A pipeline inspection system and method

Abstract

The application provides a pipeline inspection system and method. The system comprises a control end, an inspection end and a backend. The backend generates and sends a preset inspection instruction to the control end. The preset inspection instruction at least comprises pipeline information of multiple segments of preset pipelines in a to-be-inspected area. The control end controls the inspection end to move in the to-be-inspected area based on the received preset inspection instruction, so that the inspection end collects a radar time sequence image corresponding to each segment of the preset pipeline. The backend acquires the radar time sequence image corresponding to each segment of the preset pipeline and inputs a pre-trained pipeline fault identification model to acquire an identification result output by the pipeline fault identification model. The backend determines the preset pipeline that needs to be repaired based on the identification result and forms a corresponding repair task, so as to improve the efficiency of pipeline fault troubleshooting and reduce the inspection cost.

Description

Technical Field

[0001] This application relates to the field of urban construction technology, and more specifically, to a pipeline inspection system and method. Background Technology

[0002] Underground pipeline management is one of the important tasks to ensure the normal operation of urban infrastructure. However, due to various reasons, underground pipelines may malfunction, affecting the normal operation of the city. In particular, the underground pipeline networks in old urban areas have been used for a long time and the materials are seriously aged, making them frequent areas for pipeline failures.

[0003] Existing methods for troubleshooting pipeline faults mostly involve regular manual inspections and maintenance. However, this approach is not only wasteful of manpower, but also ineffective because it cannot be done by digging into the ground to detect faults. Summary of the Invention

[0004] The purpose of this application is to provide a pipeline inspection system and method to improve the efficiency of pipeline fault diagnosis and reduce inspection costs.

[0005] In a first aspect, the present invention provides a pipeline inspection system, comprising a control terminal, an inspection terminal, and a back-end. The back-end generates preset inspection instructions and sends them to the control terminal. The preset inspection instructions include at least pipeline information for multiple preset pipeline segments within the inspection area. Based on the received preset inspection instructions, the control terminal controls the inspection terminal to move within the inspection area, enabling the inspection terminal to acquire radar time-series images corresponding to each preset pipeline segment. The back-end acquires the radar time-series images corresponding to each preset pipeline segment and inputs them into a pre-trained pipeline fault identification model to obtain the identification results output by the pipeline fault identification model. Based on the identification results, the back-end determines the preset pipelines requiring repair and generates corresponding repair tasks. Specifically, for each preset pipeline segment… For pipelines, the backend obtains the recognition results output by the pipeline fault recognition model through the following steps: For all radar time-series images corresponding to the preset pipeline, select the first radar time-series image and the second radar time-series image; perform wavelet transform on the first radar time-series image and the second radar time-series image to obtain the first reconstructed time-series image and the second reconstructed time-series image, respectively; determine the difference image based on the difference between the first reconstructed time-series image and the second reconstructed time-series image; extract the spatial features of the difference image, the first reconstructed time-series image, and the second reconstructed time-series image through the feature extractor of the pipeline fault recognition model; and identify the spatial features through the classifier of the pipeline fault recognition model to output the corresponding pipeline fault recognition result.

[0006] In an optional implementation, the pipeline information includes at least a category identifier for each pre-defined pipeline segment. The category identifier is used to distinguish the geometric shapes between pipelines. The backend is also used to determine the corresponding pipeline fault identification model based on the pre-defined pipeline category identifier.

[0007] In an optional implementation, for each radar time-series image, the backend converts the radar time-series image into a reconstructed time-series image in the following manner: the radar time-series image is converted to grayscale; the grayscale-converted radar time-series image is subjected to two-dimensional discrete wavelet transform to obtain the corresponding approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix; and reconstruction is performed based on the obtained approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix to obtain the reconstructed time-series image.

[0008] In an optional implementation, the backend constructs a wavelet coefficient structure based on the obtained approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix; the backend performs an inverse wavelet transform on the wavelet coefficient structure to obtain the corresponding reconstructed image.

