A method and device for identifying the intersection position of a girth weld and a spiral weld

Abstract

The present application discloses a method and device for identifying the intersection position of a girth weld and a spiral weld, and belongs to the technical field of oil and gas pipelines. The method comprises: obtaining an original detection signal image corresponding to a target oil and gas pipeline; based on at least one girth weld in the target oil and gas pipeline, intercepting at least one standard detection signal image corresponding to each girth weld from the original detection signal image according to target parameters; for a target girth weld in at least one girth weld, calling an intersection recognition model, processing the standard detection signal image corresponding to the target girth weld, and obtaining an intersection recognition result corresponding to the target girth weld. The embodiment of the present application realizes the automatic identification of the intersection between the girth weld and the spiral weld, improves the efficiency of extracting the intersection information between the girth weld and the spiral weld, and saves manpower and time costs.

Description

Technical Field

[0001] The embodiments of the present application belong to the technical field of oil and gas pipelines, and in particular, relate to a method and device for identifying the intersection position of a girth weld and a spiral weld. Background Art

[0002] During the operation and maintenance of oil and gas pipelines, it is necessary to conduct girth weld defect inspection, on-site excavation inspection, and repair work.

[0003] At present, the pipeline magnetic leakage internal detection system is mainly used to check for girth weld defects, and then the defective girth welds are excavated on site based on the inspection results. In the on-site engineering implementation, adjacent welds or nearby welds are easily excavated by mistake due to factors such as ground measurement and positioning errors. Therefore, it is necessary to confirm the accuracy of the excavated welds. In order to confirm the accuracy of the excavated welds, it is necessary to match the intersection position of the girth weld and the spiral weld for verification. In the process of girth weld risk inspection and excavation, the relevant technology mainly relies on manual identification of the intersection position of the girth weld and the spiral weld in the signal data of the magnetic leakage internal detection system.

[0004] However, the number of girth welds involved in oil and gas pipelines is in the millions. Manual identification of the intersection of girth welds and spiral welds requires a lot of manpower and time resources, which is not conducive to improving efficiency. Summary of the invention

[0005] The embodiment of the present application provides a method and device for identifying the intersection position of a girth weld and a spiral weld, which can be used to automatically identify the intersection between the girth weld and the spiral weld, improve the efficiency of extracting the intersection information between the girth weld and the spiral weld, and save manpower and time costs. The technical solution is as follows:

[0006] On the one hand, an embodiment of the present application provides a method for identifying the intersection position of a girth weld and a spiral weld, the method comprising:

[0007] Acquire the original detection signal image corresponding to the target oil and gas pipeline;

[0008] Based on at least one girth weld in the target oil and gas pipeline, intercepting standard detection signal images corresponding to the at least one girth weld from the original detection signal image according to target parameters;

[0009] For a target girth weld among the at least one girth weld, calling an intersection recognition model, processing a standard detection signal image corresponding to the target girth weld, and obtaining an intersection recognition result corresponding to the target girth weld;

[0010] The intersection recognition result corresponding to the target girth weld is used to indicate the intersection between the target girth weld and the spiral weld in the target oil and gas pipeline obtained by the intersection recognition model.

[0011] On the other hand, an embodiment of the present application provides a device for identifying the intersection position of a girth weld and a spiral weld, the device comprising:

[0012] An image acquisition module is used to acquire the original detection signal image corresponding to the target oil and gas pipeline;

[0013] An image standardization module, for intercepting a standard detection signal image corresponding to the at least one girth weld in the target oil and gas pipeline from the original detection signal image according to target parameters;

[0014] An intersection recognition module is used to call an intersection recognition model for a target girth weld among the at least one girth weld, process a standard detection signal image corresponding to the target girth weld, and obtain an intersection recognition result corresponding to the target girth weld;

[0015] The intersection recognition result corresponding to the target girth weld is used to indicate the intersection between the target girth weld and the spiral weld in the target oil and gas pipeline obtained by the intersection recognition model.

