A rapid detection device and method for drainage pipes

Abstract

The present invention discloses a rapid detection device and method for drainage pipelines. The device includes an underwater unmanned vehicle, a control system, an image acquisition device, and a water quality detection device. The underwater unmanned vehicle is provided with a control system, an image acquisition device, and a water quality detection device. The image acquisition device and the water quality detection device are respectively connected to the control system. The control system includes a central control subsystem and a motion subsystem. The motion subsystem is connected to the central control subsystem. The image acquisition device includes a stable pan-tilt head, a lighting module, and a camera. The stable pan-tilt head, the lighting module, and the camera are respectively connected to the central control subsystem. The water quality detection device includes a probe and a driver. The probe is connected to the driver, and the driver is connected to the central control subsystem. The method is a detection method applied to the above device. The present invention can achieve rapid detection of pipeline anomalies. The present invention can be widely applied in the field of management and maintenance of drainage pipe networks.

Description

Technical Field

[0001] The present invention relates to the field of management and maintenance of drainage pipe networks, and particularly to a rapid detection device and method for drainage pipes. Background Art

[0002] As an important infrastructure of a city, the quality of management and maintenance of the drainage pipe network has an extremely important impact on the development of the city. Pipeline detection includes two parts: detection of pipeline defects and detection of abnormal water quality in the pipeline. Currently, the detection of drainage pipeline defects mainly uses methods such as CCTV and QV, while the detection of abnormal water quality in the pipeline is mainly achieved by installing on-line water quality monitoring instruments in the pipeline. When detecting pipeline defects, it is necessary to block the upstream and downstream pipe sections, which severs the connection between pipeline defect detection and abnormal water quality detection in the pipeline and weakens the interpretability of pipeline detection results. Moreover, traditional pipeline defect detection cannot perform continuous real-time detection. It relies on manual viewing of images for analysis, resulting in low detection efficiency. And arranging on-line water quality monitoring instruments can only monitor the water quality of fixed-node water pipelines. To monitor the water quality of the entire pipe network, a large number of instruments must be installed. Therefore, generally speaking, the detection cost of pipelines is very high. It is very necessary to develop a highly efficient and low-cost pipeline rapid detection device to fill the gap in the current pipeline rapid detection field, which is used in the preliminary census stage of pipeline detection and greatly improves the pertinence of pipeline detection and repair work.

[0003] With the rapid development of computer technology, the application of image recognition technology is becoming more and more extensive. Nowadays, image recognition technology has been widely applied in fields or products such as public transportation, medicine, smart phones, and plant recognition. Indicators such as the resolution, speed, and accuracy of image recognition have reached the application requirements for pipeline defect recognition.

[0004] With the development of the production technology of on-line water quality instruments, many current on-line water quality monitoring instruments can be produced in a miniaturized manner and can be installed on small unmanned platforms for water quality monitoring. Water quality indicators represented by conductivity are sensitive and have a wide detection range, etc., and can be used as representative indicators of water quality anomalies.

[0005] Generally speaking, it is highly feasible to apply image recognition technology and water quality detection technology represented by conductivity to the field of pipeline rapid detection. Summary of the Invention

[0006] In order to solve the above technical problems, the purpose of the present invention is to provide a rapid detection device and method for drainage pipes, which can achieve rapid detection of pipeline anomalies.

[0007] The first technical solution adopted by the present invention is: a rapid detection device for drainage pipes, including an underwater unmanned vehicle, a control system, an image acquisition device, and a water quality detection device. The underwater unmanned vehicle is provided with a control system, an image acquisition device, and a water quality detection device. The image acquisition device and the water quality detection device are respectively connected to the control system. The control system includes a central control subsystem and a motion subsystem. The motion subsystem is connected to the central control subsystem. The image acquisition device includes a stable pan-tilt head, a lighting module, and a camera. The stable pan-tilt head, the lighting module, and the camera are respectively connected to the central control subsystem. The water quality detection device includes a probe and a driver. The probe is connected to the driver, and the driver is connected to the central control subsystem.

