A method for diagnosing sewer problems based on residual spectra

Abstract

The application discloses a kind of residual spectrum-based sewer problem diagnosis method, specifically related to sewer problem diagnosis field, including to sunny day collection to be checked pipe two ends adjacent inspection well water sample in pipe, after pretreatment, carry out three-dimensional fluorescence instrument detection and obtain corresponding EMM matrix, obtain upstream inspection well EEM matrix m i And downstream inspection well EEM matrix m i+1 ; after deducting first and second Raman scattering and Rayleigh scattering signal to EEM matrix, downstream inspection well EEM matrix m i+1 Subtract upstream inspection well EEM matrix m i After, obtain the residual spectrum of the pipe segment to be checked;According to the residual fluorescence characteristic peak situation of residual spectrum, determine the problem type that exists in the pipe segment i is checked out.Judgment is distinguished with spectral library comparison by constructing the residual spectrum of upstream and downstream inspection well three-dimensional fluorescence matrix difference and combining spectral feature threshold value, realize the rapid identification and tracing of pipe segment mixed connection pollution and low pollution water infiltration.

Description

Technical Field

[0001] This invention relates to the field of drainage pipe problem diagnosis technology, and more specifically, to a drainage pipe problem diagnosis method based on residual spectroscopy. Background Technology

[0002] In response to the current problems of direct discharge of urban domestic sewage and generally low efficiency of collection and treatment facilities, in 2023, five departments, including the Ministry of Housing and Urban-Rural Development, issued a document requiring the acceleration of the renovation and upgrading of drainage pipe networks to enhance the safety, resilience, and operational efficiency of urban infrastructure. Currently, my country's urban drainage pipe networks suffer from numerous problems such as mixed rainwater and sewage connections and structural defects, leading to sewage mixing and infiltration of low-pollution water, which seriously affects the operational efficiency of urban drainage pipe networks. At the same time, due to early construction and chaotic management of the pipe networks, there are problems such as unclear baseline data and difficulties in sorting out the network. Therefore, it is urgent to develop efficient and accurate drainage pipe problem investigation technologies to provide targeted support for drainage pipe quality improvement and efficiency enhancement projects.

[0003] The main purpose of investigating drainage pipeline problems is to identify the external water (mixed sewage, backflow of seawater, infiltrated groundwater or sewage) in each pipe section, thereby clarifying the main targets of subsequent pipeline renovation. However, the current mainstream geophysical exploration methods rely on visual imaging equipment such as CCTV and QV. Before conducting the investigation, it is necessary to shut off the water supply to the pipeline and dredge the silt, which consumes a lot of manpower and time, and it is difficult to identify the type and source of external water in the pipeline. The traditional water quality characteristic factor method relies on changes in water quality characteristic factors such as COD, NH3-N, and TN, which is also difficult to accurately trace the source of external water in the pipeline.

[0004] CN116858817 discloses a diagnostic method for mixed industrial wastewater connections based on fluorescence spectroscopy. This method uses the residual spectra of adjacent manholes in the pipe section to diagnose mixed industrial wastewater connections. However, this method does not consider the inflow and infiltration of low-pollution water in the pipe section, and it is difficult to make corresponding problem diagnoses for different types of residual spectra. CN119784746 discloses a diagnostic method for the inflow and infiltration of low-pollution water in sewage pipes based on fluorescence spectral images. This method is mainly based on a background pollution source spectrum library and machine learning model, but it is difficult to support the diagnosis of mixed pollution points and pollution source tracing in rainwater pipes. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a drainage pipeline problem diagnosis method based on residual spectroscopy. By constructing a residual spectrum from the difference of three-dimensional fluorescence matrices of upstream and downstream inspection wells and combining it with spectral feature threshold discrimination and spectral library comparison, the method can quickly identify and trace the source of mixed pollution in pipe sections and low-pollution water infiltration, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for diagnosing drainage pipe problems based on residual spectroscopy, comprising:

[0007] S1. Collect water samples from adjacent manholes at both ends of the pipeline to be investigated on a sunny day. After preprocessing, conduct three-dimensional fluorescence detection and obtain the corresponding EMM matrix. Obtain the upstream manhole EEM matrix mi and the downstream manhole EEM matrix mi+1.