[0009] In an optional implementation, before the backend constructs the wavelet coefficient structure, it further includes setting the values ​​of the approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix that are less than a preset value to zero.

[0010] In an optional implementation, the backend obtains the approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix in the following manner:

[0011]

[0012] Among them, A k , m The value in the k-th column and m-th row of the approximation coefficient matrix. The value in the k-th column and m-th row of the horizontal detail coefficient matrix. This represents the value in the k-th column and m-th row of the vertical detail coefficient matrix. The value in the k-th column and m-th row of the diagonal detail coefficient matrix. These are high-pass and low-pass filters, respectively, P l , n This represents the grayscale value in the k-th column and m-th row of the radar time-series image.

[0013] In an optional implementation, the first radar time-series image and the second radar time-series image are two radar time-series images that are spaced apart or adjacent. The identification result output by the pipeline fault identification model is used to indicate the pipeline defect type and defect level. The backend forms the corresponding maintenance task in the following way: it determines whether the preset pipeline needs maintenance based on the pipeline defect type and defect level indicated by the identification result; if so, it forms the maintenance task based on the radar time-series image of the preset pipeline and the corresponding identification result.

[0014] In an optional implementation, the backend generates preset inspection instructions in the following manner:

[0015] The system responds to user clicks on target area controls via the inspection interface, identifying the area to be inspected; it also responds to user clicks on target pipe category controls via the inspection interface, identifying the category identifier of the preset pipe; based on the area to be inspected and the category identifier of the preset pipe, it searches the pipeline database for pipe information for multiple preset pipe segments, including the location data of each preset pipe segment; and based on the category identifier of the preset pipe and the corresponding pipe information, it generates a preset inspection instruction.

[0016] Secondly, the present invention provides a pipeline inspection method, the method comprising:

[0017] The process involves acquiring radar time-series images corresponding to each pre-defined pipeline segment and inputting them into a pre-trained pipeline fault identification model to obtain the identification results output by the model. Based on these results, the pre-defined pipelines requiring repair are identified, and corresponding repair tasks are generated. Specifically, for each pre-defined pipeline segment, the identification results output by the pipeline fault identification model are obtained through the following steps: For all radar time-series images corresponding to the pre-defined pipeline, a first radar time-series image and a second radar time-series image are selected; wavelet transform is performed on the first and second radar time-series images to obtain a first reconstructed time-series image and a second reconstructed time-series image, respectively; based on the difference between the first and second reconstructed time-series images, a difference image is determined; spatial features of the difference image, the first reconstructed time-series image, and the second reconstructed time-series image are extracted using the feature extractor of the pipeline fault identification model; and the spatial features are identified using the classifier of the pipeline fault identification model to output the corresponding pipeline fault identification result.

[0018] In an optional implementation, for each radar time-series image, the backend converts the radar time-series image into a reconstructed time-series image in the following manner:

[0019] The radar time-series image is converted to grayscale; the grayscale-converted radar time-series image is then subjected to two-dimensional discrete wavelet transform to obtain the corresponding approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix; based on the obtained approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix, a reconstruction is performed to obtain the reconstructed time-series image.

[0020] This application provides a pipeline inspection system and method, wherein the backend generates preset inspection instructions and sends them to the control end; based on the received preset inspection instructions, the control end controls the inspection end to move within the inspection area, enabling the inspection end to acquire radar time-series images corresponding to each preset pipeline segment; the backend acquires the radar time-series images corresponding to each preset pipeline segment and inputs them into a pre-trained pipeline fault identification model to obtain the identification results output by the pipeline fault identification model; based on the identification results, the backend determines the preset pipelines that need repair and forms corresponding repair tasks; wherein, for each preset pipeline segment, the backend obtains the pipeline fault identification model output through the following steps. The identification results are as follows: For all radar time-series images corresponding to the preset pipeline, the first radar time-series image and the second radar time-series image are selected; wavelet transform is performed on the first radar time-series image and the second radar time-series image to obtain the first reconstructed time-series image and the second reconstructed time-series image, respectively; based on the difference between the first reconstructed time-series image and the second reconstructed time-series image, the difference image is determined; the spatial features of the difference image, the first reconstructed time-series image, and the second reconstructed time-series image are extracted by the feature extractor of the pipeline fault identification model; the spatial features are identified by the classifier of the pipeline fault identification model to output the corresponding pipeline fault identification result. The pipeline inspection system provided in this application uses radar scanning to achieve non-contact pipeline inspection and combines AI model to achieve intelligent identification of pipeline faults and defect levels. The inspection process is highly automated, avoiding ground excavation and manual inspection, which not only saves manpower and resources, but also improves inspection efficiency and reduces inspection costs. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of a pipeline inspection system provided in an embodiment of this application.