[0008] Further, the stable pan-tilt head is installed on the top of the underwater unmanned vehicle, the camera is installed on the stable pan-tilt head, and the lighting module is installed in front of the camera.

[0009] Further, the driver is installed inside the underwater unmanned vehicle, and the probe is installed at the bottom of the underwater unmanned vehicle.

[0010] Further, the central control subsystem includes a data transmission module, a navigation module, an image recognition module, and a water quality analysis module. The navigation module, the image recognition module, and the water quality analysis module are respectively connected to the data transmission module. The lighting module, the stable pan-tilt head, and the motion subsystem are respectively connected to the navigation module. The camera is connected to the image recognition module, and the water quality analysis module is connected to the driver.

[0011] Further, the motion subsystem includes a driving motor and wheels. The driving motor is connected to the wheels, and the driving motor is connected to the navigation module.

[0012] Further, the wheels are installed on both sides of the bottom of the underwater unmanned vehicle, and the number of wheels is not less than 4.

[0013] The second technical solution adopted by the present invention is: a rapid detection method for drainage pipes, including the following steps:

[0014] Determine the detection range and the device placement location;

[0015] Obtain the current image based on image acquisition and identify the liquid level in the detection pipe;

[0016] Determine the detection mode according to the liquid level in the detection pipe;

[0017] According to the preset requirements of the detection mode, perform image acquisition and water quality index determination based on the image acquisition device and the water quality detection device;

[0018] Based on the image recognition module, quickly identify the image of the data obtained by image acquisition, and obtain the rapid identification result of pipeline defects;

[0019] Based on the data obtained from the water quality index measurement by the water quality analysis module, water quality anomaly identification is carried out to obtain the water quality anomaly identification result.

[0020] Furthermore, it also includes:

[0021] Combining the pipeline defect rapid identification result and the water quality anomaly identification result to obtain the pipeline rapid detection result;

[0022] Output the pipeline rapid detection result and conduct manual review of some defective pipelines and manual tracing of water quality anomalies to obtain the review result;

[0023] Based on the review result, adjust the parameters of the image rapid classifier in the image recognition module and the parameters of the water quality anomaly recognition model in the water quality analysis module.

[0024] Furthermore, the construction method of the image rapid classifier in the image recognition module includes the following steps:

[0025] Select the defective pipeline sample image;

[0026] Extract the invariant moment features of the defect type and grade of the defective pipeline sample image to obtain the invariant moment features;

[0027] Normalize the input vector of the invariant moment features to convert the invariant moment features to between [0, 1];

[0028] Perform fuzzy processing on the invariant moment features and classify the fuzzy processed invariant moment features based on the neural network to obtain the predicted classification result;

[0029] According to the predicted classification result and the expected target output corresponding to the invariant moment features, obtain the training error;

[0030] Furthermore, the water quality anomaly recognition model in the water quality analysis module is specifically a water quality anomaly recognition model constructed based on the pipeline experimental platform, including the following steps:

[0031] Preset the experimental conditions and simulate to obtain the water quality of the experimental pipeline;

[0032] Measure the distribution of the conductivity of the water quality of the experimental pipeline to obtain the conductivity at different positions in the pipeline;

[0033] Based on the conductivity at different positions in the pipeline, construct the conductivity distribution law, obtain the conductivity fluctuation range under different abnormal water quality conditions, and select appropriate upper and lower limit fluctuation values as the standard for the water quality analysis model to judge whether there is abnormal water quality access, to obtain the water quality anomaly recognition model;

[0034] Based on the water quality anomaly recognition model, identify the actual abnormal pipeline water quality and conduct error analysis on the recognition result to obtain the training error;

[0035] Adjust the parameters of the water quality anomaly recognition model based on the training error to obtain a trained water quality anomaly recognition model.

