phase interference at that point being completely canceled, confirming that the band as a typical interference enhancement region. After performing the same operation on all interference bands, a conjugate interference suppression spectrum is generated. Based on the spectrum, intensity restoration processing is performed on the spectral curves of the original pollution response region to remove peak distortion, absorption valley tailing, and boundary blurring caused by reflection enhancement, ultimately restoring a standard spectral structure that matches the true absorption characteristics of the pollutant. This scanning mechanism is driven by pollution response and has the ability to suppress directional pollution, selectively identify pollution and repair structural pollution. It can improve the spectral detection accuracy of this invention to a high stability level with an error of less than 3% under extreme interference background, providing high-confidence basic data for subsequent pollutant component identification and source tracing modeling.

[0042] S006, within the stable time window after spectral reconstruction is completed, performs time-reversal light field modulation, injects micro-patterns with the opposite phase to the interference, and controls the microbubble array to disturb the oil film thickness on the water surface, generating a detection threshold barrier with time evolution characteristics, thereby achieving rapid suppression of abnormal spectral signals and closed-loop control of the detection process; After completing the spectral distortion suppression and absorption peak restoration operations during the critical pollution period, to further enhance the stability and robustness of pollution detection, a time-reversal optical field needs to be deployed during the stable period, and an active intervention structure needs to be introduced to achieve dynamic extinguishing and closed-loop control of the pollution interference signal. The specific steps are as follows: A spectrally stable period after conjugate scanning correction was selected as the activation window for the inverted light field. Quantitative analysis of the slope of the reflection curves, the rate of change of the first derivative, and the stability of the absorption peak-valley spacing at multiple consecutive sampling nodes within this period confirmed that the spectral curve was in a steady-state phase with minimal disturbance. A light field time axis was constructed centered on this steady-state time segment. This light field time axis added a set of symmetrical time markers to the original sampling sequence. Each marker corresponds to an original light source excitation time point, forming a forward and reverse paired structure to drive the construction of the reverse illumination path. In this time structure, the original forward illumination is used to trigger normal pollution absorption characteristics, while the reverse time sequence is used to excite a light field with phase cancellation capabilities, achieving temporal retrospective intervention of historical interference phase trajectories, thereby constructing a reversible interference buffer layer in the spatial path.

[0043] Simultaneously with the establishment of the inversion light field, a microstructure pattern is injected into the incident light path using a precisely controlled nanoscale optical interferometer. This pattern embeds microscale interference fringes in space, with the phase shift direction opposite to that in the pollutant interference band. Each fringe is reconstructed through a micro-reflective layer within the optical crystal material. Based on the quantized phase difference data of each main band in the previously enhanced interference region, a corresponding reverse phase shift is introduced at a specific spatial location. The pattern achieves high-precision positioning at the subwavelength scale, avoiding the introduction of secondary interference. The injection process employs a spatial gating method to ensure that the pattern is activated only in the incident region corresponding to the target interference band, and pattern stability is maintained through continuous illumination and high-frequency flicker control. This pattern has real-time response capability, changing synchronously with each forward and reverse switch of the inversion light field, achieving optical pattern coupling synchronized with the pollution evolution cycle. This actively cancels the superposition effect of optical interference peaks in the incident path, improving spectral stability and feature recognition accuracy.

[0044] To further disrupt the stability of the interference path formed by the oil film layer on the water surface, microbubble perturbation is employed to dynamically disturb the oil film thickness distribution. Microbubbles are released directionally through a high-density array of micro-nozzles positioned within a support above the outlet. The nozzle size is controlled within 100 micrometers, and the release rate and position can be precisely controlled via current. As the bubbles rise to the surface, they are driven by buoyancy, forming a localized thickness disturbance region upon contact with the oil film surface, with a radius of up to 5 millimeters. Each disturbance point acts spatially on a single reflection path within the oil film interference layer, causing an asymmetric refraction shift of the incident light along that path, thus breaking the periodic stability of the interference fringes. The bubble release rhythm is guided by the pollution concentration evolution spectrum; the disturbance region is densified when the concentration increases, and the release frequency is reduced when the concentration decreases. The perturbation time interval can be precisely set within 200 milliseconds to ensure real-time response to sudden pollution changes. Microbubble perturbation and optical field inversion are executed in tandem, creating a synergistic intervention mechanism between physical perturbation and optical control of the pollution interference signal, significantly reducing the probability of spectral misinterpretation.