[0008] S2. After subtracting the first and second-order Raman and Rayleigh scattering signals from the EEM matrix, subtract the upstream inspection well's EEM matrix mi+1 from the downstream inspection well's EEM matrix to obtain the residual spectrum of the pipe section to be investigated. ;

[0009] S3, Based on residual spectrum The residual fluorescence characteristic peaks are used to determine the type of problem existing in the investigated pipe section i.

[0010] S4. Based on the diagnosed problem type of the pipeline section, conduct source tracing of corresponding mixed pollution or low-pollution water through three-dimensional fluorescence spectroscopy.

[0011] In a preferred embodiment, the steps of well water sample collection, pretreatment, and testing described in S1 include:

[0012] S1-1. Well water sampling includes collecting the overlying water in the well, which includes water samples 5-10 cm below the surface. During sampling, the sediment at the bottom of the well should not be disturbed to avoid affecting subsequent three-dimensional fluorescence monitoring.

[0013] S1-2. To avoid fluctuations in water quality, sampling should be conducted during peak water usage periods. This includes collecting three batches of water samples (10-50 mL each) from the same inspection well within 10 minutes. The water samples are then mixed in equal volumes to ensure representativeness.

[0014] S1-3, Pretreatment includes filtering solid impurities from water samples through a 0.45μm filter membrane;

[0015] S1-4. For the filtered water sample, a three-dimensional fluorescence spectrometer is used for detection, with an excitation wavelength range of 200-450nm and an emission wavelength range of 250-500nm.

[0016] In a preferred embodiment, the steps of subtracting the first- and second-order Raman and Rayleigh scattering signals in S2 include:

[0017] S2-1. Use a UV-spectrophotometer to perform a full scan of absorbance on the filtered water sample, measuring the wavelength range of 200-550nm, and obtain its absorbance data.

[0018] S2-2. Based on the acquired absorbance data, the EEM matrix is ​​corrected for fluorescence internal filtering effect, and its first and second-order Raman scattering and Rayleigh scattering signals are subtracted.

[0019] S2-3. Based on matrix operations, the fluorescence matrix data after subtracting the scattering signal is processed. The downstream inspection well EEM matrix mi+1 is subtracted from the upstream inspection well EEM matrix mi to obtain the residual spectrum. .

[0020] In a preferred embodiment, the identification of the residual fluorescence characteristic peaks and the types of problems in S3 includes:

[0021] S3-1, If ​​the residual spectrum The residual fluorescence peak is below the preset threshold, which means the residual spectrum... The absence of obvious residual fluorescence peaks indicates that there is no low-pollution water inflow or infiltration in this pipe section, and there is no mixed sewage input into the downstream inspection well;

[0022] S3-2, If the residual spectrum If a large area of ​​negative residual fluorescence peaks appears in the spectrum, it is determined that there is low-pollution water infiltration in the pipe section. Specifically, when there are at least N wavelength points in the residual spectrum with fluorescence intensity lower than a preset negative threshold, and these wavelength points are located in a continuous range with a maximum wavelength span not exceeding L nm, the region is determined to be a large area of ​​negative residual fluorescence peaks. Here, N represents the lower limit of the number of negative points (including ≥10); L represents the allowable wavelength span range (including ≤50 nm), which is used to indicate the continuity and local aggregation characteristics of negative wavelength points in the spectrum.

[0023] S3-3, If the residual spectrum If a residual fluorescence peak appears in the downstream inspection well that is different from the one in the upstream inspection well EEM matrix mi, it is determined that there is one or more new types of mixed pollution input in the downstream inspection well of the pipe section.

[0024] S3-4, If the residual spectrum If residual fluorescence peaks at the same position appear in the EEM matrix mi of the upstream inspection well, it is determined that there is a mixed pollution input of the same type in the downstream inspection well of the pipe section.

[0025] In a preferred embodiment, the step of tracing the source of mixed contaminated or low-pollution water in S4 includes:

[0026] S4-1. If the judgment result of the problem type is S3-2, then the residual spectrum is compared with the low-pollution water spectral library to clarify the type of low-pollution water.

