Sewer infiltration defect simulation device, model training method and prediction method

Abstract

The present application relates to sewage pipeline infiltration detection technical field, disclose a sewage pipeline infiltration defect simulation device, model training method and prediction method, the present application through the linkage of camera monitoring system and infiltration flow monitoring system, realize the space-time synchronous collection of pipeline defect place infiltration form image and infiltration flow data, solved the problem that the prior art cannot acquire the infiltration form and the infiltration flow data simultaneously under the laboratory condition;On this basis, the training data is generated by using the synchronous collected data and the deep learning model is trained, the prediction model which can predict the pipeline infiltration form category and the projection area of the outburst according to the infiltration flow and other infiltration related parameters is obtained, thereby the quantitative mapping relationship from the measurable parameter to the infiltration defect is established, and the technical blank that the pipeline infiltration defect cannot be non-excavation quantitative evaluation in the prior art is filled.

Description

Technical Field

[0001] This invention relates to the field of sewage pipeline infiltration detection technology, specifically to a sewage pipeline infiltration defect simulation device, model training method, and prediction method. Background Technology

[0002] Urban sewage pipe networks are a crucial component of urban water environment infrastructure, and their operational efficiency directly impacts the influent quality and pollutant reduction efficiency of sewage treatment plants. During actual construction and long-term service, sewage pipes are prone to localized defects due to factors such as substandard construction techniques, uneven foundation settlement, tree root intrusion, loose pipe joints, or corrosion damage. These defects allow groundwater, rainwater, and other external water to seep into the pipes through the damaged sections, reducing the effective flow cross-section for sewage transport, lowering the influent pollutant concentration at sewage treatment plants, and consequently affecting the network's operational efficiency.

[0003] Currently, for the problem of groundwater infiltration, existing technologies mainly rely on closed-circuit television (CCTV) inspection of pipelines for qualitative or semi-quantitative identification of internal pipeline defects. However, these methods can only obtain the visible defect morphology of the pipeline inner wall and cannot quantitatively assess the actual infiltration flow rate of groundwater into the pipeline and its dynamic evolution process. Because sewage pipelines are usually buried at great depths and operate in complex environments, the background sewage flow rate is difficult to control stably due to fluctuations in upstream flow, and the groundwater infiltration flow rate is difficult to measure directly under operating conditions, making in-situ experimental research on infiltration mechanisms extremely difficult. Furthermore, existing research mostly focuses on leakage descriptions under single defect morphologies, lacking multi-factor collaborative simulation methods for the infiltration process under multi-media coupling effects of pipelines, soil, and groundwater. There are also no experimental methods or devices that can establish a quantitative mapping relationship between infiltration flow rate and infiltration defect levels (such as dripping, line leakage, seepage, and gushing). Summary of the Invention

[0004] This invention provides a device for simulating infiltration defects in sewage pipelines, a model training method, and a prediction method. It addresses the technical problems in existing technologies, such as the inability to quantitatively simulate groundwater infiltration processes under laboratory conditions, the inability to simultaneously acquire infiltration morphology and flow rate data, and the inability to establish a quantitative mapping relationship between infiltration-related parameters and pipeline infiltration defects. These issues prevent trenchless quantitative assessment of pipeline infiltration defects. This invention can accurately simulate the groundwater infiltration process in drainage pipelines, simultaneously acquire dynamic data on infiltration flow rate and morphology, reveal the impact mechanism of groundwater infiltration into pipelines, and provide technical support for trenchless, quantitative assessment of pipeline defect levels.

[0005] In a first aspect, the present invention provides a device for simulating infiltration defects in sewage pipelines, comprising: a simulated soil tank for filling with soil and providing infiltration simulation space; a defective simulated pipeline, which is a replaceable pipeline specimen constructed within the simulated soil tank, and its pipe wall is pre-installed with defects; a water supply device connected to the simulated soil tank for providing water head pressure; a camera monitoring system for acquiring images of the infiltration morphology at the pipeline defect; an infiltration flow monitoring system for acquiring infiltration flow data at the pipeline outlet, wherein the acquisition sequence of the camera monitoring system and the infiltration flow monitoring system is synchronized; a data processing unit for processing the infiltration morphology images and infiltration flow data acquired at the same time to generate training data; the training data is used to train a deep learning model to obtain a sewage pipeline infiltration defect prediction model; the sewage pipeline infiltration defect prediction model is used to predict the infiltration morphology category and outflow projection area of ​​the pipeline under test based on the infiltration correlation parameters of the pipeline under test, wherein the infiltration correlation parameters include the infiltration flow rate.

