Diagnostic Method for Inflow and Infiltration of Low-Pollution Water in Sewage Pipelines Based on Fluorescence Spectrum Images

Abstract

The present invention relates to the technical field of municipal engineering, and provides a method for diagnosing the inflow and infiltration of low-pollution water in sewage pipelines based on fluorescence spectral images, including: collecting different types of sewage and different types of low-pollution water in the sewage pipeline inspection area, mixing the collected different types of sewage and different types of low-pollution water in different proportions according to the mixing situation of sewage and low-pollution water in the sewage pipeline inspection area, using a three-dimensional fluorescence spectrometer to detect the mixed water sample, obtaining the fluorescence spectral image corresponding to the mixed water sample, training a water sample type diagnosis model and a water sample type proportion diagnosis model based on the fluorescence spectral image corresponding to the mixed water sample, the sewage type and proportion, and the low-pollution water type and proportion, and diagnosing the inflow and infiltration of low-pollution water in the sewage pipeline to be diagnosed based on the above models. In this way, the diagnosis of the inflow and infiltration of low-pollution water when multiple types of sewage coexist in the sewage pipeline can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of municipal engineering, and particularly to a method for diagnosing the inflow and infiltration of low-pollution water in sewage pipes based on fluorescence spectral images. Background Art

[0002] During the long-term operation of urban sewage pipe systems, they are often affected by the inflow and infiltration of low-pollution water (such as groundwater, mountain spring water, seawater, river water, etc.). After this low-pollution water enters the sewage pipes, it will cause a decrease in the organic matter concentration of the influent to the sewage treatment plant, affect the biochemical treatment efficiency, and may increase the operating cost of the sewage treatment plant. In addition, overloaded operation of the pipes may cause overflow pollution and affect the safety of the water environment. Therefore, accurately diagnosing the problem of inflow and infiltration of low-pollution water in sewage pipes is of great significance for ensuring the normal operation of urban sewage systems.

[0003] Currently, the diagnostic methods for the inflow and infiltration of low-pollution water in sewage pipes mainly include flow monitoring, water quality analysis, tracer detection, etc. Although these methods can reflect the pipeline leakage situation to a certain extent, they often have limitations such as long monitoring periods, low spatial resolution, and high costs. For example, traditional water quality analysis methods mainly rely on indicators such as chemical oxygen demand and total organic carbon, and it is difficult to accurately distinguish the sources of sewage and low-pollution water with these indicators. Therefore, there is an urgent need to develop more efficient, rapid, and economical diagnostic means to improve the diagnostic efficiency and accuracy. Summary of the Invention

[0004] In a first aspect, an embodiment of the present invention provides a method for diagnosing the inflow and infiltration of low-pollution water in sewage pipes based on fluorescence spectral images, the method comprising:

[0005] Collect different types of sewage and different types of low-pollution water within the sewage pipe inspection area;

[0006] According to the mixing situation of sewage and low-pollution water within the sewage pipe inspection area, mix the collected different types of sewage and different types of low-pollution water in different proportions, and use a three-dimensional fluorescence spectrometer to detect the mixed water samples to obtain the fluorescence spectral images corresponding to the mixed water samples;

[0007] Using the fluorescence spectral images corresponding to the mixed water samples as sample data, and the sewage types and low-pollution water types corresponding to the mixed water samples as sample labels, construct a water sample type diagnostic training data set, and use the water sample type diagnostic training data set for model training to obtain a water sample type diagnostic model;

[0008] According to the sewage type and low-pollution water type corresponding to the mixed water sample, the mixed water samples are grouped. For any one group, using the fluorescence spectral image corresponding to the mixed water sample in the group as sample data, and the sewage type ratio and low-pollution water type ratio corresponding to the mixed water sample in the group as sample labels, a diagnostic training data set for water sample type ratio is constructed. The diagnostic training data set for water sample type ratio is used for model training to obtain a diagnostic model for water sample type ratio corresponding to the group.

[0009] For the sewage pipeline to be diagnosed, during the peak water use period on a sunny day, each inspection well is opened one by one to collect the water sample in the inspection well, and a three-dimensional fluorescence spectrometer is used to detect the water sample in the inspection well to obtain the fluorescence spectral image corresponding to the water sample in the inspection well. The fluorescence spectral image corresponding to the water sample in the inspection well is input into the water sample type diagnostic model for water sample type diagnosis. According to the diagnosed sewage type and low-pollution water type of the water sample in the inspection well, the water sample in the inspection well is grouped. The fluorescence spectral image corresponding to the water sample in the inspection well is input into the diagnostic model for water sample type ratio corresponding to the group for water sample type ratio diagnosis. According to the diagnosed sewage type ratio and low-pollution water type ratio of the water sample in the inspection well, the inflow and infiltration diagnosis of low-pollution water into the sewage pipeline to be diagnosed is carried out.