[0045] By coordinating the time-reversed optical field, inverse phase pattern, and microbubble perturbation structure, a dynamic detection threshold barrier with adaptive adjustment capabilities is constructed within the pollution response stability window. This barrier uses changes in the pollution response signal intensity as feedback. When a nonlinear upward trend in the reflection intensity of the target band exceeding a set threshold is detected, it automatically activates pattern modulation amplitude enhancement and microbubble frequency encryption functions, while simultaneously extending the inversion optical field's action period. The entire process continuously adjusts its response intensity based on the dynamic changes in pollution, forming a real-time closed-loop relationship between pollution identification and interference suppression. The threshold barrier activation has a threshold memory characteristic, maintaining the suppressed state until the pollution signal intensity returns to a stable state, ensuring uninterrupted monitoring response during periods of high pollution incidence. Compared to traditional static threshold judgment mechanisms, this structure possesses a triple defense line of spatial perturbation, phase cancellation, and time-dynamic judgment, effectively constructing a physical isolation boundary for pollution identification. It can control the intervention time of abnormal interference signals to within one second, significantly improving the response efficiency and monitoring accuracy of high-risk discharge outlet pollution events.

[0046] This invention constructs a joint spectral and temporal observation layer, introducing a dual-polarization, multi-angle optical acquisition mechanism to effectively acquire holographic reflection information of water bodies under complex interference scenarios. Then, through coherent phase analysis and spectral decomposition techniques, it accurately identifies the spectral nonlinear distortion characteristics caused by oil film interference, avoiding misjudging hazardous emissions as natural fluctuations. Simultaneously, by leveraging cross-domain multi-source data fusion, it constructs a dynamic trajectory of pollutant concentrations and further restores the true absorption spectrum through dual-mirror scheduling and phase conjugate intervention of the light source. Finally, by injecting inverse-phase micro-patterns within the spectral stability window and coordinating with bubble perturbation, it achieves physical reduction of interference noise and adaptive construction of the threshold barrier, forming a closed-loop control of the entire pollution detection process. Compared with existing technologies, this scheme not only significantly improves the sensitivity and response speed for identifying atypical pollution signals but also significantly enhances the stability and reliability of the detection system in the face of oil film interference and complex water quality emergencies.

[0047] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms, characterized in that, Includes the following steps: S001, Establish a joint observation layer of spectral and time series, obtain water body reflected light signals through dual polarization scanning and multi-angle calibration in the incident path, and generate a multi-dimensional spectral baseline matrix; S002 performs coherent phase analysis based on a multidimensional spectral baseline matrix, decomposes the spectral signal in the reflection path layer by layer, extracts the distribution feature map of the optical interference order, and generates a shielding mask for the optical interference band. S003, using the band region marked by the mask, replaces the high-saturation spectral segment with a low-exposure sequence to construct a time series stability score map and marks the nonlinear distortion positioning anchor point; S004 combines nonlinear distortion positioning anchor points with electrode signal data, flow pulse signal data, and remote sensing imaging data as inputs to perform cross-domain data fusion analysis, reconstruct the dynamic evolution trajectory of pollutant concentration, and locate the pollution peak range. S005, based on the evolution trajectory of pollutant concentration, performs dynamic scheduling of the time and frequency structure of the light source, uses double mirror time marker reordering to trigger phase conjugate scanning, suppresses the gain of optical interference signal and restores the true absorption spectral characteristics; S006, within the stable time window after spectral reconstruction is completed, performs time-reversal light field modulation, injects micro-patterns with the opposite phase to the interference, and controls the microbubble array to disturb the oil film thickness on the water surface, generating a detection threshold barrier with time evolution characteristics.