[0016] On the other hand, an embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the method for identifying the intersection position of the girth weld and the spiral weld as described above.

[0017] On the other hand, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for identifying the intersection position of the girth weld and the spiral weld as described above is implemented.

[0018] On the other hand, an embodiment of the present application provides a computer program product, which, when executed on a computer device, enables the computer device to execute the above-mentioned method for identifying the intersection position of the girth weld and the spiral weld.

[0019] The technical solution provided in the embodiments of the present application can bring the following beneficial effects:

[0020] By making a normalized detection signal image based on the magnetic flux leakage detection signal image, and constructing a training sample based on the normalized detection signal image to train the neural network model, and using the trained neural network model to identify the intersection between the girth weld and the spiral weld, the automatic identification of the intersection between the girth weld and the spiral weld is achieved. Moreover, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, the embodiment of the present application improves the efficiency of extracting the intersection information between the girth weld and the spiral weld, saving manpower and time costs. In addition, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, which relies on subjective judgment and experience, the embodiment of the present application uses a neural network model to identify the intersection between the girth weld and the spiral weld, which helps to improve the accuracy of intersection identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 It is a flow chart of a method for identifying the intersection position of a girth weld and a spiral weld provided by an embodiment of the present application;

[0023] Figure 2 is a schematic diagram of a standard detection signal image corresponding to a girth weld provided by an embodiment of the present application;

[0024] Figure 3 is a parameter diagram of an intersection recognition model provided by an embodiment of the present application;

[0025] Figure 4 is a schematic diagram of a model recognition error provided by an embodiment of the present application;

[0026] Figure 5 It is a block diagram of a device for identifying the intersection position of a girth weld and a spiral weld provided by an embodiment of the present application;

[0027] Figure 6 It is a block diagram of a device for identifying the intersection position of a girth weld and a spiral weld provided in another embodiment of the present application. DETAILED DESCRIPTION

[0028] Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of methods consistent with some aspects of the present application as detailed in the appended claims.

[0029] Please refer to Figure 1 , which shows a flow chart of a method for identifying the intersection position of a girth weld and a spiral weld provided by an embodiment of the present application. The method may include the following steps.

[0030] Step 110: Acquire the original detection signal image corresponding to the target oil and gas pipeline.

[0031] The target oil and gas pipeline refers to the oil and gas pipeline that needs to be inspected for girth weld defects. Usually, the target oil and gas pipeline is long, and is laid in multiple sections. The joint welds between every two sections are called girth welds. In the process of girth weld defect inspection, it is necessary to determine whether there is an intersection between the girth weld and the spiral weld in the target oil and gas pipeline, and the position of the intersection when there is an intersection. From the introduction of the above background technology, it can be seen that the relevant technology mainly relies on manual identification of the intersection between the girth weld and the spiral weld. This method requires a lot of manpower and time resources and is inefficient. Based on this, an embodiment of the present application provides a method for automatically determining the intersection between the girth weld and the spiral weld based on a neural network model, which helps to improve the efficiency and accuracy of girth weld defect inspection.

[0032] Step 120 , based on at least one girth weld in the target oil and gas pipeline, intercept at least one standard detection signal image corresponding to the girth weld from the original detection signal image according to the target parameters.

[0033] The embodiment of the present application is based on the original signal image of internal detection of the magnetic flux leakage internal detection system (for the convenience of description, it is referred to as the "original detection signal image"), and extracts the image area corresponding to each circumferential weld in the target oil and gas pipeline to produce a planned standard detection signal image. Then, the neural network model is trained and the intersection recognition is performed based on the standard detection signal image.