[0027] S4-2. If the specific problem type is the judgment result of S3-3 or S3-4, then the residual spectrum is compared with the pollution source spectrum library to clarify the pollution source type.

[0028] S4-3, The source tracing method includes prediction based on machine learning models or cosine similarity calculation, and the similarity calculation formula includes:

[0029] The source tracing method includes prediction based on machine learning models or cosine similarity calculation, and the similarity calculation formula includes:

[0030]

[0031] in This represents the spectral vector of a known pollution source in the spectral library; Indicates the section of pipe to be diagnosed The residual spectral vector; This indicates that a known pollution source in the spectral library is in the 1st... Fluorescence intensity values ​​at each wavelength point; Indicates the current pipe segment to be identified. The residual spectrum in the first Fluorescence intensity values ​​at each wavelength point; This represents the dot product of two spectral vectors; Represents the magnitude of the spectral vectors in the spectral library; Represents the magnitude of the residual spectral vector; This represents the total number of spectral vectors for a known pollution source; the similarity range is... The closer the value is to 1, the more similar the spectra are.

[0032] In a preferred embodiment, during the source tracing process of residual spectra in step S4, a hierarchical discrimination mechanism for spectral matching is constructed, and a spectral variation response strategy is introduced to enhance the robustness and distinguishability of pollution source type determination. The hierarchical discrimination mechanism includes a first discrimination layer, a second discrimination layer, and a third discrimination layer.

[0033] The first discrimination layer: Based on the cosine similarity calculation results between the residual spectral vector and the spectral vectors of each pollution source in the spectral library, the spectral vectors with similarity greater than the preset initial screening threshold are selected to form a candidate set of pollution sources;

[0034] The second discriminant layer: For each spectrum in the candidate pollution source set, calculate the Euclidean distance and Pearson correlation coefficient between it and the residual spectral vector, and construct a dual-scoring channel with these two as adversarial factors, wherein:

[0035] The Euclidean distance value is used to quantify the overall deviation of the spectrum at the numerical level.

[0036] The Pearson correlation coefficient value is used to characterize the consistency of the trend of spectral changes;

[0037] The two scoring channels together form the adversarial scoring vector of the pollution source spectrum;

[0038] The third discrimination layer: Based on the adversarial scoring vector corresponding to each pollution source spectrum, calculate its spectrum variability index. The spectrum variability is defined as the maximum absolute difference between the normalized values ​​of each scoring channel. The maximum absolute difference is used to measure the stability and adaptability fluctuation of the spectrum under the multi-channel discrimination dimension.

[0039] Final judgment logic: In the candidate set of pollution sources, the pollution source type with a spectral variability index lower than the preset steady-state threshold is selected as the final judgment result; if the variability of all candidate spectra exceeds the preset steady-state threshold, the output is no stable matching source, and it is indicated that the current spectral library does not have sufficient coverage and training samples need to be added or the pollution source types need to be expanded.

[0040] The technical effects and advantages of this invention are as follows:

[0041] This invention addresses the difficulty in accurately identifying mixed stormwater and sewage connections and low-pollution water infiltration in current urban drainage networks. It proposes a discrimination method based on residual spectroscopy, which constructs a residual spectrum by using the difference between the EEM matrices of upstream and downstream inspection wells to achieve rapid identification of external water types and sources, providing targeted support for network quality improvement and renovation.

[0042] This method performs first- and second-order Raman scattering and Rayleigh scattering signal subtraction on the detection data before constructing the residual spectrum, and combines it with internal filtering effect correction, which can significantly eliminate background interference, improve the sensitivity and stability of three-dimensional fluorescence detection, and thus obtain highly reliable residual spectral features for subsequent judgment.

[0043] This invention establishes a quantitative identification criterion by introducing a preset threshold and continuous interval discrimination conditions into the residual spectrum. This enables stable identification of low-pollution water inflow and infiltration, overcoming the limitation of traditional water quality characteristic factor detection being insensitive to low concentrations and slow-release pollutants, and improving the completeness and adaptability of the diagnosis.

[0044] After identifying the problem in the pipe section, this method further combines the comparison of residual spectral data with the pollution source spectral library or low-pollution water spectral library. With the help of similarity calculation or model prediction, it can achieve accurate classification of pollution types and determination of sources, expanding the diagnostic results from a single "problem exists" to "clear type and source".

