Pipeline deformation image noise reduction method

Abstract

The invention provides a pipeline deformation image noise reduction method, and relates to the technical field of pipeline deformation measurement. The method comprises the following steps: making and pasting a mark target, calibrating image noise, constructing a reference image set and a noise image set, constructing and training a digital image noise model, and outputting a target pipeline noise reduction image through the trained digital image noise model. According to the method, the corresponding pipeline deformation images in different noise environments are identified through the mark target information, and the digital image noise model capable of automatically performing noise reduction processing on the pipeline deformation images is constructed, so that accurate deformation calculation of the pipeline deformation images in different environment states is efficiently and accurately realized; the method can be applied to optical deformation measurement in a complex illumination environment and deformation measurement under other photoelectric noise interference.

Description

Technical Field

[0001] This invention mainly relates to the field of pipeline deformation measurement technology, and specifically to a method for denoising pipeline deformation images. Background Technology

[0002] Piping systems are a crucial component of industrial systems, widely used in energy, chemical, and other fields. Current inspection methods primarily rely on traditional approaches such as ultrasound and contact sensors to assess the health status of pipelines. However, for complex pipeline systems consisting of multiple pipes, inspections are often conducted using vehicles, robots, or manual labor, resulting in low efficiency and failing to meet today's demands for rapid and efficient inspections.

[0003] Existing optical non-contact measurement methods can obtain deformation information of multiple pipelines within the measured area. However, in actual engineering processes, due to ambient light interference and other photoelectric noise interference, the acquired images often contain image noise, resulting in low measurement accuracy. This fails to meet the actual engineering requirements and seriously restricts the development of pipeline safety requirements in the domestic energy and chemical industry. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for denoising pipeline deformation images to address the shortcomings of the prior art.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A method for denoising pipeline deformation images, comprising the following steps:

[0006] S1. Make multiple marker targets for identifying different pipelines under test, attach the multiple marker targets to the surface of different pipelines under test and take pictures to obtain multiple original images of the pipelines under test containing the marker targets.

[0007] S2. Measure the distance between each pipeline under test using a distance measuring tool, and measure the strain data of each pipeline under test using strain gauges pre-set on each pipeline under test containing a target marker, to obtain the deformation distance data and deformation strain data of each pipeline under test, and take pictures of each pipeline under test after deformation according to different combinations of shooting parameters, to obtain deformation images of the pipeline under test corresponding to different combinations of shooting parameters.

[0008] S3. Based on all the original images of the pipeline under test and their corresponding deformation images, strain calculation is performed. The deformation spacing data and deformation strain data corresponding to the calculated deformation images of the pipeline under test are compared with the deformation spacing data and deformation strain data corresponding to the pipeline under test measured in the same pipeline under test. From the comparison results, the deformation images of the pipeline under test that are similar to the measured data are selected as reference images. A reference image set is constructed through multiple reference images. The remaining deformation images of the pipeline under test are used as noise images. A noise image set is constructed through multiple noise images.

[0009] S4. Construct a digital image noise model based on a neural network, and train the digital image noise model using the reference image set and the noise image set to obtain the trained digital image noise model;

[0010] S5. The imported target pipeline image is denoised using the trained digital image noise model, and the denoised target pipeline image is output.

[0011] The beneficial effects of this invention are: by identifying pipeline deformation images under different noise environments through target information, and constructing a digital image noise model that can automatically denoise pipeline deformation images, it can efficiently and accurately calculate the deformation of pipeline deformation images under different environmental conditions. It is suitable for optical deformation measurement under complex lighting conditions and deformation measurement under other photoelectric noise interference. Attached Figure Description

[0012] Figure 1 A flowchart of a pipeline deformation image noise reduction method provided in an embodiment of the present invention;

[0013] Figure 2 This is a schematic diagram of a marker target provided in an embodiment of the present invention. Detailed Implementation

[0014] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0015] In many industrial sectors, such as petrochemicals and natural gas transportation, pipelines may carry high-pressure, flammable, explosive, or toxic media. Pipeline deformation can lead to leaks, causing serious accidents such as fires, explosions, or environmental pollution. Therefore, pipeline deformation detection is crucial. However, current optical non-contact measurement methods are subject to interference from ambient light and other photoelectric noise, resulting in image noise in the acquired images and thus low measurement accuracy, failing to meet practical engineering requirements.