[0034] In the process of making the standard detection signal image, the actual length of the pipeline is used as a reference, and the standard detection signal image corresponding to each girth weld is intercepted from the original detection signal image according to the target parameters. Optionally, the target parameters include: the interception range of the upstream area of ​​the girth weld, the interception range of the downstream area of ​​the girth weld, based on which, the standard detection signal image corresponding to the girth weld includes: the area where the girth weld is located, the upstream area of ​​the girth weld, and the downstream area of ​​the girth weld, wherein the area where the girth weld is located is located between the upstream area of ​​the girth weld and the downstream area of ​​the girth weld. Optionally, the interception range of the upstream area of ​​the girth weld is equal to the interception range of the downstream area of ​​the girth weld, so that the length of the upstream area of ​​the girth weld is equal to the length of the downstream area of ​​the girth weld. Optionally, the interception range is between 0.3 meters and 0.5 meters. For example, taking the example that the interception range of the upstream area of ​​the girth weld and the interception range of the downstream area of ​​the girth weld are both 0.5 meters, the original detection signal image is intercepted, and the obtained standard detection signal image includes Figure 2 The standard detection signal image shown. Optionally, after obtaining the standard detection signal image, the standard detection signal image is stored in a picture format such as jpg or png for subsequent use.

[0035] Based on the normalized standard detection signal image, a training sample of the model can be constructed, and the model can be trained according to the constructed training sample, so that the trained model can be used to identify the intersection between the girth weld and the spiral weld later. In the embodiment of the present application, the neural network model used for identifying the intersection between the girth weld and the spiral weld is called an intersection recognition model. Based on this, in one example, the training process of the intersection recognition model is as follows: select n girth welds from at least one girth weld, n is a positive integer; perform intersection labeling processing on the standard detection signal images corresponding to the n girth welds, respectively, to obtain the intersection labeling results corresponding to the n girth welds; construct n training samples according to the standard detection signal images corresponding to the n girth welds, and the intersection labeling results corresponding to the n girth welds; train the intersection recognition model based on the n training samples to obtain a trained intersection recognition model.

[0036] The training of the intersection recognition model adopts a batch training method. Each batch of training requires selecting n girth welds from at least one girth weld. Optionally, the n girth welds are randomly selected from at least one girth weld. Then, the standard detection signal images corresponding to the n girth welds are respectively subjected to intersection labeling processing to obtain the intersection labeling results corresponding to the n girth welds. Afterwards, based on the intersection labeling results corresponding to the n girth welds and the standard detection signal images corresponding to the n girth welds, n training samples are constructed. Each training sample includes a standard detection signal image corresponding to a girth weld and an intersection labeling result. Finally, the intersection labeling model is trained using n training samples to obtain a trained intersection labeling model.

[0037] The intersection marking result is used to indicate the intersection between the girth weld and the spiral weld obtained by the intersection marking process. Optionally, the intersection marking result includes intersection existence data and intersection position data. The intersection existence data is used to indicate whether the intersection between the girth weld and the spiral weld exists. Assuming that E represents the intersection existence data, the value of E is {0,1}, where E is 0, indicating that the intersection does not exist, and E is 1, indicating that the intersection exists. The intersection position data is used to indicate the intersection position between the girth weld and the spiral weld and / or the direction parameter of the spiral weld relative to the target girth weld. Assuming that Px and Py represent the intersection position between the girth weld and the spiral weld, and Pz represents the direction parameter of the spiral weld relative to the target girth weld, the types of Pz include: the upstream spiral weld intersects the girth weld in the positive direction (corresponding to Pz value 1), the upstream spiral weld intersects the girth weld in the reverse direction (corresponding to Pz value 2), the downstream spiral weld intersects the girth weld in the positive direction (corresponding to Pz value 3), and the downstream spiral weld intersects the girth weld in the reverse direction (corresponding to Pz value 4). Based on this, the intersection annotation results include (E, Px, Py, Pz).

[0038] For example, Figure 2 In the first figure, the standard detection signal image corresponding to the girth weld indicates that there is an intersection between the girth weld 210 and the spiral weld 220, and E is 1. Figure 2 In the first figure, taking the upper left corner of the standard detection signal image as the origin, the intersection position between the girth weld and the upstream spiral weld is: Px1{196}, Py1{58}, and the intersection position between the girth weld and the downstream spiral weld is: Px2{190}, Py2{156}. Figure 2 In the first figure, the standard detection signal image corresponding to the girth weld indicates that the upstream spiral weld intersects the girth weld positively, and the downstream spiral weld intersects the girth weld positively, then Pz is {1,3}.