[0045] The residual spectral diagnosis of the present invention does not require complex pre-processing steps such as water cut-off and dredging, nor does it rely on single-factor fluctuations such as COD or ammonia nitrogen for judgment. Instead, it quickly completes the comparison through spectral differences, thereby significantly reducing the input of manpower and equipment, shortening the investigation cycle, and improving the efficiency and feasibility of pipeline network diagnosis.

[0046] This method is compatible with different pipeline operation scenarios. It can be applied to the identification of mixed connections in sewage pipelines as well as the infiltration detection of rainwater pipelines. It has cross-scenario application capabilities and has built a stable and scalable drainage pipeline diagnosis mode, laying a technical foundation for the intelligent operation and maintenance of urban pipeline networks. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the sewage leakage diagnosis method for sewage pipes based on clean water introduction, provided in an embodiment of the present invention. Detailed Implementation

[0048] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Refer to the instruction manual appendix Figure 1 An embodiment of the present invention provides a method for diagnosing drainage pipe problems based on residual spectroscopy, comprising:

[0050] S1. Collect water samples from adjacent manholes at both ends of the pipeline to be investigated on a sunny day. After preprocessing, conduct three-dimensional fluorescence detection and obtain the corresponding EMM matrix. Obtain the upstream manhole EEM matrix mi and the downstream manhole EEM matrix mi+1.

[0051] S2. After subtracting the first and second-order Raman and Rayleigh scattering signals from the EEM matrix, subtract the upstream inspection well's EEM matrix mi+1 from the downstream inspection well's EEM matrix to obtain the residual spectrum of the pipe section to be investigated. ;

[0052] S3, Based on residual spectrum The residual fluorescence characteristic peaks are used to determine the type of problem existing in the investigated pipe section i.

[0053] S4. Based on the diagnosed problem type of the pipeline section, conduct source tracing of corresponding mixed pollution or low-pollution water through three-dimensional fluorescence spectroscopy.

[0054] The steps for sampling, pretreatment, and testing water samples from inspection wells as described in S1 include:

[0055] S1-1. Well water sampling includes collecting the overlying water in the well, which includes water samples 5-10 cm below the surface. During sampling, the sediment at the bottom of the well should not be disturbed to avoid affecting subsequent three-dimensional fluorescence monitoring.

[0056] S1-2. To avoid fluctuations in water quality, sampling should be conducted during peak water usage periods. This includes collecting three batches of water samples (10-50 mL each) from the same inspection well within 10 minutes. The water samples are then mixed in equal volumes to ensure representativeness.

[0057] S1-3, Pretreatment includes filtering solid impurities from water samples through a 0.45μm filter membrane;

[0058] S1-4. For the filtered water sample, a three-dimensional fluorescence spectrometer is used for detection, with an excitation wavelength range of 200-450nm and an emission wavelength range of 250-500nm.

[0059] The steps for subtracting first- and second-order Raman and Rayleigh scattering signals described in S2 include:

[0060] S2-1. Use a UV-spectrophotometer to perform a full scan of absorbance on the filtered water sample, measuring the wavelength range of 200-550nm, and obtain its absorbance data.

[0061] S2-2. Based on the acquired absorbance data, the EEM matrix is ​​corrected for fluorescence internal filtering effect, and its first and second-order Raman scattering and Rayleigh scattering signals are subtracted.

[0062] S2-3. Based on matrix operations, the fluorescence matrix data after subtracting the scattering signal is processed. The downstream inspection well EEM matrix mi+1 is subtracted from the upstream inspection well EEM matrix mi to obtain the residual spectrum. .

[0063] The identification of residual fluorescence characteristic peaks and the types of problems described in S3 includes:

[0064] S3-1, If ​​the residual spectrum The residual fluorescence peak is below the preset threshold, which means the residual spectrum... The absence of obvious residual fluorescence peaks indicates that there is no low-pollution water inflow or infiltration in this pipe section, and there is no mixed sewage input into the downstream inspection well;

[0065] S3-2, If the residual spectrum If a large area of ​​negative residual fluorescence peaks appears in the spectrum, it is determined that there is low-pollution water infiltration in the pipe section. Specifically, when there are at least N wavelength points in the residual spectrum with fluorescence intensity lower than a preset negative threshold, and these wavelength points are located in a continuous range with a maximum wavelength span not exceeding L nm, the region is determined to be a large area of ​​negative residual fluorescence peaks. Here, N represents the lower limit of the number of negative points (including ≥10); L represents the allowable wavelength span range (including ≤50 nm), which is used to indicate the continuity and local aggregation characteristics of negative wavelength points in the spectrum.