[0023] Icon labels:

[0024] Backend -10, Control End -20, Inspection End -30. Detailed Implementation

[0025] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0026] Figure 1This is a schematic diagram of a pipeline inspection system provided in an embodiment of this application. The pipeline inspection system provided in this embodiment includes a control end, an inspection end, and a back-end.

[0027] The inspection and control terminals can be deployed in the area to be inspected. The control terminal can be a mobile electronic device such as a mobile phone or remote control, or a non-mobile electronic device such as a computer used by the inspection personnel. The inspection terminal includes a data acquisition module, a positioning module, and a processing module. The data acquisition module can be a drone or robot equipped with low-resolution radar. The positioning module can be GPS, etc. The processing module is used to respond to commands from the control terminal, enabling the movement and data acquisition control of the inspection terminal.

[0028] The backend can be deployed in a fixed area, such as a server or computer. The backend includes a database to store official website information for each city / district. It also includes pipeline fault identification models for different pipeline types. These models can be generated based on ConvLSTM training. In a specific embodiment, the pipelines are classified according to their geometric shape, which may include tubular, hollow, and layered structures.

[0029] This application also provides an application and a visual interactive interface for a pipeline inspection system.

[0030] Users can select the area to be inspected and the type of pipeline through the inspection interface provided by the control terminal or the backend to complete the issuance of inspection tasks.

[0031] Specifically, the backend or control terminal can generate preset inspection instructions in the following ways:

[0032] The system responds to user clicks on target area controls within the inspection interface, identifying the area to be inspected. It also responds to user clicks on target pipe category controls within the inspection interface, determining the category identifier of the preset pipe. Based on the area to be inspected and the category identifier of the preset pipe, it searches the pipeline database for pipe information for multiple preset pipe segments, including the location data of each segment. Finally, based on the category identifier and corresponding pipe information, it generates a preset inspection command.

[0033] The inspection interface provides a map display area. Each city can be divided into urban districts, displayed as area controls. Users can zoom in and out of the map and select the target area control. The inspection interface also displays pipe category controls corresponding to the pipe category. Users can select one or more pipe categories as needed.

[0034] The backend can directly search the pipeline database for pipeline information for each pre-defined pipeline segment within the identified inspection area and pipeline category identifier. This pipeline information can include pipeline location data, material, diameter, category, etc. Location data can specifically include the coordinates of the pipeline centerline stakes and weld seams, the elevation data of the pipeline's location, and the pipeline's burial depth underground, etc.

[0035] The backend generates a preset inspection command and sends it to the control terminal. The preset inspection command includes at least the pipeline information of multiple preset pipelines in the area to be inspected.

[0036] Based on the received preset inspection instructions, the control terminal controls the inspection terminal to move within the inspection area, enabling the inspection terminal to collect radar time-series images corresponding to each preset pipeline segment.

[0037] For each pre-set pipeline segment, the inspection terminal can move along the centerline of the pipeline and collect corresponding radar timing images at a specified frequency.

[0038] The radar here operates in B-can scanning mode.

[0039] The backend acquires radar time-series images corresponding to each pre-defined pipeline segment and inputs them into a pre-trained pipeline fault identification model to obtain the identification results output by the model. Specifically, the backend can determine the corresponding pipeline fault identification model based on the pre-defined pipeline category identifier to obtain the identification results. The identification results output by the pipeline fault identification model are used to indicate the pipeline defect type and defect level.