[0010] In some realizable ways of the first aspect, different types of sewage are determined according to the types of water users in the sewage pipeline inspection area, including domestic sewage, catering wastewater, commercial wastewater, and / or papermaking wastewater; different types of low-pollution water are determined according to the geographical features in the sewage pipeline inspection area, including groundwater, mountain spring water, seawater, and / or river water.

[0011] In some realizable ways of the first aspect, the mixing situation of sewage and low-pollution water in the sewage pipeline inspection area is determined according to the local natural environmental conditions and the functional type of the drainage area; specifically, when the groundwater level in the sewage pipeline inspection area is higher than the buried depth of the sewage pipeline, different types of sewage in the sewage pipeline inspection area are mixed with groundwater; when there is mountain spring water in the sewage pipeline inspection area, different types of sewage in the sewage pipeline inspection area are mixed with mountain spring water; when the sewage pipeline inspection area is a residential building, domestic sewage is mixed with different types of low-pollution water; when there are catering and commercial businesses in the sewage pipeline inspection area at the same time, catering wastewater and commercial wastewater are mixed with different types of low-pollution water.

[0012] In some realizable ways of the first aspect, when mixing different types of sewage and different types of low-pollution water in different proportions, the proportion of low-pollution water is gradually increased from 0% to 95%, and at least in an increment interval of 5%.

[0013] In some realizable ways of the first aspect, before detecting the mixed water sample using a three-dimensional fluorescence spectrometer, the mixed water sample is filtered using a filter membrane with a pore size of 0.22 - 0.45 μm; when detecting the mixed water sample using a three-dimensional fluorescence spectrometer, a quartz cuvette is used, the voltage of the three-dimensional fluorescence spectrometer is set to 400 V, the excitation wavelength range is 200 - 450 nm, the emission wavelength range is 250 - 550 nm, and the slit width is 2 nm.

[0014] In some realizable ways of the first aspect, a water sample type diagnosis training dataset is used for model training to obtain a water sample type diagnosis model, including:

[0015] Select a machine learning model and, based on this, use the water sample type diagnosis training dataset for model training. When the accuracy of the trained model does not meet the standard, replace the machine learning model or adjust the model hyperparameters until the accuracy meets the standard, and finally obtain a water sample type diagnosis model.

[0016] In some realizable ways of the first aspect, the machine learning models that can be selected when using the water sample type diagnosis training dataset for model training include support vector machine, convolutional neural network, random forest, k-nearest neighbor algorithm, and gradient boosting machine.

[0017] In some realizable ways of the first aspect, a water sample type proportion diagnosis training dataset is used for model training to obtain a water sample type proportion diagnosis model corresponding to the group, including:

[0018] Select a machine learning model and, based on this, use the water sample type proportion diagnosis training dataset for model training. When the accuracy of the trained model does not meet the standard, replace the machine learning model or adjust the model hyperparameters until the accuracy meets the standard, and finally obtain a water sample type proportion diagnosis model corresponding to the group.

[0019] In some realizable ways of the first aspect, the machine learning models that can be selected when using the water sample type proportion diagnosis training dataset for model training include support vector regression, random forest regression, partial least squares regression, and neural network regression.

[0020] In some realizable ways of the first aspect, the sewage pipeline to be diagnosed includes multiple sewage pipe segments to be diagnosed; the fluorescence spectral image corresponding to the water sample in the inspection well is input into the water sample type diagnosis model for water sample type diagnosis. According to the diagnosed sewage type and low-pollution water type of the water sample in the inspection well, the water sample in the inspection well is grouped. The fluorescence spectral image corresponding to the water sample in the inspection well is input into the water sample type proportion diagnosis model corresponding to the group for water sample type proportion diagnosis. According to the diagnosed sewage type proportion and low-pollution water type proportion of the water sample in the inspection well, low-pollution water inflow and infiltration diagnosis of the sewage pipeline to be diagnosed is carried out, including:

[0021] Synchronously input the fluorescence spectral images corresponding to the water samples in adjacent inspection wells of the same sewage pipe section to be diagnosed into the water sample type diagnosis model for water sample type diagnosis. Based on the diagnosed sewage types and low-pollution water types of the water samples in adjacent inspection wells of the same sewage pipe section to be diagnosed, determine the sewage type and low-pollution water type of the receiving water body of the sewage pipe section to be diagnosed, and conduct group classification accordingly. Input the fluorescence spectral images corresponding to the water samples in adjacent inspection wells of the same sewage pipe section to be diagnosed into the water sample type proportion diagnosis model corresponding to the group for water sample type proportion diagnosis, and conduct low-pollution water inflow and infiltration diagnosis on the sewage pipe section to be diagnosed based on the diagnosed sewage type proportion and low-pollution water type proportion of the water samples in adjacent inspection wells of the same sewage pipe section to be diagnosed.