[0006] This invention achieves spatiotemporal synchronous acquisition of infiltration morphology images and infiltration flow data at pipeline defects through the linkage of a camera monitoring system and an infiltration flow monitoring system. This solves the problem in existing technologies that cannot simultaneously acquire infiltration morphology and infiltration flow data under laboratory conditions. Based on this, training data is generated using the synchronously acquired data and a deep learning model is trained to obtain a prediction model that can predict the pipeline infiltration morphology type and outflow projection area based on infiltration-related parameters such as infiltration flow. This establishes a quantitative mapping relationship from measurable parameters to infiltration defects, filling the technical gap in existing technologies that cannot perform trenchless quantitative assessment of pipeline infiltration defects.

[0007] In one optional embodiment, the water supply device is a liftable water tank, which controls the water head pressure through the principle of hydrostatic pressure, with the pressure fluctuation error being less than or equal to a preset pressure fluctuation threshold. This embodiment, by setting the water supply device as a liftable water tank and utilizing the principle of hydrostatic pressure to control the water head pressure, and keeping the pressure fluctuation error within a preset threshold range, can provide stable and adjustable water head pressure conditions for simulation experiments. This allows for a realistic reproduction of the groundwater infiltration environment under different burial depths, ensuring the reliability and repeatability of experimental data.

[0008] In one optional implementation, the infiltration flow monitoring system includes an electronic balance positioned at the outlet of the defect-simulated pipe to collect infiltration flow data using a weighing method. This implementation enables high-precision real-time measurement of groundwater infiltration, avoiding errors caused by indirect calculations in traditional methods, and providing accurate flow label data for subsequent deep learning model training.

[0009] In one optional embodiment, the camera monitoring system includes an industrial camera used in conjunction with a tracer to capture images of the infiltration morphology at pipe defects. This embodiment, by using an industrial camera in conjunction with a tracer to capture images of the infiltration morphology at pipe defects, can clearly capture the flow pattern and outflow range of groundwater as it seeps into the pipe through the surrounding soil, providing high-quality image data for subsequent image processing to extract the outflow projection area.

[0010] In one optional implementation, the data processing unit is equipped with an image processing algorithm; the image processing algorithm is used to process the infiltration morphology image and extract the outflow projection area. This implementation, by incorporating an image processing algorithm into the data processing unit to automatically process the infiltration morphology image and extract the outflow projection area, can transform the infiltration morphology from qualitative observation into a quantitative indicator, providing objective and consistent training labels for the deep learning model and avoiding the subjectivity and inconsistency caused by manual interpretation.

[0011] In one optional implementation, the infiltration correlation parameters also include hydraulic head difference, soil parameters, and defect size. This implementation incorporates hydraulic head difference, soil parameters, and defect size into the infiltration correlation parameters, enabling the deep learning model to learn the infiltration patterns under the coupled effects of multiple factors, thus improving the model's adaptability to different operating conditions and its prediction accuracy.

[0012] In one optional implementation, the deep learning model is a two-branch deep neural network, including a shared feature layer, a classification branch, and a regression branch. The classification branch outputs the infiltration morphology category, and the regression branch outputs the outflow projection area. The training data includes a spatiotemporal correlation matrix composed of infiltration correlation parameters, infiltration morphology categories, and outflow projection areas at the same time. The deep learning model is trained using the infiltration correlation parameters as input features and the infiltration morphology category and outflow projection area as training labels to obtain a sewage pipeline infiltration defect prediction model. This implementation adopts a two-branch deep neural network structure containing a shared feature layer, a classification branch, and a regression branch, and uses the spatiotemporal correlation matrix composed of infiltration correlation parameters, infiltration morphology categories, and outflow projection areas at the same time as training data. This allows the model to simultaneously learn the classification task of infiltration morphology categories and the regression task of outflow projection area. The shared feature extraction layer promotes information complementarity between multiple tasks, thereby synchronously outputting the qualitative category and quantitative area of ​​infiltration defects, providing more comprehensive predictive information for the refined determination of pipeline defect levels.