[0036] The beneficial effects of the method of the present invention are as follows: Through the detection device and the detection method, the rapid detection of drainage pipeline anomalies is realized, thereby reducing the cost invested in the daily detection of urban pipe networks. Pipeline maintenance personnel can perform targeted manual rechecks and pipe repairs based on the recognition results, greatly improving the operation and maintenance quality of the pipe networks. Description of the Drawings

[0037] Figure 1 is a device diagram of a rapid drainage pipeline detection device of the present invention;

[0038] Figure 2 is a structural block diagram of a rapid drainage pipeline detection device of the present invention;

[0039] Figure 3 is a step flowchart of a detection method applying the rapid drainage pipeline detection device of the present invention;

[0040] Figure 4 is a step flowchart of a method for constructing an image rapid classifier adopted by an image recognition module of a rapid drainage pipeline detection device of the present invention;

[0041] Figure 5 is a step flowchart of a method for constructing a water quality anomaly recognition model adopted by a water quality analysis module of a rapid drainage pipeline detection device of the present invention.

[0042] Reference Numerals: 1, underwater unmanned vehicle; 2, central control subsystem; 3, data transmission module; 4, navigation module; 5, image recognition module; 6, water quality analysis module; 7, motion subsystem; 8, wheels; 9, image acquisition device; 10, camera; 11, stable pan-tilt head; 12, lighting module; 13, water quality detection device; 14, driver; 15, probe; 16, drive motor. Detailed Embodiments

[0043] The following further elaborates on the present invention in detail with reference to the drawings and specific embodiments. For the step numbers in the following embodiments, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0044] Refer to Figure 1 and Figure 2, the present invention provides a rapid detection device for drainage pipes, including an underwater unmanned vehicle 1, a control system, an image acquisition device, and a water quality detection device. The underwater unmanned vehicle 1 is provided with a control system, an image acquisition device 9, and a water quality detection device 13. The image acquisition device 9 and the water quality detection device 13 are respectively connected to the control system. The control system includes a central control subsystem 2 and a motion subsystem 7. The motion subsystem 7 is connected to the central control subsystem 2. The image acquisition device 9 includes a stable pan-tilt head 11, a lighting module 12, and a camera 10. The stable pan-tilt head 11, the lighting module 12, and the camera 10 are respectively connected to the central control subsystem 2. The water quality detection device 13 includes a probe 15 and a driver 14. The probe 15 is connected to the driver 14, and the driver 14 is connected to the central control subsystem 2.

[0045] Specifically, the image acquisition device 9 uses a camera with a pixel count of not less than 8 million pixels. The video quality it captures is not less than 1080p 30fps, and it has an infrared night vision shooting function. The stable pan-tilt head 11 uses a three-axis anti-shake pan-tilt head. The horizontal adjustment direction of this pan-tilt head is 0 - 360°, and the elevation angle adjustment direction includes at least [-45°, 90°], mainly realizing the control of the camera shooting direction and the anti-shake function. The image acquisition device 9 is equipped with a lens protection device, which uses a high-strength transparent lens protection cover and an automatic cleaning tool. The automatic cleaning tool uses a wiper-type cleaning tool. When the lens protection cover is contaminated, the device automatically turns on the cleaning tool for cleaning. The driver 14 is integrated on the internal circuit board of the underwater robot 1. The probe 15 uses a cylindrical probe.

[0046] Further as a preferred embodiment, the stable pan-tilt head 11 is installed on the top of the underwater unmanned vehicle 1, the camera 10 is installed on the stable pan-tilt head 11, and the lighting module 12 is installed in front of the camera 10.

[0047] Further as a preferred embodiment, the driver 14 is installed inside the underwater unmanned vehicle 1, and the probe 15 is installed at the bottom of the underwater unmanned vehicle 1.

[0048] Further as a preferred embodiment, the central control subsystem 2 includes a data transmission module 3, a navigation module 4, an image recognition module 5, and a water quality analysis module 6. The navigation module 4, the image recognition module 5, and the water quality analysis module 6 are respectively connected to the data transmission module 3. The lighting module 12, the stable pan-tilt head 11, and the motion subsystem 7 are respectively connected to the navigation module 4. The camera 10 is connected to the image recognition module 5, and the water quality analysis module 6 is connected to the driver 14.