[0022] In a second aspect, an embodiment of the present invention provides a low-pollution water inflow and infiltration diagnosis system for sewage pipes based on fluorescence spectral images. The system includes:

[0023] An acquisition module, configured to acquire different types of sewage and different types of low-pollution water in the sewage pipe inspection area;

[0024] A detection module, configured to mix the acquired different types of sewage and different types of low-pollution water in different proportions according to the mixing situation of sewage and low-pollution water in the sewage pipe inspection area, and use a three-dimensional fluorescence spectrometer to detect the mixed water sample to obtain the fluorescence spectral image corresponding to the mixed water sample;

[0025] A training module, configured to use the fluorescence spectral image corresponding to the mixed water sample as sample data, and the sewage type and low-pollution water type corresponding to the mixed water sample as sample labels to construct a water sample type diagnosis training data set, and use the water sample type diagnosis training data set for model training to obtain a water sample type diagnosis model;

[0026] The training module is further configured to conduct group classification on the mixed water sample according to the sewage type and low-pollution water type corresponding to the mixed water sample. For any one group, use the fluorescence spectral image corresponding to the mixed water sample in it as sample data, and the sewage type proportion and low-pollution water type proportion corresponding to the mixed water sample in it as sample labels to construct a water sample type proportion diagnosis training data set, and use the water sample type proportion diagnosis training data set for model training to obtain a water sample type proportion diagnosis model corresponding to the group;

[0027] The diagnostic module is used to, for the sewage pipeline to be diagnosed, open inspection wells one by one during the peak water consumption period on sunny days, collect water samples in the inspection wells, use a three-dimensional fluorescence spectrometer to detect the water samples in the inspection wells, obtain the fluorescence spectral images corresponding to the water samples in the inspection wells, input the fluorescence spectral images corresponding to the water samples in the inspection wells into the water sample type diagnostic model for water sample type diagnosis, classify the water samples in the inspection wells according to the diagnosed sewage types and low-pollution water types of the water samples in the inspection wells, input the fluorescence spectral images corresponding to the water samples in the inspection wells into the water sample type proportion diagnostic model corresponding to the group for water sample type proportion diagnosis, and perform low-pollution water inflow and infiltration diagnosis on the sewage pipeline to be diagnosed according to the diagnosed sewage type proportion and low-pollution water type proportion of the water samples in the inspection wells.

[0028] Compared with the prior art, the present invention has at least the following technical effects:

[0029] (1) The present invention determines the low-pollution water types and the inflow and infiltration proportions in the sewage pipeline through fluorescence spectra, solving the problems that the current sewage pipeline geophysical exploration technologies (such as QV, CCTV, etc.) have high diagnostic costs and it is difficult to diagnose and trace the low-pollution water.

[0030] (2) The present invention first diagnoses the sewage types and low-pollution water types of the main receiving water bodies in the pipe section through the water sample type diagnostic model, which can select an appropriate water sample type proportion diagnostic model for subsequent water sample type proportion diagnosis, thereby improving the diagnostic efficiency and accuracy.

[0031] (3) The present invention analyzes the proportion of low-pollution water inflow and infiltration in the sewage pipe section through the machine learning model, overcoming the problem that the fluorescence characteristic peak intensity and the dilution ratio are not linearly related under the influence of the inner filter effect and coexisting peak interference.

[0032] It should be understood that the content described in the invention content part is not intended to limit the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Combined with the drawings and referring to the following detailed description, the above and other features, advantages and aspects of the embodiments of the present invention will become more obvious. The drawings are used to better understand the present invention and do not constitute a limitation to the present invention. In the drawings, the same or similar reference numerals represent the same or similar elements, where:

[0034] Figure 1 is a flowchart of a method for diagnosing low-pollution water inflow and infiltration in a sewage pipeline based on fluorescence spectral images provided by an embodiment of the present invention;

[0035] Figure 2 is a ROC graph of the prediction effect of the water sample type diagnostic model provided by an embodiment of the present invention;

[0036] Figure 3 This is the prediction effect diagram of the water sample type ratio diagnosis model provided by the embodiments of the present invention;

[0037] Figure 4 This is the structural diagram of a sewage pipeline low-pollution water inflow and infiltration diagnosis system based on fluorescence spectral images provided by the embodiments of the present invention. Detailed implementation manners

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] In addition, the term "and / or" in the present invention is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in the present invention generally represents an "or" relationship between the front and rear associated objects.