[0013] In one optional implementation, the data processing unit further determines the pipeline infiltration defect level based on the infiltration morphology category and outflow projection area output by the sewage pipeline infiltration defect prediction model. This implementation automatically determines the pipeline infiltration defect level by the data processing unit based on the infiltration morphology category and outflow projection area output by the model, realizing full automation from raw data acquisition to defect level assessment, and completing quantitative classification diagnosis of defects without manual intervention.

[0014] Secondly, this invention provides a method for training a prediction model for sewage pipeline infiltration defects, comprising the following steps: conducting a simulation experiment using the apparatus described in the first aspect above or any corresponding embodiment, simultaneously acquiring infiltration morphology images at the pipeline defect and infiltration flow data at the pipeline outlet; processing the infiltration morphology images to extract the outflow projection area; aligning the infiltration correlation parameters, infiltration morphology categories, and outflow projection areas acquired at the same time to construct a spatiotemporal correlation matrix; training a deep learning model based on the spatiotemporal correlation matrix to establish a quantitative mapping relationship from infiltration correlation parameters to infiltration morphology categories and outflow projection areas, thereby obtaining a prediction model for sewage pipeline infiltration defects. This invention, by using the aforementioned apparatus to conduct simulation experiments and simultaneously acquire infiltration morphology images and infiltration flow data, and then constructing a spatiotemporal correlation matrix to train a deep learning model, can establish a quantitative mapping relationship from measurable parameters such as infiltration flow rate to infiltration morphology categories and outflow projection areas, providing a reliable prediction model for trenchless intelligent diagnosis of pipeline infiltration defects.

[0015] Thirdly, this invention provides a method for predicting infiltration defects in sewage pipelines, comprising the following steps: obtaining infiltration correlation parameters of the pipeline to be tested, including infiltration flow rate; inputting the infiltration correlation parameters into a sewage pipeline infiltration defect prediction model obtained in the second aspect above or any corresponding embodiment, and outputting the infiltration morphology category and outflow projection area of ​​the pipeline to be tested. This invention, by obtaining infiltration correlation parameters such as the infiltration flow rate of the pipeline to be tested and inputting them into a pre-trained prediction model, can directly output the infiltration morphology category and outflow projection area of ​​the pipeline, achieving rapid quantitative diagnosis of infiltration defects in underground pipelines without excavating the ground or interrupting pipeline operation, significantly reducing detection costs and engineering risks.

[0016] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the sewage pipeline infiltration defect prediction model training method of the second aspect or any corresponding embodiment thereof, or the sewage pipeline infiltration defect prediction method of the third aspect or any corresponding embodiment thereof. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a first structural block diagram of a sewage pipeline infiltration defect simulation device according to an embodiment of the present invention; Figure 2 This is a second structural block diagram of a sewage pipeline infiltration defect simulation device according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating a method for training a prediction model for infiltration defects in sewage pipelines according to an embodiment of the present invention. Figure 4 This is a flowchart illustrating a method for predicting infiltration defects in sewage pipelines according to an embodiment of the present invention. Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention; Among them, 1 represents a liftable water tank; 2 represents a transparent simulated soil tank; 3 represents a defect simulated pipeline; 4 represents an industrial camera; 5 represents an electronic balance; and 6 represents a lifting platform. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0020] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0021] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0022] During long-term service, urban sewage pipelines are prone to localized damage due to construction defects, foundation settlement, loose joints, or corrosion. This allows groundwater and other external water to seep into the pipeline, reducing the influent concentration to sewage treatment plants and affecting the network's operational efficiency. Currently, closed-circuit television (CCTV) inspection is mainly used to qualitatively or semi-quantitatively identify internal pipeline defects. However, this method can only obtain the visible morphology of defects on the inner wall and cannot quantitatively assess the actual infiltration flow of groundwater into the pipeline. Due to the deep burial depth and complex operating environment of pipelines, it is difficult to directly separate and measure groundwater infiltration flow under operating conditions, making in-situ field studies extremely challenging. Furthermore, existing technologies lack multi-factor collaborative simulation methods for the infiltration process under multi-media coupling of pipelines, soil, and groundwater, and cannot establish a quantitative mapping relationship between measurable parameters such as infiltration flow and pipeline infiltration defects. This results in a lack of effective experimental equipment and methods to support the trenchless quantitative assessment of pipeline infiltration defects. Based on this, the present invention provides a sewage pipeline infiltration defect simulation device, model training method and prediction method to solve the technical problems in the prior art that it is impossible to quantitatively simulate the groundwater infiltration process under laboratory conditions, impossible to simultaneously obtain infiltration morphology and infiltration flow data, and thus impossible to establish a quantitative mapping relationship between infiltration correlation parameters and pipeline infiltration defects, resulting in the inability to achieve trenchless quantitative assessment of pipeline infiltration defects.