[0039] Before training the intersection recognition model, it is also necessary to build the intersection recognition model. Figure 3As shown in the figure, the intersection recognition model includes: convolutional layer (Convolutional), residual output (Residual), average pooling (Avgpool) and output activation function (Softmax).

[0040] Next, n training samples may be used to train the intersection recognition model. Optionally, the training of the intersection recognition model based on n training samples to obtain a trained intersection recognition model includes the following steps.

[0041] (1) The intersection recognition model is called to process the standard detection signal images corresponding to the n girth welds respectively with the first step length to obtain the first intersection recognition results corresponding to the n girth welds.

[0042] For each training sample in the n training samples, the intersection recognition model starts from the upper left corner of the standard detection signal image in the training sample, and performs convolutional neural network operations on the standard detection signal image grid by grid from left to right and from top to bottom with a first step length, and obtains the first intersection recognition result of each grid, that is, the calculation result of (E, Px, Py, Pz) of each network. Optionally, the first step length is 10 to 15 pixels.

[0043] (2) Based on the first intersection recognition results corresponding to the n girth welds and the intersection marking results corresponding to the n girth welds, first recognition errors corresponding to the n girth welds are determined.

[0044] For each standard detection signal image, the first intersection recognition result corresponding to the standard detection signal image can be compared with the intersection marking result corresponding to the standard detection signal image to obtain a first recognition error of the intersection recognition model. Optionally, the first recognition error includes: the error between the intersection existence data in the first intersection recognition result and the intersection existence data in the intersection marking result. Optionally, the first recognition error also includes: the error between the intersection position data in the first intersection recognition result and the intersection position data in the intersection marking result.

[0045] (3) According to the first recognition errors corresponding to the n ring welds, the parameters of the intersection recognition model are adjusted to obtain the intersection recognition model that has completed preliminary training.

[0046] According to the first recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model can be adjusted to make the first recognition errors converge, thereby completing the preliminary training of the intersection recognition model.

[0047] (4) The intersection recognition model that has completed preliminary training is called, and the image areas with the smallest first recognition errors in the standard detection signal images corresponding to the n ring welds are processed respectively with a second step length to obtain the second intersection recognition results corresponding to the n ring welds, and the second step length is smaller than the first step length.

[0048] In an embodiment of the present application, for the graphic area with the smallest first recognition error when the intersection recognition model performs intersection recognition on the standard detection signal image during preliminary training, the intersection recognition model that has completed preliminary training is called, and these image areas are processed separately with a second step length to obtain a second intersection recognition result. Wherein, the second step length is less than the first step length, and optionally, the second step length is 1 pixel. Optionally, when the bottom grid of the standard detection signal image is less than one step length from the bottom of the image, the content above the standard detection signal image is looped at the bottom of the standard detection signal image and supplemented to a complete step length.

[0049] (5) Based on the second intersection recognition results corresponding to the n girth welds and the intersection marking results corresponding to the n girth welds, the second recognition errors corresponding to the n girth welds are determined.

[0050] For each standard detection signal image, the second intersection recognition result corresponding to the standard detection signal image can be compared with the intersection marking result corresponding to the standard detection signal image to obtain a second recognition error of the intersection recognition model. Optionally, the second recognition error includes: an error between the intersection position data in the second intersection recognition result and the intersection position data in the intersection marking result.

[0051] (6) According to the second recognition errors corresponding to the n ring welds, the parameters of the initially trained intersection recognition model are adjusted to obtain a trained intersection recognition model.

[0052] According to the second recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model can be adjusted to make the second recognition errors converge, thereby completing the preliminary training of the intersection recognition model.

[0053] like Figure 4 As shown, it shows a schematic diagram of the model recognition error provided by the embodiment of the present application. By completing the training of the intersection recognition model, the error of the intersection clock position between the recognition result and the manual annotation result is less than 1 hour, and the error of the intersection clock position of 98% is less than 30 minutes. Therefore, the intersection recognition model provided by the embodiment of the present application can achieve a relatively accurate intersection recognition effect after completing the training.