[0049] Specifically, the navigation module 4 uses GPS signals for positioning and is equipped with no less than 4 distance sensors to achieve the wall obstacle avoidance function. The recognition range of the distance sensors is 0.1 - 1.0 m, and the recognition accuracy is 0.01 m. The data transmission module 3 uses a 4G mobile network or a local wireless network for transmission, and is mainly responsible for the transmission of control information, analysis data, and image information. Preferably, in the case of ideal signal strength, the local wireless network is preferentially used for communication. The lighting module 12 uses an LED lamp with a power of not less than 10 W, the lighting direction is consistent with the camera shooting direction, the lighting distance is not less than 5 m, and the effective lighting brightness is not less than 200 nit. The image recognition module 5 and the water quality analysis module 6 select ARM chips.

[0050] Further as a preferred embodiment, the motion subsystem 7 includes a drive motor and wheels 8. The drive motor 16 is connected to the wheels 8, and the drive motor 16 is connected to the navigation module 4.

[0051] Further as a preferred embodiment, the wheels 8 are installed on both sides of the bottom of the underwater unmanned vehicle 1, and the number of the wheels 8 is not less than 4.

[0052] Referring to Figure 3 , a detection method applied to the above-mentioned drainage pipeline rapid detection device includes the following steps:

[0053] S1. Preparation before detection, determining the detection range and the placement location.

[0054] Specifically, the single - time detection range does not exceed 2 km of pipe length, and the placement location is set at the upstream inspection well of the pipe network.

[0055] S2. Data acquisition, obtaining the current image based on the image acquisition device 9, identifying the liquid level in the detection pipe according to the current image, and selecting the detection mode according to the level of the current liquid level. When the liquid level in the pipe is not higher than the minimum detection liquid level of the water quality detection device, the device switches to the ground detection mode; when the liquid level in the pipe is higher than the position where the camera 10 is located, the device switches to the underwater detection mode; when the liquid level in the pipe is between the above two situations, the device switches to the water - surface detection mode. According to the requirements of the detection mode, image acquisition and water quality index determination are carried out based on the image acquisition device 9 and the water quality detection device 13.

[0056] Specifically, the comprehensive density of the rapid drainage pipeline detection device is at least twice the density of water. When the liquid in the pipeline is full and the flow rate is 1 m / s, it can stay in the pipeline by its own gravity. When the drainage pipeline rapid detection device is in the ground detection mode, the traveling speed of the device in the pipeline is 0.05 - 0.25 m / s. When the drainage pipeline rapid detection device is in the water detection mode and the underwater detection mode, the traveling speed of the device in the pipeline is 0.25 - 0.50 m / s. Additionally, the image extraction interval for image acquisition does not exceed 1 s, and the acquisition interval for water quality index measurement does not exceed 0.5 s.

[0057] S3. Data analysis and result output stage: Based on the image recognition module 5, perform rapid image recognition on the data obtained from image acquisition to obtain the rapid pipeline defect recognition result. Based on the water quality analysis module 6, perform water quality anomaly recognition on the data obtained from water quality index measurement to obtain the water quality anomaly recognition result. Finally, output the rapid pipeline detection result. Additionally, for the water quality anomaly recognition model, simply put, it is the upper and lower limits of the conductivity. If it exceeds a certain range, there may be abnormal water quality, and thus it is inferred that there may be defects in the pipe network.

[0058] S4. Manual review: Output the rapid pipeline detection result and conduct manual review of some defective pipelines and manual tracing of water quality anomalies. Use the review results to adjust the parameters of the rapid image classifier and the parameters of the water quality anomaly recognition model.

[0059] Specifically, adjusting the parameters of the water quality anomaly recognition model means adjusting the upper and lower limits of the conductivity fluctuation.

[0060] Further as a preferred embodiment, refer to Figure 4 , the construction method of the rapid image classifier adopted by the image recognition module 5 includes the following steps:

[0061] The first step: Select pipeline sample images, collect no less than 500 pipeline sample images, which include various pipeline defect categories of structural defects and functional defects. The image pixels are not less than 500W, and the number of images for each type of defect is not less than 100.