[0023] This embodiment provides a device for simulating sewage pipe infiltration defects. Figure 1 This is a structural diagram of a sewage pipeline infiltration defect simulation device according to an embodiment of the present invention, as shown below. Figure 1 As shown, the device includes: Simulated soil troughs are used to fill soil and provide space for simulated infiltration; Defect simulation pipe 3 is a replaceable pipe specimen constructed in a simulated soil tank, with pre-installed defects on its pipe wall; A water supply device, connected to a simulated earthen tank, is used to provide water head pressure; A camera monitoring system is used to collect images of the seepage patterns at pipeline defects; The infiltration flow monitoring system is used to collect infiltration flow data at the outlet of the pipeline, and the acquisition time of the camera monitoring system is synchronized with that of the infiltration flow monitoring system. The data processing unit processes infiltration morphology images and infiltration flow data collected at the same time to generate training data. The training data is used to train a deep learning model to obtain a sewage pipeline infiltration defect prediction model. The sewage pipeline infiltration defect prediction model is used to predict the infiltration morphology type and outflow projection area of ​​the pipeline under test based on the infiltration correlation parameters of the pipeline under test. The infiltration correlation parameters include infiltration flow.

[0024] In an optional implementation, the simulated soil tank is a transparent simulated soil tank 2, see [link to implementation details]. Figure 2 It is made of high-strength transparent materials, such as plexiglass; the soil inside the simulated soil trough is filled with a layered compaction process to restore the physical parameters of the soil around the pipe, such as moisture content, density and permeability coefficient, to simulate the real buried environment.

[0025] In one optional implementation, the defect simulation pipe 3 includes at least one of the typical defect forms such as interface leakage, pipe wall perforation and cracks, and the outlet end of the defect simulation pipe 3 is connected to the infiltration flow monitoring system.

[0026] In one optional embodiment, the water supply device is a liftable water tank 1, see [link to previous embodiment]. Figure 2 Equipped with a vertical lifting mechanism and water level sensor, water pressure sensor, etc., it controls the water head pressure through the principle of water column static pressure. The pressure fluctuation error is less than or equal to the preset pressure fluctuation threshold, for example, the pressure fluctuation error is ≤0.1kPa, so as to achieve stable water head infiltration conditions.

[0027] In one alternative implementation, the infiltration flow monitoring system includes an electronic balance 5, see [link to relevant documentation]. Figure 2 An electronic balance 5 is installed at the outlet of the defect simulation pipe 3 to collect infiltration flow data by weighing. The electronic balance 5 can be a high-precision electronic balance, installed at the end of the outlet of the defect simulation pipe 3, and uses the weighing method to collect the mass change of the infiltrated water in real time. High-precision instantaneous infiltration flow data is obtained through time differential calculation.

[0028] In another optional implementation, the infiltration flow monitoring system includes an electronic balance 5 and a flow acquisition module. The electronic balance 5 is installed at the outlet of the defect simulation pipe 3 to collect effluent quality data by weighing. The flow acquisition module is linked to the electronic balance 5, using a 24-bit analog-to-digital converter and a synchronous clock to continuously record the effluent quality changes at a sampling frequency of not less than 100Hz, and calculates the instantaneous infiltration flow in real time using a differential algorithm. The flow acquisition module is also used to receive the output signal of the head sensor and is time-synchronized with the trigger signal of the camera monitoring system, with a synchronization error of less than 1ms, to ensure the spatiotemporal consistency of infiltration flow, head pressure, and infiltration morphology data.

[0029] In one alternative implementation, the video monitoring system includes an industrial camera 4, see [link to relevant documentation]. Figure 2 It is used in conjunction with tracers to capture images of the infiltration morphology at pipeline defects. The industrial camera 4 is a high frame rate industrial camera with a frame rate of not less than 100fps, which is placed outside a transparent simulated soil tank. In conjunction with a colorimetric tracer (such as sodium fluorescein or rhodamine aqueous solution), it continuously captures the infiltration morphology of groundwater after penetrating the soil and gushing out through pipeline defects, including four typical levels: dripping, line leakage, gushing leakage and jetting leakage.