[0040] To solve the technical problems in the background art, the embodiments of the present invention provide a method and system for diagnosing low-pollution water inflow and infiltration in sewage pipelines based on fluorescence spectral images. Different types of sewage and different types of low-pollution water in the sewage pipeline inspection area can be collected. According to the mixing situation of sewage and low-pollution water in the sewage pipeline inspection area, the collected different types of sewage and different types of low-pollution water are mixed in different proportions, and a three-dimensional fluorescence spectrometer is used to detect the mixed water sample to obtain the fluorescence spectral image corresponding to the mixed water sample. According to the fluorescence spectral image corresponding to the mixed water sample, the sewage type and ratio, and the low-pollution water type and ratio, a water sample type diagnosis model and a water sample type ratio diagnosis model are trained, and based on the above models, the low-pollution water inflow and infiltration in the sewage pipeline to be diagnosed is diagnosed. In this way, the diagnosis of low-pollution water inflow and infiltration when multiple types of sewage coexist in the sewage pipeline can be realized, and the diagnosis efficiency and accuracy can be effectively improved.

[0041] The following will, with reference to the accompanying drawings, describe in detail a method and system for diagnosing low-pollution water inflow and infiltration in sewage pipelines based on fluorescence spectral images provided by the embodiments of the present invention through specific embodiments.

[0042] Figure 1 This is the flowchart of a method for diagnosing low-pollution water inflow and infiltration in sewage pipelines based on fluorescence spectral images provided by the embodiments of the present invention, as Figure 1As shown, the low-pollution water inflow and infiltration diagnosis method 100 for sewage pipelines may include:

[0043] S110, collecting different types of sewage and different types of low-pollution water within the sewage pipeline inspection area.

[0044] Among them, different types of sewage are determined according to the types of water users within the sewage pipeline inspection area, including but not limited to domestic sewage, catering wastewater, commercial wastewater, and / or papermaking wastewater, etc.; different types of low-pollution water are determined according to the geographical characteristics within the sewage pipeline inspection area, including but not limited to groundwater, mountain spring water, seawater, and / or river water, etc.

[0045] S120, according to the mixing situation of sewage and low-pollution water within the sewage pipeline inspection area, mixing the collected different types of sewage and different types of low-pollution water in different proportions, and using a three-dimensional fluorescence spectrometer to detect the mixed water sample to obtain the fluorescence spectral image corresponding to the mixed water sample.

[0046] Among them, the mixing situation of sewage and low-pollution water within the sewage pipeline inspection area is determined according to the local natural environmental conditions and the functional types of drainage areas. Specifically, when the groundwater level within the sewage pipeline inspection area is higher than the buried depth of the sewage pipeline, different types of sewage within the sewage pipeline inspection area are mixed with groundwater; when there is mountain spring water within the sewage pipeline inspection area, different types of sewage within the sewage pipeline inspection area are mixed with mountain spring water; when the sewage pipeline inspection area is a residential building, domestic sewage is mixed with different types of low-pollution water; when there are catering and commercial businesses within the sewage pipeline inspection area at the same time, catering wastewater and commercial wastewater are mixed with different types of low-pollution water.

[0047] As an example, when mixing the collected different types of sewage and different types of low-pollution water in different proportions, it is necessary to gradually increase the proportion of low-pollution water from 0% to 95%, and at least increase it at an interval of 5%. Further, to avoid errors caused by unstable light sources and filtration pretreatment of the three-dimensional fluorescence spectrometer, the mixing proportion of low-pollution water shall not be less than 5% of the total volume. Before using the three-dimensional fluorescence spectrometer to detect the mixed water sample, it is necessary to filter the mixed water sample with a filter membrane with a pore size of 0.22 - 0.45 μm. When using the three-dimensional fluorescence spectrometer to detect the mixed water sample, it is necessary to use a quartz cuvette, and set the voltage of the three-dimensional fluorescence spectrometer to 400V, the excitation wavelength range to 200 - 450 nm, the emission wavelength range to 250 - 550 nm, and the slit width to 2 nm.

[0048] S130, using the fluorescence spectral image corresponding to the mixed water sample as sample data, and the sewage type and low-pollution water type corresponding to the mixed water sample as sample labels, constructing a water sample type diagnosis training data set, and using the water sample type diagnosis training data set for model training to obtain a water sample type diagnosis model.