[0066] S3-3, If the residual spectrum If a residual fluorescence peak appears in the downstream inspection well that is different from the one in the upstream inspection well EEM matrix mi, it is determined that there is one or more new types of mixed pollution input in the downstream inspection well of the pipe section.

[0067] S3-4, If the residual spectrum If residual fluorescence peaks at the same position appear in the EEM matrix mi of the upstream inspection well, it is determined that there is a mixed pollution input of the same type in the downstream inspection well of the pipe section.

[0068] The steps for tracing the source of mixed contaminated or low-pollution water described in S4 include:

[0069] S4-1. If the judgment result of the problem type is S3-2, then the residual spectrum is compared with the low-pollution water spectral library to clarify the type of low-pollution water.

[0070] S4-2. If the specific problem type is the judgment result of S3-3 or S3-4, then the residual spectrum is compared with the pollution source spectrum library to clarify the pollution source type.

[0071] S4-3, The source tracing method includes prediction based on machine learning models or cosine similarity calculation, and the similarity calculation formula includes:

[0072]

[0073] in This represents the spectral vector of a known pollution source in the spectral library; Indicates the section of pipe to be diagnosed The residual spectral vector; This indicates that a known pollution source in the spectral library is in the 1st... Fluorescence intensity values ​​at each wavelength point; Indicates the current pipe segment to be identified. The residual spectrum in the first Fluorescence intensity values ​​at each wavelength point; This represents the dot product of two spectral vectors; Represents the magnitude of the spectral vectors in the spectral library; Represents the magnitude of the residual spectral vector; This represents the total number of spectral vectors for a known pollution source; the similarity range is... The closer the value is to 1, the more similar the spectra are.

[0074] In the process of tracing the residual spectrum in step S4, a hierarchical discrimination mechanism for spectral matching is constructed, and a spectral variation response strategy is introduced to enhance the robustness and distinguishability of pollution source type determination. The hierarchical discrimination mechanism includes a first discrimination layer, a second discrimination layer and a third discrimination layer.

[0075] The first discrimination layer: Based on the cosine similarity calculation results between the residual spectral vector and the spectral vectors of each pollution source in the spectral library, the spectral vectors with similarity greater than the preset initial screening threshold are selected to form a candidate set of pollution sources;

[0076] The second discriminant layer: For each spectrum in the candidate pollution source set, the Euclidean distance and Pearson correlation coefficient between it and the residual spectral vector are calculated, and a dual-scoring channel with these two as adversarial factors is constructed, wherein:

[0077] The Euclidean distance value is used to quantify the overall deviation of the spectrum at the numerical level.

[0078] The Pearson correlation coefficient value is used to characterize the consistency of the trend of spectral changes;

[0079] The two scoring channels together form the adversarial scoring vector of the pollution source spectrum;

[0080] The third discrimination layer: Based on the adversarial scoring vector corresponding to each pollution source spectrum, calculate its spectrum variability index. The spectrum variability is defined as the maximum absolute difference between the normalized values ​​of each scoring channel. The maximum absolute difference is used to measure the stability and adaptability fluctuation of the spectrum under the multi-channel discrimination dimension.

[0081] Final judgment logic: In the candidate set of pollution sources, the pollution source type with a spectral variability index lower than the preset steady-state threshold is selected as the final judgment result; if the variability of all candidate spectra exceeds the preset steady-state threshold, the output is no stable matching source, and it is indicated that the current spectral library does not have sufficient coverage and training samples need to be added or the pollution source types need to be expanded.