[0030] In one optional embodiment, the device further includes a head sensor, which is deployed at different depths and horizontal positions along the simulated soil trench to monitor the spatial distribution and dynamic changes of water head pressure within the trench in real time. For example, the head sensor can be installed through openings in the sidewall of the trench, with the sensor probe encased in permeable stone. The measurement range is set according to the maximum experimental head, typically 0–50 kPa, with an accuracy of no less than ±0.1%FS. Through simultaneous multi-point measurements, the seepage driving force (i.e., the head difference) within the soil and its variation over time can be accurately calculated, providing key input parameters for deep learning models.

[0031] In one alternative implementation, the infiltration-related parameters also include hydraulic head difference, soil parameters, and defect size.

[0032] In one optional implementation, the data processing unit is equipped with an image processing algorithm and a deep learning model. The image processing algorithm is used to quantitatively calculate the outflow projection area of ​​the infiltration morphology images collected by the camera monitoring system, and the deep learning model is used to perform fusion analysis on multi-source data such as infiltration flow, water head difference, and soil parameters.

[0033] In one optional implementation, an image processing algorithm is used to process the infiltration morphology image and extract the outflow projection area. The image processing algorithm includes sequentially performing grayscale conversion, filtering and noise reduction, binarization segmentation, and morphological operations on the infiltration morphology image to extract the contour of the tracer-stained region and quantitatively calculate the outflow projection area.

[0034] In one optional implementation, the deep learning model is a two-branch deep neural network, including a shared feature layer, a classification branch, and a regression branch; the classification branch is used to output the infiltration morphology category, and the regression branch is used to output the outflow projection area; the training data includes a spatiotemporal correlation matrix composed of infiltration correlation parameters, infiltration morphology categories, and outflow projection areas at the same time; the deep learning model is trained using infiltration correlation parameters as input features and infiltration morphology categories and outflow projection areas as training labels to obtain a sewage pipeline infiltration defect prediction model.

[0035] In one optional implementation, the data processing unit further determines the level of infiltration defect in the pipeline based on the infiltration morphology category and outflow projection area output by the sewage pipeline infiltration defect prediction model.

[0036] The sewage pipeline infiltration defect simulation device provided in this embodiment achieves spatiotemporal synchronous acquisition of infiltration morphology images and infiltration flow data at pipeline defects through the linkage of a camera monitoring system and an infiltration flow monitoring system. This solves the problem in the prior art that it is impossible to simultaneously acquire infiltration morphology and infiltration flow data under laboratory conditions. Based on this, training data is generated using the synchronously acquired data and a deep learning model is trained to obtain a prediction model that can predict the pipeline infiltration morphology type and outflow projection area based on infiltration-related parameters such as infiltration flow. This establishes a quantitative mapping relationship from measurable parameters to infiltration defects, filling the technical gap in the prior art that cannot perform trenchless quantitative assessment of pipeline infiltration defects.

[0037] This embodiment also provides a method for training a prediction model for infiltration defects in sewage pipelines. Figure 3 This is a flowchart of a method for training a sewage pipeline infiltration defect prediction model according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps: Step S301, using Figure 1 or Figure 2 The apparatus of the illustrated embodiment was used to conduct a simulation experiment, simultaneously acquiring images of the infiltration morphology at the pipeline defect and infiltration flow data at the pipeline outlet.

[0038] Specifically, step S301 includes the following sub-steps: Step S3011: Reconstruction of the surrounding soil environment. Install and seal the defective simulated pipeline specimen 3. Based on the soil parameters of the target research scenario, use a layered compaction process to fill and compact the simulated soil layer by layer in the simulated soil trench, recreating the real soil environment around the pipeline. Soil parameters include soil type, dry density, moisture content, and permeability coefficient, and the physical and mechanical properties of each soil layer are recorded.

[0039] Step S3012, Groundwater Head Pressure Control. The liftable water tank 1 is adjusted to a preset height via the lifting platform 6, and groundwater head pressure is applied to the bottom or side of the soil tank using the principle of hydrostatic pressure of water column; a water head sensor is used to monitor the pore water pressure in the soil tank in real time to ensure that the water head difference remains constant during the experiment.