[0040] For each pre-defined pipe segment, the backend can obtain the identification results output by the pipe fault identification model through the following steps:

[0041] For all radar time-series images corresponding to the preset pipeline, the first and second radar time-series images are selected. Wavelet transforms are performed on the first and second radar time-series images to obtain the first and second reconstructed time-series images, respectively. Based on the difference between the first and second reconstructed time-series images, a difference image is determined. Spatial features of the difference image, the first reconstructed time-series image, and the second reconstructed time-series image are extracted using the feature extractor of the pipeline fault identification model. The spatial features are then identified using the classifier of the pipeline fault identification model to output the corresponding pipeline fault identification result.

[0042] Based on the recognition results, the backend identifies the preset pipes that need repair and generates the corresponding repair tasks.

[0043] Specifically, the backend can generate corresponding maintenance tasks in the following ways:

[0044] Based on the pipeline defect type and defect level indicated by the identification results, it is determined whether the preset pipeline needs maintenance. If so, a maintenance task is formed based on the radar time-series image of the preset pipeline and the corresponding identification results.

[0045] Here, pipeline defect types can include structural defects such as ruptures, deformations, misalignments, disconnections, and leaks, as well as functional defects such as scaling and obstructions. Furthermore, different pipeline defect types can correspond to different defect levels, which can be set according to the urgency of the required repairs. For example, defect levels can be divided into three levels, from minor to severe.

[0046] Furthermore, for each type of pipeline defect, the need for repair is determined based on its defect level. For example, for leakage defects, repair is required if the defect level is Level 1. For scaling defects, repair is only required if the defect level is Level 3.

[0047] The backend can assign maintenance tasks to the corresponding maintenance personnel to ensure timely maintenance of the pipelines.

[0048] This application provides a pipeline inspection system that uses radar scanning to achieve non-contact pipeline inspection and combines it with AI models to intelligently identify pipeline faults and defect levels. The inspection process is highly automated, avoiding ground excavation and manual inspection. It can also automatically generate and issue maintenance tasks, which not only saves manpower and resources but also improves inspection efficiency and reduces inspection costs.

[0049] In one embodiment of this application, an AI model is proposed to identify radar time-series images of pipelines to determine whether the pipeline has defects and the level of defects.

[0050] Regarding the selection of the AI ​​model, this embodiment chose Convolutional Long Short-Term Memory (ConvLSTM) as the initial model. ConvLSTM combines the advantages of convolution operations and LSTM, enabling it to extract features spatially and model dependencies temporally.

[0051] The training process of the pipeline fault identification model will be explained below using a tubular pipeline as an example.

[0052] I. Sample Data Collection

[0053] A large number of radar time-series images of tubular, normal, and defective pipes were acquired. These radar time-series images were obtained from ground-penetrating radar (GPR) scans in B-CAN mode. For pipes with smaller diameters (less than 50 cm), an antenna of 400-900 MHz can be used. For pipes with larger diameters, an antenna of 200-400 MHz can be used.

[0054] II. Preprocessing of Sample Data

[0055] The radar time-series images in the sample are preprocessed with denoising, grayscale conversion, etc.

[0056] For each radar time-series image, a wavelet transform is performed to obtain the corresponding reconstructed time-series image. For each channel, based on two adjacent or one-step-series reconstructed time-series images corresponding to that channel, the difference image corresponding to that channel is determined.

[0057] Construct a sample set, in which each sample includes a first reconstructed time series image, a second reconstructed time series image, and a difference image corresponding to a pipeline.

[0058] Specifically, for each radar time-series image, the radar time-series image can be converted into a reconstructed time-series image in the following way:

[0059] Convert the radar time-series images to grayscale.

[0060] The radar time-series image after grayscale conversion is subjected to two-dimensional discrete wavelet transform to obtain the corresponding approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix and diagonal detail coefficient matrix.