[0054] Step 130, for a target girth weld in at least one girth weld, calling an intersection recognition model, processing a standard detection signal image corresponding to the target girth weld, and obtaining an intersection recognition result corresponding to the target girth weld.

[0055] The trained intersection recognition model can be used to process the standard detection signal image without intersection marking in at least one standard detection signal image, that is, to process the standard detection signal image corresponding to the target girth weld to obtain the intersection recognition result corresponding to the target girth weld. The intersection recognition result corresponding to the target girth weld is used to indicate the intersection between the target girth weld obtained by the intersection recognition model and the spiral weld in the target oil and gas pipeline. Optionally, the intersection recognition result corresponding to the target girth weld includes: intersection existence data and intersection position data, wherein: the intersection existence data is used to indicate whether there is an intersection between the target girth weld and the spiral weld; the intersection position data is used to indicate the intersection position between the target girth weld and the spiral weld when there is an intersection between the target girth weld and the spiral weld, and / or the direction parameter of the spiral weld relative to the target girth weld.

[0056] Optionally, the spiral weld includes an upstream spiral weld and a downstream spiral weld. Based on this, in the case where the target girth weld and the upstream spiral weld have an intersection, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the upstream spiral weld; the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the upstream spiral weld intersects the target girth weld in a positive direction, and the upstream spiral weld intersects the target girth weld in a reverse direction. In the case where the target girth weld and the downstream spiral weld have an intersection, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the downstream spiral weld; the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the downstream spiral weld intersects the target girth weld in a positive direction, and the downstream spiral weld intersects the target girth weld in a reverse direction.

[0057] In the embodiment of the present application, the intersection position data in the intersection marking result and the intersection position data in the intersection recognition result may include both the intersection coordinate position and the intersection clock position. Regardless of the type of intersection position included in the intersection position data, the two types can be converted to each other. Taking the intersection position data including the intersection coordinate position as an example, the conversion process between the intersection coordinate position and the intersection clock position is introduced and explained.

[0058] In one example, the intersection position data includes the intersection coordinate position; after the above step 130, it also includes: obtaining the target coordinate position corresponding to the target clock position in the standard detection signal image corresponding to the target ring weld; based on the intersection coordinate position and the target coordinate position, determining the intersection clock position corresponding to the intersection coordinate position.

[0059] The value of the clock position includes 12 values, namely 1 to 12 (where 12 is represented by the letter T), wherein the target clock position is a clock position used as a reference. Optionally, the target clock position is the position of 12 o'clock, such as Figure 2 Assume that the standard detection signal image corresponding to the target ring weld in the embodiment of the present application is as follows: Figure 2 In the first picture, the rectangular area from the rightmost side of the standard detection signal image to the left 10 pixels can be determined from top to bottom using the template matching algorithm to determine that the vertical coordinate of the intersection coordinate position corresponding to the target clock position T is 150.

[0060] Optionally, the above-mentioned determination of the intersection clock position corresponding to the intersection position coordinates based on the intersection coordinate position and the target coordinate position includes: determining the coordinate position difference between the intersection coordinate position and the target coordinate position; obtaining the image size of the standard detection signal image corresponding to the target ring weld; determining the clock position difference between the intersection clock position and the target clock position based on the coordinate position difference and the image size; and determining the intersection clock position based on the clock position difference and the target clock position.

[0061] The intersection coordinate position includes (Px, Py). The embodiment of the present application only uses Py to convert to the intersection clock position for illustration. After understanding the technical solution of the present application, those skilled in the art will easily think of the way and process of converting Px to the intersection clock position. Assuming that the image size includes the height H, the intersection coordinate position includes Py, and the target coordinate position includes Pt, the intersection clock position ClockPy is calculated as follows:

[0062] ClockPy=12(Pt-Py) / H, Py<Pt;

[0063] ClockPy=12-12(Py-Pt) / H, Py≥Pt.