[0040] Step S3013: Synchronous monitoring of the infiltration process. Open the outlet of the defect simulation pipe 3, allowing groundwater to seep into the pipe defect under pressure. Start the electronic balance 5 to continuously record the mass change at the outlet at a sampling frequency of no less than 100Hz, calculating the instantaneous infiltration flow rate and cumulative infiltration volume. Simultaneously, start the industrial camera 4, using pre-injected colorimetric tracer, to continuously acquire images of the groundwater infiltration evolution, recording four typical infiltration modes and their evolution processes: dripping, line leakage, gushing, and spouting.

[0041] Step S302: Process the infiltration morphology image and extract the outflow projection area.

[0042] Specifically, the image sequence acquired by the industrial camera 4 is processed by the image processing algorithm in the data processing unit, performing grayscale conversion, filtering and noise reduction, binarization segmentation and morphological operations in sequence to extract the contour of the tracer-stained area; based on the pixel calibration results, the projected area of ​​groundwater gushing out at the defect in each frame of the image is quantitatively calculated, and the dynamic changes of the projected area of ​​gushing out are analyzed according to the time series.

[0043] Step S303: Align the infiltration correlation parameters, infiltration morphology categories, and outflow projection areas collected at the same time to construct a spatiotemporal correlation matrix.

[0044] Specifically, data such as infiltration flow rate, head difference, soil permeability coefficient, defect size parameters, infiltration morphology type, and outflow projection area collected at the same time are aligned to form a spatiotemporal correlation matrix; this matrix covers multiple working conditions and provides training samples for subsequent models.

[0045] Step S304: Train a deep learning model based on the spatiotemporal correlation matrix, establish a quantitative mapping relationship from infiltration correlation parameters to infiltration morphology category and outflow projection area, and obtain a sewage pipeline infiltration defect prediction model.

[0046] Specifically, the construction of a deep learning mapping model includes the following: Input layer: Normalize parameters such as infiltration flow rate, head difference, soil permeability coefficient, and equivalent diameter of defects to form the input feature vector.

[0047] Feature extraction layer: Construct a shared feature layer to extract high-order coupled features from multi-source data. This shared feature layer can employ a multi-layer fully connected or convolutional structure.

[0048] Parallel output branches: These include a classification branch and a regression branch. The classification branch encodes the infiltration pattern as orthogonal feature vectors (e.g., dripping: [1,0,0,0]). Line leakage: [0,1,0,0] Leakage: [0,0,1,0] Spray: [0,0,0,1] The probability distribution of each morphology is output using the Softmax function; the regression branch uses the ReLU function to output the continuous regression value of the outflow projection area.

[0049] Joint loss function: It integrates cross-entropy loss (classification task) and mean squared error loss (regression task), and introduces dynamic weight factors to balance the learning progress of the two tasks; an adaptive moment estimator optimizer is used for model training.

[0050] Defect Level Determination: Based on the infiltration pattern and outflow projection area predicted by the model, the infiltration defect level of the pipeline is automatically determined by comparing it with the preset defect level classification standard. The preset defect level classification standard can be selected from the "Technical Specification for Inspection and Evaluation of Urban Drainage Pipelines" (CJJ 181-2012).

[0051] Model validation: Using a reserved test dataset, the classification accuracy is evaluated by the confusion matrix and the regression accuracy is evaluated by the relative error to validate the model performance.

[0052] The training method provided in this embodiment uses the above-mentioned device to conduct simulation experiments and simultaneously collect infiltration morphology images and infiltration flow data, and then constructs a spatiotemporal correlation matrix to train a deep learning model. This method can establish a quantitative mapping relationship from measurable parameters such as infiltration flow to infiltration morphology categories and outflow projection areas, providing a reliable prediction model for trenchless intelligent diagnosis of pipeline infiltration defects.

[0053] This embodiment also provides a method for predicting infiltration defects in sewage pipelines. Figure 4 This is a flowchart of a method for predicting infiltration defects in sewage pipelines according to an embodiment of the present invention, as shown below. Figure 4 As shown, the process includes the following steps: Step S401: Obtain the infiltration correlation parameters of the pipeline to be tested, including the infiltration flow rate.