[0049] Specifically, the fluorescence spectral image corresponding to the mixed water sample can be used as the sample data, and the sewage type and low-pollution water type corresponding to the mixed water sample can be used as the sample labels to construct a training data set for diagnosing water sample types. A machine learning model is selected, and based on this, the training data set for diagnosing water sample types is used for model training. When the accuracy of the trained model does not meet the standard, the machine learning model is replaced or the model hyperparameters are adjusted until the accuracy meets the standard, and finally a water sample type diagnosis model is obtained. Optionally, the machine learning models that can be selected during training mainly include support vector machines, convolutional neural networks, random forests, k-nearest neighbor algorithms, and gradient boosting machines.

[0050] S140. According to the sewage type and low-pollution water type corresponding to the mixed water sample, the mixed water samples are grouped. For any one group, the fluorescence spectral image corresponding to the mixed water sample in the group is used as the sample data, and the sewage type ratio and low-pollution water type ratio corresponding to the mixed water sample in the group are used as the sample labels to construct a training data set for diagnosing the water sample type ratio. The training data set for diagnosing the water sample type ratio is used for model training to obtain a water sample type ratio diagnosis model corresponding to the group.

[0051] Specifically, according to the sewage type and low-pollution water type corresponding to the mixed water sample, the mixed water samples can be grouped. For any one group, the fluorescence spectral image corresponding to the mixed water sample in the group is used as the sample data, and the sewage type ratio and low-pollution water type ratio corresponding to the mixed water sample in the group are used as the sample labels to construct a training data set for diagnosing the water sample type ratio. A machine learning model is selected, and based on this, the training data set for diagnosing the water sample type ratio is used for model training. When the accuracy of the trained model does not meet the standard, the machine learning model is replaced or the model hyperparameters are adjusted until the accuracy meets the standard, and finally a water sample type ratio diagnosis model corresponding to the group is obtained. Optionally, the machine learning models that can be selected during training mainly include support vector regression, random forest regression, partial least squares regression, and neural network regression.

[0052] S150. For the sewage pipeline to be diagnosed, during the peak water use period on a sunny day, the inspection wells are opened one by one, and the water samples in the inspection wells are collected. The three-dimensional fluorescence spectrometer is used to detect the water samples in the inspection wells to obtain the fluorescence spectral image corresponding to the water samples in the inspection wells. The fluorescence spectral image corresponding to the water samples in the inspection wells is input into the water sample type diagnosis model for water sample type diagnosis. According to the diagnosed sewage type and low-pollution water type of the water samples in the inspection wells, the water samples in the inspection wells are grouped. The fluorescence spectral image corresponding to the water samples in the inspection wells is input into the water sample type ratio diagnosis model corresponding to the group for water sample type ratio diagnosis. According to the diagnosed sewage type ratio and low-pollution water type ratio of the water samples in the inspection wells, the low-pollution water inflow and infiltration diagnosis of the sewage pipeline to be diagnosed is carried out.

[0053] Among them, the sewage pipeline to be diagnosed includes a plurality of sewage pipe segments to be diagnosed. In view of this, the fluorescence spectral images corresponding to the water samples in adjacent inspection wells of the same sewage pipe segment to be diagnosed can be synchronously input into the water sample type diagnosis model for water sample type diagnosis. According to the diagnosed sewage types and low-pollution water types of the water samples in adjacent inspection wells of the same sewage pipe segment to be diagnosed, the sewage types and low-pollution water types of the main receiving water body of the sewage pipe segment to be diagnosed are determined, and group division is carried out accordingly. The fluorescence spectral images corresponding to the water samples in adjacent inspection wells of the same sewage pipe segment to be diagnosed are input into the water sample type ratio diagnosis model corresponding to the group for water sample type ratio diagnosis. According to the diagnosed sewage type ratio and low-pollution water type ratio of the water samples in adjacent inspection wells of the same sewage pipe segment to be diagnosed, the low-pollution water inflow and infiltration diagnosis of the sewage pipe segment to be diagnosed is carried out, and the diagnosis result is output.

[0054] For the convenience of further understanding, the above-mentioned sewage pipeline low-pollution water inflow and infiltration diagnosis method 100 will be described in detail below with a specific embodiment:

[0055] (1) Collect different types of sewage and different types of low-pollution water in the sewage pipeline inspection area.

[0056] Here, it is assumed that the sewage pipeline inspection area is mainly residential areas and catering business areas, and at the same time, the average elevation of the sewage pipeline is lower than the groundwater and the elevation of the river water along the line.

[0057] Based on the above situation, domestic sewage and catering wastewater in the sewage pipeline inspection area are collected as the main sewage types, and groundwater and river water in the sewage pipeline inspection area are collected as the main low-pollution water types.