[0061] The approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix can be obtained in the following ways:

[0062]

[0063] Among them, A k , m The value in the k-th column and m-th row of the approximation coefficient matrix. The value in the k-th column and m-th row of the horizontal detail coefficient matrix. This represents the value in the k-th column and m-th row of the vertical detail coefficient matrix. The value in the k-th column and m-th row of the diagonal detail coefficient matrix. These are high-pass and low-pass filters, respectively, P l , n This represents the grayscale value in the k-th column and m-th row of the radar time-series image.

[0064] In one specific embodiment, the wavelet basis functions here can be the Daubechies 4 wavelet basis functions. Then, the `pywt.dwt2` function is used to perform a two-dimensional discrete wavelet transform to obtain the approximate coefficient matrix (cA), the horizontal detail coefficient matrix (cH), the vertical detail coefficient matrix (cV), and the diagonal detail coefficient matrix (cD).

[0065] Specifically, before constructing the wavelet coefficient structure, the values ​​of the approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix that are less than a preset value are set to zero.

[0066] This removes potentially noisy data from the image, thereby ensuring the quality of the reconstructed time-series image and guaranteeing accurate recognition.

[0067] Reconstruction is performed based on the obtained approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix to obtain a reconstructed time-series image.

[0068] Next, a wavelet coefficient structure can be constructed based on the obtained approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix. An inverse wavelet transform is then performed on the wavelet coefficient structure to obtain the corresponding reconstructed image.

[0069] Here, a two-dimensional discrete wavelet inverse transform can be used to upsample the approximation coefficient matrix, horizontal detail coefficient matrix, vertical detail coefficient matrix, and diagonal detail coefficient matrix respectively. The upsampled results are then subjected to low-pass and high-pass filtering, and finally, the results are summed to obtain the wavelet coefficient structure, which is then converted into a reconstructed image.

[0070] The reconstructed image can also be cropped or normalized to ensure that it is consistent with the size of the radar time-series image.

[0071] The difference image here can be selected by subtracting the first reconstructed time series image with the first timestamp from the second reconstructed time series image with the second timestamp.

[0072] Images in the sample set are labeled with their corresponding defect types and defect levels. The set is then divided into a training set (70%), a validation set (15%), and a test set (15%). Each set includes a first reconstructed temporal image, a second reconstructed temporal image, and a corresponding difference image.

[0073] III. Model Training

[0074] The ConvLSTM model here can include an input layer, an input gate, a forget gate, an output gate, cell states, and an output layer.

[0075] The input layer takes three-dimensional tensors as input data, specifically including the length values ​​of two-dimensional images and time series.

[0076] The input gate updates the current cell state based on the current input and the cell state from the previous time step using convolution operations and the sigmoid activation function. Input gate i t It can be represented as:

[0077] i t =σg (W xi *X t +W hi *H t-1 +W ci ℃ t-1 +b i );

[0078] Where, σ g X is the sigmoid activation function. t H is the sample input for the current time step. t-1 C is the hidden state from the previous time step. t-1 W is the memory unit of the previous time step. xi W hi W ci b i These are the model parameters.

[0079] The forget gate uses convolution operations and a sigmoid activation function to calculate and determine which information needs to be forgotten based on the current input and the cell state at the previous time step. The forget gate f t It can be represented as:

[0080] f t =σ g (W xf *X t +W hf *H t-1 +W cf ℃ t-1 +b f );

[0081] Among them, W xf W hf W cf b f These are the model parameters.

[0082] The output gate determines the output at the current time step using convolution operations and the tanh activation function, based on the current input, the cell state at the previous time step, and the input gate. Output gate o t It can be represented as:

[0083] o t =σ g (W xo *X t +W ho *H t-1 +W co ℃ t +b o );

[0084] Among them, W xo W ho Wco b o These are the model parameters.