2. The rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S001 includes: An observation base point is set up at the front end of the river discharge outlet area, and multiple observation units are set up in the horizontal direction. An observation arm with a spectral receiver is installed above each observation unit, and a polarization controller with beam modulation function is set at the front end of the receiver to construct a dual-polarization observation path. At the same time, a receiver component with a variable pitch angle is configured at each observation unit to form multiple fixed-angle observation channels. Each channel is connected to the main control device via optical fiber, and spectral data is collected synchronously using a unified timestamp. After acquisition, dark current correction and environmental background normalization were performed on the spectral data, and a reflectance correction factor was constructed by combining the ambient light detection results. Multi-channel difference maps are generated based on the differences in band reflection intensity between channels with different polarization directions and angles. These maps are then stitched together in the time dimension to generate a multi-dimensional spectral baseline matrix that combines spectral and temporal dimensions.

3. The rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S002 includes: Based on the multidimensional spectral baseline matrix combining spectral and temporal dimensions, the spectral reflectance data under different incident angles and polarization directions are expanded into curves. By calculating the reflection intensity deviation of each angular channel band by band, a multi-angle fluctuation distribution map in the band direction is obtained, and high fluctuation segments that show dispersion or reversal in the angular dimension are selected as candidate bands for interference response. After locking the candidate bands, phase structure deconstruction is performed on the spectral curves under different angles and polarization conditions. The relative displacement of the peaks and the intensity shift at the same wavelength are measured, the corresponding interference order is calculated, and the two-dimensional interference order distribution map is reconstructed. The interference order distribution maps at different time points are stacked in sequence to generate a three-dimensional interference response map. The main interference band is identified based on the interference response intensity threshold, and an optical interference shielding mask is generated under the condition of continuous characteristics in the spatial and temporal dimensions. The generated masking mask is subjected to consistency correction across channels, and normalized by combining ambient light intensity and angle illumination conditions, outputting a multi-channel joint optical interference mask image that has been comprehensively corrected by time, angle, polarization and environmental parameters.

4. The rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S003 includes: Extract the intervals marked as interference bands in the band region calibrated by the mask, and screen the spectral segments that cause signal saturation in the corresponding time series. Call the low-exposure parallel spectral sampling results under the same sampling time node, and back-calculate the equivalent intensity of the main channel based on the light intensity conversion coefficient and replace the high-saturation segments. After the replacement process is completed, the full-channel spectral data are normalized and aligned in their respective wavelength dimensions. The magnitude of the reflection intensity change of each wavelength point in the time series is calculated, and a time series stability score is constructed with wavelength as the horizontal axis and sampling time as the vertical axis. In the stability rating chart, wavebands with continuous non-periodic and drastic fluctuations in reflection values ​​are screened out, and wavebands with nonlinear distortion characteristics are identified by combining the time difference peaks, amplitude abrupt changes and waveform asymmetry features of multiple angle channels. The identified band points are cross-checked with the interference order distribution map and optical interference mask to confirm that they are in the high-order transition region and are covered by the mask, and are finally marked as nonlinear distortion positioning anchor points.

5. A rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 4, characterized in that, The labeling of nonlinear distortion positioning anchor points is based on the identification criteria of asynchronous peak reflection intensity, abrupt amplitude change and waveform asymmetry of the band point in multiple angular channels, and the intersection of the band point in the interference order jump region and the mask coverage region is used as the judgment condition.

6. The rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S004 includes: The calibrated nonlinear distortion positioning anchor point data is extracted as the starting point for pollution trajectory identification and synchronized with the electrode signal data in time. During the period when the conductivity changes abruptly, the potential difference jumps and the acidity and alkalinity shift synchronously, the anchor points with spectral anomalies and telecommunication response consistency are calibrated as telecommunication coupling confirmation anchor points. Based on the confirmed anchor point of telecommunications coupling, flow pulse data is extracted, and the periods of sudden increase in flow velocity, flow direction offset angle and rapid change in turbulence intensity are analyzed to determine the pollutant transport path and flow direction, and the location of the pollution source is traced back based on the relationship between flow velocity and time. After completing the flow matching, remote sensing images with the same time as the anchor point are retrieved, and layers corresponding to the wavelength of the anchor point are extracted. Gray-scale distribution analysis is used to identify areas with abnormally enhanced reflectance and to determine the consistency between the pollution diffusion boundary and the water flow direction. By fusing and analyzing anchor point data, electrode signals, flow pulse data, and remote sensing images on a unified time axis, the trend of pollution concentration over time is calculated, the dynamic evolution trajectory of pollutant concentration is constructed, and the pollution peak interval is located.