[0064] For example, assuming that Figure 2 The first image in is used as the standard detection signal image corresponding to the target girth weld, then the intersection coordinate position between the target girth weld and the upstream spiral weld includes Py=58, and the intersection coordinate position between the target girth weld and the downstream spiral weld includes Py=156. From the above example, it can be seen that the target intersection position Pt=150, combined with the above formula, it can be obtained that: the intersection clock position between the target girth weld and the upstream spiral weld includes ClockPy=12(Pt-Py) / H=5.52, corresponding to 5:31; the intersection clock position between the target girth weld and the downstream spiral weld includes ClockPy=12-12(Py-Pt) / H=11.64, corresponding to 11:40.

[0065] In summary, the technical solution provided by the embodiment of the present application, by producing a normalized detection signal image based on the magnetic flux leakage internal detection signal image, and constructing a training sample based on the normalized detection signal image to train a neural network model, and using the trained neural network model to identify the intersection between the girth weld and the spiral weld, realizes the automatic identification of the intersection between the girth weld and the spiral weld. Moreover, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, the embodiment of the present application improves the efficiency of extracting the intersection information between the girth weld and the spiral weld, saving manpower and time costs. In addition, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, which relies on subjective judgment and experience, the embodiment of the present application uses a neural network model to identify the intersection between the girth weld and the spiral weld, which helps to improve the accuracy of intersection identification.

[0066] Please refer to Figure 5 , which shows a block diagram of a device for identifying the intersection position of a girth weld and a spiral weld provided by an embodiment of the present application. The device has the function of implementing the example of the method for identifying the intersection position of the girth weld and the spiral weld, and the function can be implemented by hardware or by hardware executing corresponding software. The device can be a computer device or can be set in a computer device. The device 500 can include: an image acquisition module 510, an image standardization module 520 and an intersection recognition module 530.

[0067] The image acquisition module 510 is used to acquire the original detection signal image corresponding to the target oil and gas pipeline.

[0068] The image standardization module 520 is used to extract, based on at least one girth weld in the target oil and gas pipeline, a standard detection signal image corresponding to the at least one girth weld from the original detection signal image according to target parameters.

[0069] The intersection recognition module 530 is used to call the intersection recognition model for the target girth weld in the at least one girth weld, process the standard detection signal image corresponding to the target girth weld, and obtain the intersection recognition result corresponding to the target girth weld; wherein the intersection recognition result corresponding to the target girth weld is used to indicate the intersection situation between the target girth weld obtained by the intersection recognition model and the spiral weld in the target oil and gas pipeline.

[0070] In one example, if Figure 6As shown, the above-mentioned device 500 also includes: a sample selection module 540, which is used to select n ring welds from the at least one ring weld, wherein n is a positive integer; an intersection marking module 550, which is used to perform intersection marking processing on the standard detection signal images corresponding to the n ring welds, respectively, to obtain the intersection marking results corresponding to the n ring welds; wherein the intersection marking results are used to indicate the intersection situation between the ring weld and the spiral weld obtained by the intersection marking processing; a sample construction module 560, which is used to construct n training samples according to the standard detection signal images corresponding to the n ring welds, respectively, and the intersection marking results corresponding to the n ring welds, respectively; a model training module 570, which is used to train the intersection recognition model based on the n training samples to obtain a trained intersection recognition model.

[0071] In one example, the above-mentioned model training model 570 is used to: call the intersection recognition model, process the standard detection signal images corresponding to the n circumferential welds respectively with the first step length, and obtain the first intersection recognition results corresponding to the n circumferential welds; determine the first recognition errors corresponding to the n circumferential welds based on the first intersection recognition results corresponding to the n circumferential welds and the intersection marking results corresponding to the n circumferential welds; adjust the parameters of the intersection recognition model according to the first recognition errors corresponding to the n circumferential welds, and obtain the intersection recognition model that has completed the preliminary training; call the completed preliminary training An intersection recognition model is provided, and the image areas with the smallest first recognition errors in the standard detection signal images corresponding to the n girth welds are processed respectively with a second step length to obtain second intersection recognition results corresponding to the n girth welds, and the second step length is smaller than the first step length; based on the second intersection recognition results corresponding to the n girth welds and the intersection marking results corresponding to the n girth welds, second recognition errors corresponding to the n girth welds are determined; according to the second recognition errors corresponding to the n girth welds, the parameters of the intersection recognition model that has completed preliminary training are adjusted to obtain a trained intersection recognition model.