[0058] (2) According to the mixing situation of sewage and low-pollution water in the sewage pipeline inspection area, the collected different types of sewage and different types of low-pollution water are mixed in different proportions, and a three-dimensional fluorescence spectrometer is used to detect the mixed water sample to obtain the fluorescence spectral image corresponding to the mixed water sample.

[0059] Referring to Table 1, different groups are set, and the proportion of low-pollution water is gradually increased from 0% to 95% at intervals of 5%. If there are more than 2 types of sewage in the mixed water sample, different sewage proportions should be set according to the actual situation. In this example, the catering wastewater accounts for at most 30% of the mixed water sample. In this embodiment, the proportions (i.e., volume proportions) of different types of sewage and low-pollution water are shown in Table 1:

[0060] Table 1

[0061]

[0062] (3) Using the fluorescence spectral images corresponding to the mixed water samples as sample data and the sewage types and low-pollution water types corresponding to the mixed water samples as sample labels, a diagnostic training dataset for water sample types is constructed. The diagnostic training dataset for water sample types is used for model training to obtain a diagnostic model for water sample types.

[0063] Among them, the ROC curve of the diagnostic model for water sample types is as Figure 2 shown, and its AUC is 0.95, indicating that its diagnostic effect for water sample types is good and the model has high reliability.

[0064] (4) According to the sewage types and low-pollution water types corresponding to the mixed water samples, the mixed water samples are grouped. For any one group, using the fluorescence spectral images corresponding to the mixed water samples in it as sample data and the sewage type ratio and low-pollution water type ratio corresponding to the mixed water samples in it as sample labels, a diagnostic training dataset for water sample type ratios is constructed. The diagnostic training dataset for water sample type ratios is used for model training to obtain a diagnostic model for water sample type ratios corresponding to the group.

[0065] Among them, the prediction effect of the diagnostic model for water sample type ratios is as Figure 3 shown, and its R 2 is 0.91, indicating that its diagnostic result for water sample type ratios is good and the model has high reliability.

[0066] (5) For the sewage pipeline to be diagnosed, during the peak water use period on sunny days, open the inspection wells one by one and collect the water samples in the inspection wells, and use a three-dimensional fluorescence spectrometer to detect the water samples in the inspection wells to obtain the fluorescence spectral images corresponding to the water samples in the inspection wells. Input the fluorescence spectral images corresponding to the water samples in the inspection wells into the diagnostic model for water sample types for water sample type diagnosis. According to the diagnosed sewage types and low-pollution water types of the water samples in the inspection wells, the water samples in the inspection wells are grouped. Input the fluorescence spectral images corresponding to the water samples in the inspection wells into the diagnostic model for water sample type ratios corresponding to the group for water sample type ratio diagnosis. According to the diagnosed sewage type ratio and low-pollution water type ratio of the water samples in the inspection wells, conduct low-pollution water inflow and infiltration diagnosis on the sewage pipeline to be diagnosed.

[0067] Further, pair the fluorescence spectral images corresponding to the water samples in adjacent inspection wells of the same sewage pipeline section to be diagnosed as inputs, and the output results are shown in Table 2:

[0068] Table 2

[0069]

[0070] In Table 2, the upstream and downstream diagnosis results of the upstream and downstream inspection wells are respectively shown for each pipe section.

[0071] Referring to Table 2, for the sewage pipelines to be diagnosed, there is no obvious inflow and infiltration of low-pollution water in pipeline sections 1 to 3; there is groundwater infiltration in pipeline section 4, and the infiltration ratio is 20%, indicating that there is damage in pipeline section 4 and structural repair needs to be carried out; there is river water inflow in pipeline section 5, and the inflow ratio is 10%, and it is necessary to check the intercepting outlets along the bank.

[0072] In summary, the present invention has at least achieved the following technical effects:

[0073] (1) The present invention determines the types of low-pollution water and the inflow and infiltration ratio in the sewage pipeline through fluorescence spectroscopy, solving the problems of high diagnostic costs in current sewage pipeline geophysical exploration technologies (such as QV, CCTV, etc.) and the difficulty in diagnosing and tracing low-pollution water.

[0074] (2) The present invention preferentially diagnoses the types of sewage and low-pollution water of the main receiving water bodies in the pipeline section through the water sample type diagnosis model, and can select an appropriate water sample type ratio diagnosis model for the subsequent water sample type ratio diagnosis, thereby improving the diagnostic efficiency and accuracy.

[0075] (3) The present invention analyzes the inflow and infiltration ratio of low-pollution water in the sewage pipeline section through a machine learning model, overcoming the problem that the intensity of the fluorescence characteristic peak and the dilution ratio are not linearly related under the influence of the inner filter effect and coexisting peak interference.