[0016] Based on the above-mentioned defects, such as Figure 1As shown in the figure, this embodiment of the invention provides a method for denoising pipeline deformation images, including the following steps:

[0017] S1. Make multiple marker targets for identifying different pipelines under test, attach the multiple marker targets to the surface of different pipelines under test and take pictures to obtain multiple original images of the pipelines under test containing the marker targets.

[0018] S2. Measure the distance between each pipeline under test using a distance measuring tool, and measure the strain data of each pipeline under test using strain gauges pre-set on each pipeline under test containing a target marker, to obtain the deformation distance data and deformation strain data of each pipeline under test, and take pictures of each pipeline under test after deformation according to different combinations of shooting parameters, to obtain deformation images of the pipeline under test corresponding to different combinations of shooting parameters.

[0019] S3. Based on all the original images of the pipeline under test and their corresponding deformation images, strain calculation is performed. The deformation spacing data and deformation strain data corresponding to the calculated deformation images of the pipeline under test are compared with the deformation spacing data and deformation strain data corresponding to the pipeline under test measured in the same pipeline under test. From the comparison results, the deformation images of the pipeline under test that are similar to the measured data are selected as reference images. A reference image set is constructed through multiple reference images. The remaining deformation images of the pipeline under test are used as noise images. A noise image set is constructed through multiple noise images.

[0020] S4. Construct a digital image noise model based on a neural network, and train the digital image noise model using the reference image set and the noise image set to obtain the trained digital image noise model;

[0021] S5. The imported target pipeline image is denoised using the trained digital image noise model, and the denoised target pipeline image is output.

[0022] It should be understood that in S1, the process of making a marker target for identifying different pipelines under test includes: arranging two different colored strips in a preset order along the horizontal axial direction of the pipeline under test to obtain the marker target.

[0023] Specifically, such as Figure 2 As shown, a 10-bit target is used, where black bars represent 0 and white bars represent 1. Figure 2 The target is 0101010101, which is the target number 341. The target number at each position is a characteristic of the target stripe. This embodiment is not limited to black or white, but can also use other different color combinations.

[0024] Affix marker targets. Place coded targets on and around the surface of the pipeline system to be tested. The target coding parameters are related to the measurement area, distance, and depth of field. The marker targets should be as directly facing the camera as possible, and their size and measurement requirements should be closely adhered to the marker points. The marker targets used are coded along the horizontal axis. The marker targets serve as the basis for image noise assessment and are used for subsequent calculation of digital noise image levels.

[0025] In S2, image noise calibration uses pipe spacing and pipe strain as the comparison items for noise calibration. The spacing of key pipes is measured using laser ranging or other ranging tools. At the same time, strain gauges are attached to the main deformed pipes to monitor strain data. The spacing and strain data are recorded as calibration reference items. The spacing and strain data are recorded under different pipe conditions. Different shooting parameters are adjusted using the station monitoring camera. The position of each image should be the same as that during actual deformation monitoring to ensure that the image can be effectively compared and analyzed with the deformation monitoring image information in the later stage.

[0026] The spacing of critical pipelines is measured using a ranging tool; in this embodiment, a laser ranging device is used to measure the spacing of critical pipelines. Strain gauges are used to measure the strain data of the critical pipelines. Preferably, obtaining the deformation spacing data and deformation strain data corresponding to each pipeline under test further includes:

[0027] The distance between each pipeline under test is measured using a ranging tool according to a set number of times. Strain data for each pipeline under test is also measured using strain gauges pre-installed on each pipeline with a marked target, according to a set number of times. This yields multiple initial deformation distance data and multiple deformation strain data for each pipeline under test. The average of the initial deformation distance data is then calculated to obtain the deformation distance data for each pipeline under test. Similarly, the average of the multiple deformation strain data is calculated to obtain the deformation strain data for each pipeline under test. For example, data should be collected at least three times and averaged, and the standard deviation of the measurement should meet the deformation measurement requirements.