[0085] The cell state is updated and adjusted based on the calculation results of the input gate, forget gate, and output gate. The current cell state C... t It can be represented as:

[0086] C t =f t ℃ t-1 +i t °tanh g (W xc *X t +W hc *H t-1 +b c );

[0087] Among them, W xc W hc b c Here are the model parameters, * for convolution, ° for Hadamard product, and tanh. g It is the hyperbolic tangent activation function.

[0088] The output layer uses the tanh activation function to transform the cell state into the output result. Output layer H t It can be represented as:

[0089] H t =o t °tanh g (C t ).

[0090] At each time step, spatial information of the temporal data can be captured through convolution operations, and the convolution kernels at each time step are shared, ensuring the consistency of spatial feature extraction.

[0091] The ConvLSTM model is trained using samples from the labeled training set. The model is then tested using samples from the test set to obtain the corresponding classification accuracy evaluation results. Model parameters such as kernel size, stride, number of LSTM units, and number of layers are adjusted accordingly. Furthermore, the model is optimized using regularized loss functions, resulting in a well-trained pipeline fault recognition model.

[0092] In one embodiment of this application, a pipeline inspection method is also provided, which can be implemented through a backend, including: acquiring radar time-series images corresponding to each preset pipeline segment and inputting them into a pre-trained pipeline fault identification model to obtain the identification results output by the pipeline fault identification model; based on the identification results, determining the preset pipelines that need to be repaired and forming corresponding repair tasks; wherein, for each preset pipeline segment, the identification results output by the pipeline fault identification model are obtained through the following steps: for all radar time-series images corresponding to the preset pipeline, selecting a first radar time-series image and a second radar time-series image; performing wavelet transform on the first radar time-series image and the second radar time-series image to obtain a first reconstructed time-series image and a second reconstructed time-series image, respectively; determining a difference image based on the difference between the first reconstructed time-series image and the second reconstructed time-series image; extracting spatial features of the difference image, the first reconstructed time-series image, and the second reconstructed time-series image through the feature extractor of the pipeline fault identification model; and identifying the spatial features through the classifier of the pipeline fault identification model to output the corresponding pipeline fault identification result.

[0093] The technical effects of the above method embodiments have been demonstrated in the system embodiments, and will not be repeated here.

[0094] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0095] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0096] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0097] It should be noted that if the function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0098] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0099] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A pipeline inspection system characterized by, The system comprises a control terminal, an inspection terminal and a backend, wherein, The backend generates a preset inspection instruction and sends it to the control terminal, the preset inspection instruction at least comprising pipe information of multiple preset pipes in an underground area to be inspected, the pipe information at least comprising position coordinates and category identifiers of the pipes; The control terminal controls the inspection terminal to move along the center line of each preset pipe in the area to be inspected based on the received preset inspection instruction, so that the inspection terminal collects radar time sequence images corresponding to each preset pipe; The backend obtains the radar time sequence images corresponding to each preset pipe and inputs them into a pre-trained pipe fault identification model to obtain an identification result output by the pipe fault identification model; The backend determines preset pipes that need to be repaired based on the identification result and forms corresponding repair tasks; For each preset pipe, the backend obtains the identification result output by the pipe fault identification model through the following steps: For all radar time sequence images corresponding to the preset pipe, first and second radar time sequence images are selected; Wavelet transformation is performed on the first and second radar time sequence images to obtain first and second reconstructed time sequence images, respectively, the first and second radar time sequence images being two radar time sequence images that are spaced apart or adjacent; A difference image is determined based on the difference between the first and second reconstructed time sequence images; Spatial features of the difference image, the first reconstructed time sequence image and the second reconstructed time sequence image are extracted by a feature extractor of the pipe fault identification model corresponding to the category identifier of the preset pipe; The spatial features are identified by a classifier of the pipe fault identification model to output a corresponding pipe fault identification result, the identification result output by the pipe fault identification model being used to indicate a pipe defect type and a defect level.

2. The system of claim 1, wherein, The pipe information at least comprises a category identifier of each preset pipe, the category identifier being used to distinguish the geometric shapes of the pipes, and the backend is further configured to determine a corresponding pipe fault identification model based on the category identifier of the preset pipe.