[0076] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0077] The above is the introduction of the method embodiments. The following further illustrates the solution of the present invention through system embodiments.

[0078] Figure 4 The structural diagram of a sewage pipeline low-pollution water inflow and infiltration diagnosis system based on fluorescence spectral images provided by an embodiment of the present invention is as Figure 4 shown. The sewage pipeline low-pollution water inflow and infiltration diagnosis system 400 may include:

[0079] An acquisition module 410, configured to acquire different types of sewage and different types of low-pollution water in the sewage pipeline inspection area.

[0080] The detection module 420 is configured to mix the collected different types of sewage and different types of low-pollution water in different proportions according to the mixing condition of sewage and low-pollution water in the sewage pipeline inspection area, and use a three-dimensional fluorescence spectrometer to detect the mixed water sample to obtain a fluorescence spectral image corresponding to the mixed water sample.

[0081] The training module 430 is configured to construct a water sample type diagnosis training data set with the fluorescence spectral image corresponding to the mixed water sample as sample data and the sewage type and low-pollution water type corresponding to the mixed water sample as sample labels, and use the water sample type diagnosis training data set to perform model training to obtain a water sample type diagnosis model.

[0082] The training module 430 is further configured to divide the mixed water samples into groups according to the sewage type and low-pollution water type corresponding to the mixed water samples. For any one group, use the fluorescence spectral image corresponding to the mixed water samples in the group as sample data, and the sewage type ratio and low-pollution water type ratio corresponding to the mixed water samples in the group as sample labels to construct a water sample type ratio diagnosis training data set, and use the water sample type ratio diagnosis training data set to perform model training to obtain a water sample type ratio diagnosis model corresponding to the group.

[0083] The diagnosis module 440 is configured to, for the sewage pipeline to be diagnosed, open the inspection wells one by one during the peak water use period on sunny days and collect the water samples in the inspection wells, and use a three-dimensional fluorescence spectrometer to detect the water samples in the inspection wells to obtain a fluorescence spectral image corresponding to the water samples in the inspection wells. Input the fluorescence spectral image corresponding to the water samples in the inspection wells into the water sample type diagnosis model for water sample type diagnosis, divide the water samples in the inspection wells into groups according to the diagnosed sewage type and low-pollution water type of the water samples in the inspection wells, input the fluorescence spectral image corresponding to the water samples in the inspection wells into the water sample type ratio diagnosis model corresponding to the group for water sample type ratio diagnosis, and perform low-pollution water inflow and infiltration diagnosis on the sewage pipeline to be diagnosed according to the diagnosed sewage type ratio and low-pollution water type ratio of the water samples in the inspection wells.

[0084] It can be understood that Figure 4 each module / unit in the sewage pipeline low-pollution water inflow and infiltration diagnosis system 400 shown has the function of implementing Figure 1 each step in the sewage pipeline low-pollution water inflow and infiltration diagnosis method 100 shown, and can achieve its corresponding technical effects. For the sake of brevity, it will not be elaborated here.