[0028] The station monitoring camera was used to take pictures by adjusting different shooting parameters. In this embodiment, the camera's focal length, exposure, gain, etc. were all adjusted, and images were taken for each different combination of parameters. The parameter setting information and its corresponding images were recorded.

[0029] Use internal pressure to deform the pipeline, and record the deformation spacing and deformation strain after deformation.

[0030] Continue to adjust the camera parameters to capture distorted images, and record the parameter settings information and corresponding images for different parameter combinations;

[0031] Deformation calculations were performed using various before-and-after deformation images, and the results were compared with distance measurement results and strain gauge measurement results.

[0032] Preferably, the different shooting parameter combinations include combinations of different parameters of focal length, exposure, and gain.

[0033] In S2, the specific operations for image noise calibration are as follows:

[0034] (a) Start the camera and adjust different shooting parameters to take multiple digital images. Each image should be taken at the same position as the actual deformation monitoring.

[0035] (b) Shooting parameters include, but are not limited to, shooting-related parameters such as focal length, aperture, exposure, shutter speed, gain, and ISO;

[0036] (c) When the camera captures images, a distance measuring and strain measuring sensor (i.e. strain gauge) should be used to simultaneously measure the pipe spacing and strain information in the measured area.

[0037] In S3, image recognition processing uses digital image correlation methods to calculate the digital images acquired under different shooting parameters to obtain the digital image measurement results of pipe spacing and strain. These results are then compared with the calibration reference to obtain the shooting parameters with the highest accuracy. The image acquired under these shooting parameters is defined as the reference image, i.e., the noise baseline image, while the other images of the pipe deformation to be measured are used as noise images.

[0038] Preferably, the step of comparing the deformation spacing data and deformation strain data corresponding to the calculated deformation image of the pipeline under test with the deformation spacing data and deformation strain data corresponding to the pipeline under test measured for the same pipeline under test, and selecting the deformation image of the pipeline under test that is similar to the measured data as a reference image from the comparison results, includes:

[0039] The deformation spacing data and deformation strain data corresponding to the calculated deformation images of the pipeline under test are respectively subtracted from the deformation spacing data and deformation strain data corresponding to the pipeline under test measured in the same pipeline under test. The absolute value of the difference is taken to obtain the absolute value of the difference corresponding to each deformation image of the pipeline under test. The deformation images of the pipeline under test corresponding to the absolute value of the difference corresponding to each deformation image of the pipeline under test are selected from the preset judgment threshold and used as reference images.

[0040] In S4, the digital image noise model is trained by scaling and partitioning the reference image to create a reference image set, creating a noise image set from the non-reference images, and constructing a digital image noise model based on a convolutional neural network. The model is then trained to establish an image noise processing model.

[0041] Preferably, the step of constructing a digital image noise model based on a convolutional neural network, and training the digital image noise model using the reference image set and the noisy image set to obtain a trained digital image noise model, includes:

[0042] A digital image noise model is constructed based on a neural network. The digital image noise model includes a three-layer neural network, with the first and second layers of the neural network respectively connected to the third layer of the neural network.

[0043] The reference image set is used as the input of the first layer neural network. The target images in each reference image of the reference image set are extracted through the convolution kernel function and linear transformation layer function of the first layer neural network. The target images in each reference image and their corresponding reference images are then input into the third layer neural network.

[0044] The noisy image set is used as the input of the second layer neural network. The target images in each noisy image in the noisy image set are extracted by the convolution kernel function and linear transformation layer function of the second layer neural network. The target images in each noisy image and their corresponding noisy images are then input into the third layer neural network.

[0045] The third-layer neural network is trained by using a noise filtering function to filter noise from the target images in each reference image and their corresponding reference images, as well as the target images in each noise image and their corresponding noise images, to obtain a trained digital image noise model.

[0046] The first and second neural networks each include a convolutional layer, a normalization layer, a linear transformation layer, and a pooling layer, while the third neural network includes a decoupling layer, a linear transformation layer, and a fully connected layer.

[0047] The specific process of training a digital image noise model is as follows:

[0048] (a) Reference image set creation: The reference image selected by S3, i.e. the noisy reference image, is expanded by image processing operations such as scaling and partitioning to obtain multiple image results. All images obtained from the reference image and its image processing are labeled as the reference image set and used as part of the input training set to be input into the convolutional neural network.