[0054] Specifically, infiltration flow data can be obtained by deploying flow monitoring equipment at the pipeline site. At the same time, the head difference can be obtained by combining the groundwater level monitoring data of the pipeline area. Soil parameters and defect size information can be obtained based on the pipeline design data to form complete infiltration-related parameters.

[0055] Step S402, input the infiltration correlation parameters Figure 3 The wastewater pipeline infiltration defect prediction model obtained in the illustrated embodiment outputs the infiltration morphology category and outflow projection area of ​​the pipeline under test.

[0056] Specifically, after normalizing the acquired infiltration correlation parameters, the data is input into a trained sewage pipeline infiltration defect prediction model. The model extracts high-order coupling features from multi-source data through a shared feature layer, outputs the infiltration morphology category through a classification branch, and outputs the outflow projection area through a regression branch. Furthermore, based on the output infiltration morphology category and outflow projection area, the infiltration defect level of the pipeline under test can be determined by comparing it with a preset defect level classification standard.

[0057] In an optional implementation, the method of this embodiment further includes: determining the infiltration defect level based on the infiltration morphology type and outflow projection area of ​​the pipeline to be tested.

[0058] The sewage pipeline infiltration defect prediction method provided in this embodiment can directly output the infiltration morphology type and outflow projection area of ​​the pipeline by obtaining infiltration-related parameters such as the infiltration flow rate of the pipeline to be tested and inputting them into a pre-trained sewage pipeline infiltration defect prediction model. This enables rapid quantitative diagnosis of infiltration defects in underground pipelines without digging up the ground or stopping the pipeline operation, significantly reducing detection costs and engineering risks.

[0059] The wastewater pipeline infiltration defect simulation device, model training method, and prediction method provided in this embodiment have the following beneficial effects: A pipeline-soil-groundwater coupling simulation experimental device was constructed, integrating a liftable water tank, a simulated soil trough, a defect simulation pipeline, a camera monitoring system, and an infiltration flow monitoring system. This device enables controllable simulation and quantitative monitoring of the entire process of groundwater infiltration in buried drainage pipelines under controlled indoor conditions.

[0060] By integrating precise flow measurement using gravimetric methods with visualization technology using industrial cameras and tracers, multi-dimensional quantitative data such as infiltration flow rate, infiltration morphology, and outflow projection area are acquired simultaneously. A dual-modal synchronous monitoring method linking the camera monitoring system and the infiltration flow monitoring system is established, and a spatiotemporal correlation matrix with multiple coupled factors is constructed, forming a structured dataset that can be used for deep learning training.

[0061] A mapping model between infiltration correlation parameters and infiltration defect levels based on a deep neural network was designed. Infiltration morphology categories are encoded as orthogonal feature vectors. A dual-branch parallel output structure with a shared feature layer and classification and regression branches is employed. Combined with a joint loss function of cross-entropy and mean squared error, simultaneous prediction from infiltration correlation parameters such as infiltration flow rate to infiltration morphology categories and outflow projection area is achieved. The model output is mapped to a standardized defect level, enabling automatic determination of infiltration defect levels. This provides experimental evidence and algorithmic support for trenchless inspection and intelligent assessment of sewage pipe networks.

[0062] Through the aforementioned integrated experiment and intelligent recognition mechanism, this invention can accurately and quantitatively obtain the groundwater infiltration flow rate, infiltration morphology, and outflow projection area of ​​drainage pipelines, clearly revealing the evolution mechanism of groundwater infiltration networks and providing reliable experimental basis and data support for determining the level of infiltration defects. Based on the head pressure control of a liftable water tank, automated real-time monitoring, and a dual-branch deep learning model, it can flexibly adapt to different soil cover conditions, groundwater heads, pipeline defect forms, and flow conditions, meeting the diverse needs of research on the mechanism of external water infiltration in drainage pipelines, quantitative assessment of defects, and engineering inspection. This invention significantly improves the quantification, intelligence, and standardization of infiltration defect detection in drainage pipelines, reduces the difficulty and uncertainty of on-site in-situ testing, and reduces the blindness and maintenance costs of pipeline disease treatment. By accurately analyzing the mapping relationship between infiltration-related parameters and defect levels, it can effectively identify external water infiltration points and guide targeted repair, helping to increase the concentration of pollutants in the influent of sewage treatment plants, improve the operational efficiency of sewage systems, reduce treatment costs and energy consumption, and provide key technical support for the refined management and control of urban drainage networks and the improvement of water environment quality and efficiency.