[0085] It should be noted that the above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A diagnostic method for the inflow and infiltration of low-pollution water in sewage pipelines based on fluorescence spectral images, characterized in that, The method includes: Collecting different types of sewage and different types of low-pollution water within the sewage pipeline inspection area; According to the mixing situation of sewage and low-pollution water within the sewage pipeline inspection area, mixing the collected different types of sewage and different types of low-pollution water in different proportions, and using a three-dimensional fluorescence spectrometer to detect the mixed water sample to obtain the fluorescence spectral image corresponding to the mixed water sample; Taking the fluorescence spectral image corresponding to the mixed water sample as sample data, and taking the sewage type and low-pollution water type corresponding to the mixed water sample as sample labels, constructing a water sample type diagnosis training data set, and using the water sample type diagnosis training data set to perform model training to obtain a water sample type diagnosis model; According to the sewage type and low-pollution water type corresponding to the mixed water sample, dividing the mixed water sample into groups. For any one group, taking the fluorescence spectral image corresponding to the mixed water sample in it as sample data, and taking the sewage type proportion and low-pollution water type proportion corresponding to the mixed water sample in it as sample labels, constructing a water sample type proportion diagnosis training data set, and using the water sample type proportion diagnosis training data set to perform model training to obtain a water sample type proportion diagnosis model corresponding to the group; For the sewage pipeline to be diagnosed, during the peak water use period on a sunny day, open the inspection wells one by one and collect the water samples in the inspection wells, and use a three-dimensional fluorescence spectrometer to detect the water samples in the inspection wells to obtain the fluorescence spectral image corresponding to the water samples in the inspection wells. Input the fluorescence spectral image corresponding to the water samples in the inspection wells into the water sample type diagnosis model for water sample type diagnosis. According to the diagnosed sewage type and low-pollution water type of the water samples in the inspection wells, divide the water samples in the inspection wells into groups. Input the fluorescence spectral image corresponding to the water samples in the inspection wells into the water sample type proportion diagnosis model corresponding to the group for water sample type proportion diagnosis. According to the diagnosed sewage type proportion and low-pollution water type proportion of the water samples in the inspection wells, perform low-pollution water inflow and infiltration diagnosis on the sewage pipeline to be diagnosed; Specifically, the sewage pipeline to be diagnosed includes multiple sewage pipe segments to be diagnosed; Synchronously input the fluorescence spectral images corresponding to the water samples in the adjacent inspection wells of the same sewage pipe segment to be diagnosed into the water sample type diagnosis model for water sample type diagnosis. According to the diagnosed sewage type and low-pollution water type of the water samples in the adjacent inspection wells of the same sewage pipe segment to be diagnosed, determine the sewage type and low-pollution water type of the receiving water body of the sewage pipe segment to be diagnosed, and perform group division accordingly. Input the fluorescence spectral images corresponding to the water samples in the adjacent inspection wells of the same sewage pipe segment to be diagnosed into the water sample type proportion diagnosis model corresponding to the group for water sample type proportion diagnosis. According to the diagnosed sewage type proportion and low-pollution water type proportion of the water samples in the adjacent inspection wells of the same sewage pipe segment to be diagnosed, perform low-pollution water inflow and infiltration diagnosis on the sewage pipe segment to be diagnosed; The different types of sewage are determined according to the types of water users within the sewage pipeline inspection area, including domestic sewage, catering wastewater, commercial wastewater, and / or papermaking wastewater; The different types of low-pollution water are determined according to the geographical characteristics within the sewage pipeline inspection area, including groundwater, mountain spring water, seawater, and / or river water; The mixing situation of sewage and low-pollution water in the sewage pipeline inspection area is determined according to the local natural environmental conditions and the functional types of drainage areas; specifically, when the groundwater level in the sewage pipeline inspection area is higher than the buried depth of the sewage pipeline, different types of sewage in the sewage pipeline inspection area are mixed with groundwater; when there are mountain springs in the sewage pipeline inspection area, different types of sewage in the sewage pipeline inspection area are mixed with mountain springs; when the sewage pipeline inspection area is a residential building, domestic sewage is mixed with different types of low-pollution water; when there are restaurants and commercial merchants in the sewage pipeline inspection area at the same time, catering wastewater and commercial wastewater are mixed with different types of low-pollution water.

2. The method according to claim 1, characterized in that When mixing different types of sewage and different types of low-pollution water collected in different proportions, the proportion of low-pollution water is gradually increased from 0% to 95%, and the increase interval is at least 5%.

3. The method according to claim 1, wherein Before using a three-dimensional fluorescence spectrometer to detect the mixed water sample, filter the mixed water sample with a filter membrane with a pore size of 0.22 - 0.45 μm; when using a three-dimensional fluorescence spectrometer to detect the mixed water sample, use a quartz cuvette, set the voltage of the three-dimensional fluorescence spectrometer to 400V, the excitation wavelength range to 200 - 450 nm, the emission wavelength range to 250 - 550 nm, and the slit width to 2 nm.

4. The method according to claim 1, characterized in that The model training using the water sample type diagnosis training data set to obtain a water sample type diagnosis model includes: Select a machine learning model and, based on this, use the water sample type diagnosis training data set for model training. When the accuracy of the trained model does not meet the standard, replace the machine learning model or adjust the model hyperparameters until the accuracy meets the standard, and finally obtain a water sample type diagnosis model.

5. The method according to claim 4, wherein The machine learning models that can be selected when using the water sample type diagnosis training data set for model training include support vector machine, convolutional neural network, random forest, k-nearest neighbor algorithm, and gradient boosting machine.

6. The method according to claim 1, wherein The model training using the water sample type proportion diagnosis training data set to obtain a water sample type proportion diagnosis model corresponding to the group includes: Select a machine learning model and, based on this, use the water sample type proportion diagnosis training data set for model training. When the accuracy of the trained model does not meet the standard, replace the machine learning model or adjust the model hyperparameters until the accuracy meets the standard, and finally obtain a water sample type proportion diagnosis model corresponding to the group.

7. The method according to claim 6, characterized in that, The machine learning models that can be selected when using the water sample type proportion diagnosis training data set for model training include support vector regression, random forest regression, partial least squares regression, and neural network regression.

Images