[0062] The second step: Extract the invariant moment features of the defect types and grades of the pipeline sample images, and take the higher values of the invariant moment features.

[0063] The third step: Perform invariant moment vector normalization to convert the features in the invariant moment input vector to between [0, 1].

[0064] The fourth step: Perform fuzzy processing on the invariant moment features and classify the fuzzy processed invariant moment features based on the neural network to obtain the predicted classification result.

[0065] Step 5: Obtain the training error according to the predicted classification result and the expected target output corresponding to the invariant moment feature.

[0066] Specifically, taking the actual defect type and grade of the image corresponding to the invariant moment feature as the expected target output, and combining the predicted classification result to obtain the training error.

[0067] Step 6: Conduct initial parameter setting before training. If the recognition accuracy of various defects in the pipeline sample image is not less than 95%, the design requirement is met at this time, perform the output of the image fast classifier, and write the classifier into the image recognition module; if the recognition accuracy of various defects in the pipeline sample image is lower than 95%, then adjust the network parameters using the training error until the design requirement is met.

[0068] Further as a preferred embodiment, refer to Figure 5 The construction method of the water quality anomaly recognition model adopted by the water quality analysis module 6 includes the following steps:

[0069] Specifically, build the model based on the pipeline experimental platform, and the pipeline experimental platform simulates the drainage pipeline based on the pipeline network defects.

[0070] Step 1: Set the experimental conditions at different concentrations, including pipeline flow regime, pipeline water quality, abnormal flow rate, and abnormal flow rate water quality.

[0071] Specifically, the pipeline flow regime at least includes pipe diameter, flow rate, flow velocity, and slope; the experimental pipe diameters include 300mm - 600mm, the experimental flow rates include 30m3 / h - 3000m3 / h, the experimental flow velocities include 0.1m / s - 2.0m / s, and the experimental slope range is [-0.002, +0.002]; the pipeline water quality indicators at least include COD, BOD, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, and pH, mainly simulating urban domestic sewage and surface water class V; for urban domestic sewage, the experimental values of each index are: COD takes 200 - 300mg / L, BOD takes 100 - 200mg / L, ammonia nitrogen takes 8 - 15mg / L, total nitrogen takes 20 - 40mg / L, total phosphorus takes 4 - 8mg / L, suspended solids take 100 - 250mg / L, and pH takes 6 - 9; for surface water class V, the experimental values of each index are: COD takes 20 - 50mg / L, BOD takes 5 - 20mg / L, ammonia nitrogen takes 1 - 5mg / L, total nitrogen takes 2 - 10mg / L, total phosphorus takes 0.1 - 0.5mg / L, suspended solids take 10 - 30mg / L, and pH takes 6 - 9; the abnormal flow rate water quality is the water quality of clear water and industrial wastewater, and the clear water quality is the water quality of surface water class V; the industrial wastewater categories at least include printing and dyeing wastewater, electroplating wastewater, papermaking wastewater, and breeding wastewater, and the conductivity range of its water quality includes 0 - 20ms / cm.

[0072] Second step, conduct the measurement of the conductivity distribution, and measure the conductivity at different positions inside the pipe under the condition of abnormal flow access.

[0073] Specifically, measure the conductivity at different positions inside the pipe under the condition of abnormal flow access. The distance between the measurement points is not greater than 3 cm, and the measurement duration at each measurement point is not less than 10 s.

[0074] Third step, construct the conductivity distribution law, obtain the conductivity fluctuation range under different abnormal water quality conditions, and select appropriate upper and lower limit fluctuation values as the standard for the water quality analysis model to judge whether there is abnormal water quality access.

[0075] Fourth step, use the above model to identify the actual abnormal pipeline water quality, and conduct error analysis on the identification results. If the error exceeds 20%, adjust the model parameters using the training error; if the error does not exceed 20%, output the water quality anomaly identification model and write the model into the water quality analysis module.