[0082] In the process of tracing the source of residual spectra, it is important to note that this scheme first constructs a hierarchical discrimination mechanism to gradually enhance the accuracy and anti-interference capability of pollution source identification. The first discrimination layer is based on the cosine similarity relationship between the residual spectrum and the spectra of each pollution source in the spectral library. This similarity value is mainly used to measure the degree of similarity between the residual spectrum and the existing standard spectrum in terms of directionality (i.e., the trend of peaks and troughs in the wavelength sequence). In this stage, by constructing all spectra and residual spectra into vector structures of the same dimension and calculating the cosine value of the angle between their vectors, the system can screen out pollution source spectra with a similarity higher than the preset initial screening threshold (e.g., 0.85) to form a pollution source candidate set, which serves as the input for subsequent discrimination steps.

[0083] Further clarification is needed in the candidate spectrum screening process. This scheme introduces an "adversarial scoring mechanism," constructing two scoring channels: numerical amplitude and trend consistency. These are uniformly organized into an adversarial scoring vector to achieve more detailed spectral comparison and improved robustness. The Euclidean distance channel measures the difference in intensity amplitude between the residual spectrum and each candidate spectrum. Specifically, it sums the squares of the intensity differences between the two spectra at each wavelength and takes the square root; the result reflects the overall degree of numerical deviation. Simultaneously, the Pearson correlation channel assesses the consistency of the two spectra in peak and trough trends. The correlation coefficient is calculated using covariance, mean, and standard deviation to quantify the level of trend consistency. The results from these two scoring channels are normalized to form the adversarial scoring vector for each spectrum, providing multi-dimensional data support for subsequent variation assessment.

[0084] Regarding the identification of whether candidate spectra are stable and well-fitted, it should be noted that this scheme constructs a "spectral variability index" as a key basis for determining the consistency of multi-channel scoring, thereby further refining the selection criteria for candidate spectra. The spectral variability index is defined as the maximum absolute difference between the normalized values ​​of each channel in the adversarial scoring vector of the spectrum, that is, it measures the degree of fluctuation of the discrimination results of the same spectrum under different dimensions. If a spectrum has a large score deviation in two channels, it indicates that it may have an unstable or inconsistent response pattern with the residual spectrum in some dimensions, and belongs to a variable spectrum with unstable judgment. Conversely, if the scores of the two channels are consistent and the normalized values ​​are close, the spectrum can be regarded as having good fit.

[0085] It should be noted in the final result decision-making process that: this scheme sets a spectral variability threshold as a steady-state judgment standard, and determines the final pollution source type based on this threshold; if the variability index of a candidate spectrum is lower than the preset threshold (e.g., 0.2), the system recognizes that it has high stability and confidence in the dual-channel scoring, and outputs its corresponding pollution source type as the final judgment result; conversely, if the variability of all spectra in the candidate set is higher than the steady-state threshold, it is determined that the current spectral library does not have matching ability, the system will return a prompt that there is no stable matching source, and suggest that the user supplement training samples or expand the spectral type coverage;

[0086] In summary, the multi-channel scoring and variation index system constructed in this hierarchical adversarial structure not only improves the accuracy and stability of pollution source type determination, but also has interpretability and scalability. It can perform comprehensive dynamic identification of spectral trends, intensity, and adaptation fluctuations, demonstrating the multi-dimensional source tracing and identification capabilities of the spectral data.

[0087] This solution also includes the following embodiments:

[0088] A method for diagnosing drainage pipe problems based on residual spectroscopy includes the following steps:

[0089] S1. On a sunny day, collect water samples from adjacent manholes at both ends of the rainwater pipe to be investigated. After preprocessing, conduct three-dimensional fluorescence detection (EEM) to obtain the upstream manhole EEM matrix mi and the downstream manhole matrix mi+1.

[0090] The specific steps for sample collection and preprocessing are as follows:

[0091] (1) The investigation should be carried out on a sunny day, with no rainfall in the previous week, and the investigation time should be from 7:00 to 9:00 in the morning to ensure that there is sewage discharged into the pipeline;

[0092] (2) Use a stainless steel water sampler to collect water samples 5 cm below the liquid surface in the inspection well, without disturbing the bottom sediment during collection;

[0093] (3) When sampling a single inspection well, repeat the sampling 3 times within 10 minutes, with each sample being 30 mL. Then, mix the 3 samples with equal volumes.