7. A rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms, as described in claim 6, is characterized in that... The step of jointly inputting nonlinear distortion positioning anchor points with electrode signal data, flow pulse signal data and remote sensing imaging data is characterized by the positioning of pollution peak intervals based on the joint criteria of synchronous leap in spectral intensity in multiple channels, conductivity continuously exceeding the average value by 25%, and the increase in the gray center value of the pollution area in the remote sensing image and the maximum area.

8. The rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S005 includes: Based on the rapid rise and peak residence periods of pollutant concentration in the pollutant concentration evolution trajectory, the start and end boundaries of the light source response time window are determined, and the excitation time sequence structure of the light source is reconstructed so that the excitation frequency of the light source is higher and the emission duration is longer during periods with higher pollutant concentration, thereby improving the temporal resolution and energy input density at critical moments of pollution development. After completing the time scheduling, based on the distribution characteristics of the pollution interference band and the intrinsic absorption position band of the pollutant, the frequency structure of the light source output is adjusted, the transmission energy of the interference band is reduced, and the transmission power of the absorption band is increased and the bandwidth is reduced, so as to achieve interference suppression and absorption enhancement. After constructing the pollution-adaptive time-frequency light source structure, a dual-mirror time structure is introduced to reverse the pollution intensity curve on the time axis to generate a symmetrical sequence, and simultaneously trigger conjugate beams of opposite phase to form a time-phase conjugate coupling field to eliminate interference shifts caused by microscale optical path differences. Phase conjugate spectral scanning is initiated along the edge of the interference band, triggering forward and conjugate reverse beams and recording the phase synthesis results. Intensity restoration is performed on the spectral curve of the pollution response section to restore the true absorption spectral characteristics.

9. A rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms according to claim 1, characterized in that, Step S006 includes: After completing the suppression of spectral distortion and the restoration of absorption peaks, the spectral stable period after conjugate scanning correction is selected as the activation window for the inverted light field. A light field time axis with symmetrical time markers is constructed to drive the reverse illumination path and form a reversible interference buffer layer, thereby realizing the time-backward intervention of the historical interference phase trajectory. While the inverted light field is established, a microstructure pattern with the opposite phase shift direction to that in the pollutant interference band is injected into the incident light path through a nanoscale optical interference device. Based on the phase difference quantization data of each main band in the interference enhancement region, an inverse phase shift is introduced at a specific spatial position, and the target incident region is activated through a spatial gating method. Based on pattern injection, microbubbles are released directionally using a high-density micro-nozzle array to dynamically disturb the oil film thickness distribution on the water surface. This causes the bubbles to disrupt the periodic stable structure of the oil film interference path during their ascent, and the disturbance rhythm is guided by the pollution concentration evolution spectrum. By coordinating the time-reversed light field, inverse phase pattern, and microbubble perturbation structure, a dynamic detection threshold barrier based on changes in pollution signal intensity is constructed within the pollution response stability window, thereby achieving rapid suppression of pollution interference signals and closed-loop control of the detection process.

10. A rapid detection method for water quality at river discharge outlets based on spectroscopy and intelligent algorithms, as described in claim 9, is characterized in that... Microbubble disturbance controls the release frequency by setting the rate of change of contaminant concentration. When the contaminant concentration increases, the disturbance density is increased, and when the contaminant concentration decreases, the disturbance intensity is reduced. It is also achieved through asymmetric refraction offset formed by the contact of bubbles with the oil film.

Images