[0076] Further as a preferred embodiment, the pipeline rapid detection results include the following:

[0077] Include the rapid identification results of pipeline defects. The rapid identification results of pipeline defects include category, grade, and location. The category includes structural defects and functional defects;

[0078] Specifically, the grade includes first grade, second grade, third grade, and fourth grade; the location includes pipe section location and circumferential location; the structural defects include rupture, deformation, corrosion, misalignment, undulation, disconnection, shedding of interface material, hidden connection of branch pipe, penetration of foreign objects, and leakage; the functional defects include sedimentation, scaling, obstacles, residual wall / dam root, tree roots, and scum; the grade is distinguished according to the severity of the pipeline defects. The first grade represents minor defects, the second grade represents medium defects, the third grade represents serious defects, and the fourth grade represents major defects. The severity of the defects increases continuously with the increase of the grade; the circumferential position where the pipeline defect is located is represented by the method of a clock. The first two digits represent the starting position from which point, and the last two digits represent the ending position to which point (exact hour). If the defect is at a certain point, the first two digits are replaced by 00, and the last two digits represent the defect point position. For example, if the circumferential position of the pipeline defect is exactly the upper half of the pipeline, the corresponding circumferential position code is 0903; if the circumferential position of the pipeline defect is exactly a point at the top of the pipeline, the corresponding circumferential position code is 0012.

[0079] The water quality anomaly identification result is the access position.

[0080] The above is a specific description of the preferred embodiment of the present invention. However, the present invention is not limited to the described embodiment. Those skilled in the art can make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A rapid detection device for drainage pipes, characterized in that, It includes an underwater unmanned vehicle, a control system, an image acquisition device and a water quality detection device. The underwater unmanned vehicle is equipped with a control system, an image acquisition device and a water quality detection device. The image acquisition device and the water quality detection device are respectively connected to the control system. The control system includes a central control subsystem and a motion subsystem. The motion subsystem is connected to the central control subsystem. The image acquisition device includes a stable pan-tilt head, a lighting module and a camera. The stable pan-tilt head, the lighting module and the camera are respectively connected to the central control subsystem. The water quality detection device includes a probe and a driver. The probe is connected to the driver. The driver is connected to the central control subsystem; Among them, the central control subsystem includes a data transmission module, a navigation module, an image recognition module and a water quality analysis module. The navigation module, the image recognition module and the water quality analysis module are respectively connected to the data transmission module. The lighting module, the stable pan-tilt head and the motion subsystem are respectively connected to the navigation module. The camera is connected to the image recognition module. The water quality analysis module is connected to the driver; The detection method applied to the drainage pipeline rapid detection device specifically includes the following steps: Determine the detection range and the device placement location; Obtain the current image based on the image acquisition device and identify the liquid level in the detection pipe; Determine the detection mode according to the liquid level in the detection pipe; When the liquid level in the pipe is not higher than the minimum detection liquid level of the water quality detection device, the device switches to the ground detection mode; when the liquid level in the pipe is higher than the position where the camera is located, the device switches to the underwater detection mode; when the liquid level in the pipe is between the above two situations, the device switches to the water detection mode; According to the preset requirements of the detection mode, perform image acquisition and water quality index determination based on the image acquisition device and the water quality detection device; Based on the image recognition module, perform rapid image recognition on the data obtained from the image acquisition to obtain a rapid pipeline defect recognition result; Based on the water quality analysis module, perform water quality anomaly recognition on the data obtained from the water quality index determination to obtain a water quality anomaly recognition result; The water quality anomaly recognition model in the water quality analysis module is specifically a water quality anomaly recognition model constructed based on a pipeline experimental platform, including the following steps: Preset experimental conditions and simulate the water quality of the experimental pipeline; Measure the distribution of the conductivity of the water quality of the experimental pipeline to obtain the conductivity at different positions in the pipeline; Based on the conductivity at different positions in the pipeline, construct a conductivity distribution law, obtain the conductivity fluctuation range under different abnormal water quality conditions, and select appropriate upper and lower limit fluctuation values as the standard for the water quality analysis model to judge whether there is abnormal water quality access, and obtain the water quality anomaly recognition model; Based on the water quality anomaly recognition model, identify the actual abnormal pipeline water quality and perform error analysis on the recognition result to obtain the training error; Based on the training error, adjust the parameters of the water quality anomaly recognition model to obtain the trained water quality anomaly recognition model.