[0049] (b) Data from non-reference images are used as a set of noisy images and are also input into the convolutional neural network as part of the training set;

[0050] (c) The pooling layer of the digital image noise model distinguishes pooling based on the position of the target. The pooling degree in the target area is lower, while the pooling degree in the non-target area can be set relatively higher.

[0051] (d) Calculate the pipeline spacing and strain of each image after denoising in the noisy image set. Use the calculation deviation as the noise error of the image for the next round of convolutional neural network iteration. The iteration ends when the final calculation deviation is less than the error threshold. The corresponding convolutional neural network model is obtained, which is the digital noise model. The model association term is the noise parameter of the noise image marker target. Establish a digital image noise denoising function based on the model association term and train to obtain the digital image noise model.

[0052] The training requirements for the digital image noise model are as follows:

[0053] The number of layers in a digital image noise model is related to the full resolution of the input image. When the resolution is high, the number of identical first and second level layers can be increased to improve the accuracy of the neural network.

[0054] The separation of the target and the complete image at the input end is achieved by extracting the target and removing the remaining positional information to obtain an image with the same resolution but containing only the target information. This image and the original complete image are used as two layers of input data for the convolutional neural network.

[0055] The training results of the convolutional neural network are calculated using the convergence degree and the deformation calculation deviation after comparing the output image with the reference image. The training of the convolutional neural network is complete when the convergence degree meets the convergence threshold and the deformation calculation deviation is less than the calculation deviation threshold. In this embodiment, the convergence threshold is 0.02; the displacement deformation calculation deviation is 0.01 mm; and the strain deformation calculation deviation is 100 με.

[0056] In S5, image noise reduction processing, in actual production monitoring, uses the target marker as the noise evaluation input threshold. A noise processing model generates a noise filtering function for the current digital image, performing digital image noise reduction processing on the current image. The specific procedure is as follows:

[0057] (a) Capture the current deformation image; capture the current deformation image during the deformation and monitoring process;

[0058] (b) Determine the model association terms, process the standard target in the current deformed image, and obtain the noise parameters of the current digital image;

[0059] (c) Substitute the obtained current digital image noise parameters into the digital noise model, and obtain the corresponding noise reduction model function based on the digital image noise model;

[0060] (d) Denoise the current deformed image by performing convolution processing according to the denoising model function to remove image noise and obtain a denoised deformed image that can be used for deformation calculation and analysis.

[0061] In the above embodiments, the pipeline deformation images corresponding to different noise environments are identified by using target information, and a digital image noise model that can automatically denoise the pipeline deformation images is constructed. This enables efficient and accurate deformation calculation of pipeline deformation images under different environmental conditions, and is suitable for optical deformation measurement under complex lighting conditions and deformation measurement under other photoelectric noise interference.

[0062] Preferably, the strain calculation based on all the original images of the pipeline under test and their corresponding deformation images includes:

[0063] The strain is calculated using a strain mapping formula to obtain the deformation spacing data and deformation strain data corresponding to the deformation image of the pipeline under test, based on the original image of the pipeline under test and its corresponding deformation image. The strain mapping formula is as follows:

[0064]

[0065] Where, x i ′ Let x and y be the x-coordinates of the deformed image. i ′ Let d be the y-coordinate of the deformed image. x Let d be the gradient deformation of the pipeline under test in the x-direction. y Let represent the gradient deformation in the y-direction of the pipeline under test, u represent the displacement deformation in the x-direction of the pipeline under test, and v represent the displacement deformation in the y-direction of the pipeline under test. The deformation of the pipeline under test relative to the x-axis in the x-direction is considered. The deformation of the pipeline under test in the y-direction relative to the x-axis is considered. The deformation of the pipeline under test is due to its rotation relative to the y-axis in the x-direction. The deformation of the pipeline under test is due to rotation relative to the y-axis in the y-direction.