3. The system of claim 1, wherein, For each radar time sequence image, the backend converts the radar time sequence image into a reconstructed time sequence image in the following manner: The radar time sequence image is converted to grayscale; The radar time sequence image after grayscale conversion is subjected to two-dimensional discrete wavelet transformation to obtain an approximation coefficient matrix, a horizontal detail coefficient matrix, a vertical detail coefficient matrix and a diagonal detail coefficient matrix; The approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix are reconstructed to obtain a reconstructed time sequence image.

4. The system of claim 3, wherein, The backend constructs a wavelet coefficient structure based on the obtained approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix; The backend performs inverse wavelet transformation on the wavelet coefficient structure to obtain a corresponding reconstructed image.

5. The system of claim 4, wherein, Before the backend constructs the wavelet coefficient structure, it also includes setting to zero the values in the approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix that are less than a preset value.

6. The system of claim 3, wherein, The backend obtains the approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix in the following manner: ； ； ； ； in, The approximation coefficient matrix is ​​the first... The value in the m-th row of the column. For the horizontal detail coefficient matrix, the first The value in the m-th row of the column. For the vertical detail coefficient matrix, the first The value in the m-th row of the column. For the diagonal detail coefficient matrix, the first... The value in the m-th row of the column. , These are high-pass and low-pass filters, respectively. For radar time-series images, the first The grayscale value in the m-th row of column m.

7. The system of claim 1, wherein, The backend forms the corresponding maintenance task in the following manner: Determine whether the preset pipeline needs to be maintained based on the pipeline defect type and the defect level indicated by the identification result; If so, form a maintenance task based on the radar time sequence image of the preset pipeline and the corresponding identification result.

8. The system of claim 2, wherein, The backend generates the preset inspection instruction in the following manner: Determine the area to be inspected in response to the user's click operation on the target area control through the inspection interface; Determine the category identifier of the preset pipeline in response to the user's click operation on the target pipeline category control through the inspection interface; Search for the pipeline information of the corresponding multiple segments of the preset pipeline from the pipe network database based on the area to be inspected and the category identifier of the preset pipeline, the pipeline information further including the position data of each segment of the preset pipeline; Generate the preset inspection instruction based on the category identifier of the preset pipeline and the corresponding pipeline information.

9. A method of inspecting a pipeline, the method comprising: The method is suitable for the pipeline inspection system of any one of claims 1-8, and the method comprises: Obtain the radar time sequence image corresponding to each segment of the preset pipeline, and input the pre-trained pipeline fault identification model to obtain the identification result output by the pipeline fault identification model; Determine the preset pipeline that needs to be maintained and form the corresponding maintenance task based on the identification result; For each segment of the preset pipeline, the identification result output by the pipeline fault identification model is obtained in the following steps: Select the first radar time sequence image and the second radar time sequence image from all radar time sequence images corresponding to the preset pipeline; Perform wavelet transform on the first radar time sequence image and the second radar time sequence image to obtain the first reconstructed time sequence image and the second reconstructed time sequence image, respectively; Determine the difference image based on the difference between the first reconstructed time sequence image and the second reconstructed time sequence image; Extract the spatial features of the difference image, the first reconstructed time sequence image and the second reconstructed time sequence image through the feature extractor of the pipeline fault identification model; Identify the spatial features through the classifier of the pipeline fault identification model to output the corresponding pipeline fault identification result.

10. The method of claim 9, wherein, For each radar time sequence image, the backend converts the radar time sequence image into a reconstructed time sequence image in the following manner: Perform gray scale conversion on the radar time sequence image; Perform two-dimensional discrete wavelet transform on the radar time sequence image after gray scale conversion to obtain the approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix; Reconstruct based on the obtained approximation coefficient matrix, the horizontal detail coefficient matrix, the vertical detail coefficient matrix and the diagonal detail coefficient matrix to obtain the reconstructed time sequence image.

Images