2. The rapid detection device for drainage pipelines according to claim 1, wherein The stable pan-tilt head is installed on the top of the underwater unmanned vehicle. The camera is installed on the stable pan-tilt head. The lighting module is installed in front of the camera.

3. The rapid detection device for drainage pipes according to claim 2, characterized in that, The driver is installed inside the underwater unmanned vehicle. The probe is installed at the bottom of the underwater unmanned vehicle.

4. The rapid detection device for a drainage pipeline according to claim 3, characterized in that, The motion subsystem includes a driving motor and wheels. The driving motor is connected to the wheels and is also connected to the navigation module.

5. The rapid detection device for drainage pipes according to claim 4, characterized in that, The wheels are installed on both sides of the bottom of the underwater unmanned vehicle, and the number of wheels is not less than four.

6. A rapid detection method for drainage pipelines, characterized in that, It includes the following steps: Determine the detection range and the device placement location; Obtain the current image based on the image acquisition device and identify the liquid level in the detection pipe; Determine the detection mode according to the liquid level in the detection pipe; When the liquid level in the pipe is not higher than the minimum detection liquid level of the water quality detection device, the device switches to the ground detection mode; when the liquid level in the pipe is higher than the position where the camera is located, the device switches to the underwater detection mode; When the liquid level in the pipe is between the above two situations, the device switches to the water detection mode; According to the preset requirements of the detection mode, perform image acquisition and water quality index determination based on the image acquisition device and the water quality detection device; Based on the image recognition module, quickly recognize the images of the data obtained from the image acquisition, and obtain the quick recognition result of pipeline defects; Based on the water quality analysis module, identify water quality anomalies for the data obtained from the water quality index determination, and obtain the water quality anomaly recognition result; The water quality anomaly recognition model in the water quality analysis module is specifically a water quality anomaly recognition model constructed based on the pipeline experimental platform, including the following steps: Preset the experimental conditions and simulate the water quality of the experimental pipeline; Measure the distribution of the conductivity of the water quality of the experimental pipeline to obtain the conductivity at different positions in the pipeline; Based on the conductivity at different positions in the pipeline, construct the conductivity distribution law, obtain the conductivity fluctuation range under different abnormal water quality conditions, and select appropriate upper and lower limit fluctuation values as the standard for the water quality analysis model to judge whether there is abnormal water quality access, and obtain the water quality anomaly recognition model; Based on the water quality anomaly recognition model, identify the actual abnormal pipeline water quality, and perform error analysis on the recognition result to obtain the training error; Based on the training error, adjust the parameters of the water quality anomaly recognition model to obtain the trained water quality anomaly recognition model.

7. The rapid detection method for a drainage pipeline according to claim 6, wherein It also includes: Combine the quick recognition result of pipeline defects and the water quality anomaly recognition result to obtain the quick pipeline detection result; Output the quick pipeline detection result and conduct manual review of some defective pipelines and manual tracing of water quality anomalies to obtain the review result; Based on the review result, adjust the parameters of the image quick classifier in the image recognition module and the parameters of the water quality anomaly recognition model in the water quality analysis module.

8. The rapid detection method for a drainage pipe according to claim 7, characterized in that The construction method of the image quick classifier in the image recognition module includes the following steps: Select the sample images of defective pipelines; Extract the invariant moment features of the defect types and grades of the sample images of defective pipelines to obtain the invariant moment features; Normalize the input vector of the invariant moment features and convert the invariant moment features to the range between [0, 1]; Perform fuzzy processing on the invariant moment features and classify the fuzzy processed invariant moment features based on the neural network to obtain the predicted classification result; According to the predicted classification result and the expected target output corresponding to the invariant moment feature, obtain the training error; Based on the training error, adjust the parameters of the image quick classifier to obtain the trained image quick classifier.

Images