[0094] (4) The mixed water sample was filtered on-site using a syringe and a needle filter;

[0095] (5) After the samples are refrigerated at 0-4℃, they are transported to the laboratory for three-dimensional fluorescence detection;

[0096] S2. After subtracting the first and second-order Raman scattering and Rayleigh scattering signals from the EEM matrix, subtract the upstream inspection well's EEM matrix mi+1 from the downstream inspection well's EEM matrix to obtain the residual spectrum of the pipe section to be investigated. ;

[0097] The specific steps for obtaining the residual spectrum are as follows:

[0098] (1) The absorbance of the filtered water sample was detected by a UV-spectrophotometer in the range of 200-550 nm.

[0099] (2) The fluorescence internal filtering effect was corrected for the EEM matrix using ultraviolet-visible spectroscopic data, and its first- and second-order Raman scattering and Rayleigh scattering signals were subtracted;

[0100] (3) Based on matrix operations, the fluorescence matrix data after subtracting the scattering signal is processed. The downstream inspection well EEM matrix mi+1 is subtracted from the upstream inspection well EEM matrix mi to obtain the residual spectrum. ;

[0101] Table 1 shows the EEM spectra and residual spectra of the upstream and downstream inspection wells corresponding to the investigated pipe section:

[0102]

[0103] S3, Based on residual spectrum Based on the residual fluorescence characteristic peaks, determine the specific type of problem in the investigated pipe section i;

[0104] Based on the residual spectrum type, the diagnostic results of the investigated rainwater pipes are shown in Table 2:

[0105]

[0106] S4. Based on the diagnosis of drainage pipe problems, conduct source tracing of corresponding mixed pollution or low-pollution water based on three-dimensional fluorescence spectroscopy;

[0107] Based on the diagnosis of pipe segment problems, the source of mixed pollution and low-pollution water were traced for pipe segments 2-3 and 4-5 respectively based on similarity calculation. According to the calculation, the residual spectrum of pipe segment 2-3 had the highest similarity with the domestic sewage spectrum, reaching 0.89; the residual spectrum of pipe segment 4-5 had the highest similarity with the groundwater spectrum, reaching 0.82.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for diagnosing drainage pipe problems based on residual spectroscopy, characterized in that, include: S1. Collect water samples from adjacent manholes at both ends of the pipeline to be investigated on a sunny day. After preprocessing, conduct three-dimensional fluorescence detection and obtain the corresponding EMM matrix. Obtain the upstream manhole EEM matrix mi and the downstream manhole EEM matrix mi+1. S2. After subtracting the first and second-order Raman and Rayleigh scattering signals from the EEM matrix, subtract the upstream inspection well's EEM matrix mi+1 from the downstream inspection well's EEM matrix to obtain the residual spectrum of the pipe section to be investigated. ; S3, Based on residual spectrum The residual fluorescence characteristic peaks are used to determine the type of problem existing in the investigated pipe section i. S4. Based on the diagnosed problem type of the pipeline section, conduct source tracing of corresponding mixed pollution or low-pollution water through three-dimensional fluorescence spectroscopy; The identification of residual fluorescence characteristic peaks and the types of problems described in S3 includes: S3-1, If ​​the residual spectrum The residual fluorescence peak is lower than the preset threshold, indicating that there is no low-pollution water inflow or infiltration in this pipe section, and there is no mixed sewage input in the downstream inspection well. S3-2, If the residual spectrum If a negative residual fluorescence peak appears in the sample, it is determined that there is low-pollution water infiltration in the pipe section. S3-3, If the residual spectrum If a residual fluorescence peak appears in the downstream inspection well that is different from the one in the upstream inspection well EEM matrix mi, it is determined that there is one or more new types of mixed pollution input in the downstream inspection well of the pipe section. S3-4, If the residual spectrum If residual fluorescence peaks with the same position appear in the EEM matrix mi of the upstream inspection well, it is determined that there is a mixed pollution input of the same type in the downstream inspection well of the pipe section. The steps for tracing the source of mixed contaminated or low-pollution water described in S4 include: S4-1. If the judgment result of the problem type is S3-2, then the residual spectrum is compared with the low-pollution water spectral library to clarify the type of low-pollution water. S4-2. If the specific problem type is the judgment result of S3-3 or S3-4, then the residual spectrum is compared with the pollution source spectrum library to clarify the pollution source type. S4-3. Source tracing methods include prediction based on machine learning models or cosine similarity calculation. The similarity calculation formula includes: ； in This represents the spectral vector of a known pollution source in the spectral library; Represents the residual spectral vector of the pipe segment i to be diagnosed; This represents the fluorescence intensity value of a known pollution source in the spectral library at the j-th wavelength point; This represents the fluorescence intensity value at the j-th wavelength point of the residual spectrum of the current pipe segment i to be identified; This represents the dot product of two spectral vectors; Represents the magnitude of the spectral vectors in the spectral library; The magnitude of the residual spectral vector is represented by n; n represents the total number of spectral vectors for a known pollution source; the similarity range is... ; In the process of tracing the source of residual spectra in step S4, a hierarchical discrimination mechanism for spectrum matching is constructed, and a spectrum variation response strategy is introduced. The hierarchical discrimination mechanism includes a first discrimination layer, a second discrimination layer, and a third discrimination layer. The first discrimination layer: Based on the cosine similarity calculation results between the residual spectral vector and the spectral vectors of each pollution source in the spectral library, the spectral vectors with similarity greater than the preset initial screening threshold are selected to form a candidate set of pollution sources; The second discriminant layer: For each spectrum in the candidate pollution source set, the Euclidean distance and Pearson correlation coefficient between it and the residual spectral vector are calculated, and a dual-scoring channel with these two as adversarial factors is constructed, wherein: The Euclidean distance value is used to quantify the overall deviation of the spectrum at the numerical level. The Pearson correlation coefficient value is used to characterize the consistency of the trend of spectral changes; The two scoring channels together form the adversarial scoring vector of the pollution source spectrum; The third discrimination layer: Based on the adversarial scoring vector corresponding to each pollution source spectrum, calculate its spectrum variability index. The spectrum variability is defined as the maximum absolute difference between the normalized values ​​of each scoring channel. The maximum absolute difference is used to measure the stability and adaptability fluctuation of the spectrum under the multi-channel discrimination dimension. Final judgment logic: In the candidate set of pollution sources, the pollution source type with a spectral variability index lower than the preset threshold is selected as the final judgment result; if the variability of all candidate spectra exceeds the preset threshold, the output is no stable matching source, and it is indicated that the current spectral library does not have the coverage capability, and training samples need to be added or the pollution source types need to be expanded.