[0063] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0064] The following is a detailed reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from memory 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device. The processor 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0065] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0066] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a memory 508, or installed from a ROM 502. When the computer program is executed by the processor 501, it performs the functions defined in the wastewater pipeline infiltration defect prediction model training method or the wastewater pipeline infiltration defect prediction method of the embodiments of the present invention.

[0067] Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0068] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the wastewater pipeline infiltration defect prediction model training method or wastewater pipeline infiltration defect prediction method shown in the above embodiments is implemented.

[0069] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0070] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A device for simulating infiltration defects in sewage pipes, characterized in that, The device includes: Simulated soil troughs are used to fill soil and provide space for simulated infiltration; The defect-simulated pipeline is a replaceable pipeline specimen constructed within the simulated soil tank, with pre-installed defects on its pipe wall. A water supply device, connected to the simulated earthen trough, is used to provide water head pressure; A camera monitoring system is used to collect images of the seepage patterns at pipeline defects; An infiltration flow monitoring system is used to collect infiltration flow data at the outlet of the pipeline, and the acquisition timing of the camera monitoring system is synchronized with that of the infiltration flow monitoring system. The data processing unit is used to process the infiltration morphology image and the infiltration flow data collected at the same time to generate training data; the training data is used to train a deep learning model to obtain a sewage pipeline infiltration defect prediction model; the sewage pipeline infiltration defect prediction model is used to predict the infiltration morphology category and outflow projection area of ​​the pipeline under test based on the infiltration correlation parameters of the pipeline under test, the infiltration correlation parameters including the infiltration flow rate.

2. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The water supply device is a liftable water tank that controls the water head pressure through the principle of static pressure of water column, and the pressure fluctuation error is less than or equal to the preset pressure fluctuation threshold.

3. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The infiltration flow monitoring system includes an electronic balance, which is installed at the outlet of the defect simulation pipeline and is used to collect infiltration flow data by weighing.

4. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The camera monitoring system includes an industrial camera, used in conjunction with a tracer to capture images of the infiltration morphology at pipeline defects.

5. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The data processing unit is equipped with an image processing algorithm; the image processing algorithm is used to process the infiltration morphology image and extract the outflow projection area.

6. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The infiltration-related parameters also include hydraulic head difference, soil parameters, and defect size.

7. The sewage pipeline infiltration defect simulation device according to claim 5, characterized in that, The deep learning model is a dual-branch deep neural network, including a shared feature layer, a classification branch, and a regression branch; the classification branch is used to output the infiltration morphology category, and the regression branch is used to output the outflow projection area. The training data includes a spatiotemporal correlation matrix consisting of infiltration correlation parameters, infiltration morphology categories, and outflow projection areas at the same time. The deep learning model is trained using the infiltration correlation parameters as input features and the infiltration morphology category and the outflow projection area as training labels to obtain a sewage pipeline infiltration defect prediction model.

8. The sewage pipeline infiltration defect simulation device according to claim 1, characterized in that, The data processing unit also determines the level of pipeline infiltration defects based on the infiltration morphology category and outflow projection area output by the sewage pipeline infiltration defect prediction model.

9. A training method for a prediction model of infiltration defects in sewage pipelines, characterized in that, Includes the following steps: The device described in any one of claims 1-8 was used to conduct a simulation experiment, and images of the infiltration morphology at the pipeline defect and infiltration flow data at the pipeline outlet were collected simultaneously. The infiltration morphology image is processed to extract the outflow projection area; Align the infiltration correlation parameters, infiltration morphology categories, and outflow projection areas collected at the same time to construct a spatiotemporal correlation matrix; A deep learning model is trained based on the spatiotemporal correlation matrix to establish a quantitative mapping relationship from the infiltration correlation parameters to the infiltration morphology category and the outflow projection area, thereby obtaining a sewage pipeline infiltration defect prediction model.

10. A method for predicting infiltration defects in sewage pipelines, characterized in that, Includes the following steps: Obtain the infiltration correlation parameters of the pipeline under test, wherein the infiltration correlation parameters include the infiltration flow rate; The infiltration correlation parameters are input into the sewage pipeline infiltration defect prediction model obtained by the method described in claim 9, and the infiltration morphology category and outflow projection area of ​​the pipeline under test are output.

Images