[0066] Specifically, the convolution kernel function is:

[0067]

[0068] Where i = 0, ..., hs, j = 0, ..., ws, t = 0, ..., n-1, s is the window size of the convolution kernel function in the first or second layer of the neural network, n is the convolutional layer level of the first or second layer of the neural network, m is the number of the previous layer, x is the output of the previous layer, K is the output of the current layer, I is the convolution result of the image layer, h and w are the resolution of the input reference image or noisy image, and b is the convolution bias.

[0069] Specifically, the linear transformation layer function is:

[0070]

[0071] Where α is randomly generated by Gaussian to improve the convergence speed, αx is the translation process of the image when x is less than 0, and x is the original data that is retained when x is greater than or equal to 0.

[0072] Specifically, the noise filtering function is:

[0073]

[0074] Where M is the total number of layers, m is the current layer number, ω is the weight matrix of the reference image or the noisy image, and x is the layer output matrix of the reference image or the noisy image.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] (1) Since the present invention uses a convolutional neural network, the corresponding convolutional neural network denoising function can be determined according to the target information under different noise environments. The image is denoised by the convolutional neural network, thereby improving the accuracy and stability of deformation measurement.

[0077] (2) Since the present invention uses a convolutional neural network method to filter and reduce noise in digital images, it is suitable for optical deformation measurement under complex lighting conditions.

[0078] (3) Because the system requirements of this invention are simple and only require taking pictures on site, the cost is relatively low, the measurement results are abundant, and the measurement limitations are small. It is particularly suitable for the detection of large and complex pipeline systems and provides a reliable measurement basis for the detection of complex pipeline systems.

[0079] (4) Since the present invention only requires taking pictures on site and the data can be processed online, the measurement is convenient during the detection process, the calculation is fully automated, the detection cycle is short, and the efficiency and accuracy of optical deformation measurement are greatly improved.

[0080] (5) The present invention has excellent noise suppression effect on various noise interferences of imaging environment and imaging parameters.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for denoising deformed pipeline images, characterized in that, Includes the following steps: S1. Make multiple marker targets for identifying different pipelines under test, attach the multiple marker targets to the surface of different pipelines under test and take pictures to obtain multiple original images of the pipelines under test containing the marker targets. S2. Measure the distance between each pipeline under test using a distance measuring tool, and measure the strain data of each pipeline under test using strain gauges pre-set on each pipeline under test containing a target marker, to obtain the deformation distance data and deformation strain data of each pipeline under test, and take pictures of each pipeline under test after deformation according to different combinations of shooting parameters, to obtain deformation images of the pipeline under test corresponding to different combinations of shooting parameters. S3. Based on all the original images of the pipeline under test and their corresponding deformation images, strain calculation is performed. The deformation spacing data and deformation strain data corresponding to the calculated deformation images of the pipeline under test are compared with the deformation spacing data and deformation strain data corresponding to the pipeline under test measured in the same pipeline under test. From the comparison results, the deformation images of the pipeline under test that are similar to the measured data are selected as reference images. A reference image set is constructed through multiple reference images. The remaining deformation images of the pipeline under test are used as noise images. A noise image set is constructed through multiple noise images. S4. Construct a digital image noise model based on a neural network, and train the digital image noise model using the reference image set and the noise image set to obtain the trained digital image noise model; S5. The imported target pipeline image is denoised using the trained digital image noise model, and the denoised target pipeline image is output.

2. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The marker target, manufactured to identify different pipelines under test, includes: Two different colored strips are arranged in a preset order along the horizontal axis of the pipeline to be tested to obtain the marker target.

3. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The different shooting parameter combinations include combinations of different parameters such as focal length, exposure, and gain.

4. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The strain calculation based on all the original images of the pipeline under test and their corresponding deformation images includes: The strain is calculated using a strain mapping formula to obtain the deformation spacing data and deformation strain data corresponding to the deformation image of the pipeline under test, based on the original image of the pipeline under test and its corresponding deformation image. The strain mapping formula is as follows: Where, x i ′ represents the x-coordinate of the deformed image, y-coordinate is 0. i ' is the y-coordinate of the deformed image, d x Let d be the gradient deformation of the pipeline under test in the x-direction. y Let represent the gradient deformation in the y-direction of the pipeline under test, u represent the displacement deformation in the x-direction of the pipeline under test, and v represent the displacement deformation in the y-direction of the pipeline under test. The deformation of the pipeline under test relative to the x-axis in the x-direction is considered. The deformation of the pipeline under test in the y-direction relative to the x-axis is considered. The deformation of the pipeline under test is due to its rotation relative to the y-axis in the x-direction. The deformation of the pipeline under test is due to rotation relative to the y-axis in the y-direction.

5. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The process involves comparing the calculated deformation spacing data and deformation strain data corresponding to the deformation image of the pipeline under test with the deformation spacing data and deformation strain data corresponding to the same pipeline under test, and selecting the deformation image of the pipeline under test that is closest to the measured data as a reference image from the comparison results, including: The deformation spacing data and deformation strain data corresponding to the calculated deformation images of the pipeline under test are respectively subtracted from the deformation spacing data and deformation strain data corresponding to the pipeline under test measured in the same pipeline under test. The absolute value of the difference is taken to obtain the absolute value of the difference corresponding to each deformation image of the pipeline under test. The deformation images of the pipeline under test corresponding to the absolute value of the difference corresponding to each deformation image of the pipeline under test are selected from the preset judgment threshold and used as reference images.

6. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The method for constructing a digital image noise model based on a convolutional neural network involves training the digital image noise model using the reference image set and the noisy image set to obtain the trained digital image noise model, including: A digital image noise model is constructed based on a neural network. The digital image noise model includes a three-layer neural network, with the first and second layers of the neural network respectively connected to the third layer of the neural network. The reference image set is used as the input of the first layer neural network. The target images in each reference image of the reference image set are extracted through the convolution kernel function and linear transformation layer function of the first layer neural network. The target images in each reference image and their corresponding reference images are then input into the third layer neural network. The noisy image set is used as the input of the second layer neural network. The target images in each noisy image in the noisy image set are extracted by the convolution kernel function and linear transformation layer function of the second layer neural network. The target images in each noisy image and their corresponding noisy images are then input into the third layer neural network. The third-layer neural network is trained by using a noise filtering function to filter noise from the target images in each reference image and their corresponding reference images, as well as the target images in each noise image and their corresponding noise images, to obtain a trained digital image noise model.

7. The pipeline deformation image noise reduction method according to claim 6, characterized in that, The convolution kernel function is: Where i = 0, ..., hs, j = 0, ..., ws, t = 0, ..., n-1, s is the window size of the convolution kernel function in the first or second layer of the neural network, n is the convolutional layer level of the first or second layer of the neural network, m is the number of the previous layer, x is the output of the previous layer, K is the output of the current layer, I is the convolution result of the image layer, h and w are the resolution of the input reference image or noisy image, b is the convolution bias, and r, k and l are the layer level and window function parameters of the convolutional neural network, respectively.

8. The pipeline deformation image noise reduction method according to claim 6, characterized in that, The linear transformation layer function is: Where α is randomly generated by Gaussian to improve the convergence speed, αx is the translation process of the image when x is less than 0, and x is the original data that is retained when x is greater than or equal to 0.

9. The pipeline deformation image noise reduction method according to claim 6, characterized in that, The noise filtering function is: Where M is the total number of layers, m is the current layer number, ω is the weight matrix of the reference image or the noisy image, and x is the layer output matrix of the reference image or the noisy image.

10. The pipeline deformation image noise reduction method according to claim 1, characterized in that, The method of measuring the distance between each pipeline under test using a ranging tool, and measuring the strain data of each pipeline under test using strain gauges pre-set on each pipeline under test containing a target marker, to obtain the deformation spacing data and deformation strain data corresponding to each pipeline under test, further includes: The distance between each pipeline under test is measured using a distance measuring tool according to a set number of times, and the strain data of each pipeline under test is measured using strain gauges pre-set on each pipeline under test with marked targets according to a set number of times. Multiple initial data of deformation distance and multiple data of deformation strain are obtained for each pipeline under test. The average value of the multiple initial data of deformation distance is obtained to obtain the deformation distance data for each pipeline under test, and the average value of the multiple data of deformation strain is obtained to obtain the deformation strain data for each pipeline under test.

Images