2. The drainage pipe problem diagnosis method based on residual spectroscopy according to claim 1, characterized in that: The steps for well water sample collection, pretreatment, and testing in S1 include: S1-1. Well water sampling includes collecting the overlying water in the well, which includes water samples 5-10 cm below the surface. During sampling, the sediment at the bottom of the well should not be disturbed to avoid affecting subsequent three-dimensional fluorescence monitoring. S1-2. To avoid fluctuations in water quality, sampling should be carried out during peak water usage periods. This includes collecting three batches of water samples from the same inspection well within 10 minutes, with each batch containing 10-50 mL. The water samples should then be mixed in equal volumes. S1-3, Pretreatment includes filtering solid impurities from water samples through a 0.45μm filter membrane; S1-4. For the filtered water sample, a three-dimensional fluorescence spectrometer is used for detection, with an excitation wavelength range of 200-450nm and an emission wavelength range of 250-500nm.

3. The drainage pipe problem diagnosis method based on residual spectroscopy according to claim 2, characterized in that: The steps for subtracting first- and second-order Raman and Rayleigh scattering signals in S2 include: S2-1. Use a UV-spectrophotometer to perform a full scan of absorbance on the filtered water sample, measuring the wavelength range of 200-550nm, and obtain its absorbance data. S2-2. Based on the acquired absorbance data, the EEM matrix is ​​corrected for fluorescence internal filtering effect, and its first and second-order Raman scattering and Rayleigh scattering signals are subtracted. S2-3. Based on matrix operations, the fluorescence matrix data after subtracting the scattering signal is processed. The downstream inspection well EEM matrix mi+1 is subtracted from the upstream inspection well EEM matrix mi to obtain the residual spectrum. .

Images