[0072] In one example, the first recognition error includes: an error between the intersection existence data in the first intersection recognition result and the intersection existence data in the intersection marking result; the second recognition error includes: an error between the intersection position data in the second intersection recognition result and the intersection position data in the intersection marking result.

[0073] In one example, the standard detection signal image corresponding to the girth weld includes: the area where the girth weld is located, the upstream area of ​​the girth weld, and the downstream area of ​​the girth weld; wherein, the area where the girth weld is located is located between the upstream area of ​​the girth weld and the downstream area of ​​the girth weld, and the length of the upstream area of ​​the girth weld is equal to the length of the downstream area of ​​the girth weld.

[0074] In one example, the intersection identification result corresponding to the target girth weld includes: intersection existence data and intersection position data, wherein: the intersection existence data is used to indicate whether there is an intersection between the target girth weld and the spiral weld; the intersection position data is used to indicate the intersection position between the target girth weld and the spiral weld, and / or the direction parameters of the spiral weld relative to the target girth weld when there is an intersection between the target girth weld and the spiral weld.

[0075] In one example, the spiral weld includes an upstream spiral weld and a downstream spiral weld; when there is an intersection between the target girth weld and the upstream spiral weld, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the upstream spiral weld; when there is an intersection between the target girth weld and the downstream spiral weld, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the downstream spiral weld.

[0076] In one example, the spiral weld includes an upstream spiral weld and a downstream spiral weld; when there is an intersection between the target girth weld and the upstream spiral weld, the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the upstream spiral weld intersects the target girth weld in a positive direction, and the upstream spiral weld intersects the target girth weld in a reverse direction; when there is an intersection between the target girth weld and the downstream spiral weld, the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the downstream spiral weld intersects the target girth weld in a positive direction, and the downstream spiral weld intersects the target girth weld in a reverse direction.

[0077] In one example, the intersection position data includes the intersection coordinate position; Figure 6 As shown, the above-mentioned device 500 also includes: a position acquisition module 580, which is used to obtain the target coordinate position corresponding to the target clock position in the standard detection signal image corresponding to the target ring weld; a position conversion module 590, which is used to determine the intersection clock position corresponding to the intersection coordinate position based on the intersection coordinate position and the target coordinate position.

[0078] In one example, if Figure 6As shown, the position conversion module 590 is used to: determine the coordinate position difference between the intersection coordinate position and the target coordinate position; obtain the image size of the standard detection signal image corresponding to the target ring weld; determine the clock position difference between the intersection clock position and the target clock position based on the coordinate position difference and the image size; determine the intersection clock position based on the clock position difference and the target clock position.

[0079] In summary, the technical solution provided by the embodiment of the present application, by producing a normalized detection signal image based on the magnetic flux leakage internal detection signal image, and constructing a training sample based on the normalized detection signal image to train a neural network model, and using the trained neural network model to identify the intersection between the girth weld and the spiral weld, realizes the automatic identification of the intersection between the girth weld and the spiral weld. Moreover, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, the embodiment of the present application improves the efficiency of extracting the intersection information between the girth weld and the spiral weld, saving manpower and time costs. In addition, compared with manually identifying and marking the intersection between the girth weld and the spiral weld, which relies on subjective judgment and experience, the embodiment of the present application uses a neural network model to identify the intersection between the girth weld and the spiral weld, which helps to improve the accuracy of intersection identification.

[0080] It should be noted that the device provided in the above embodiment only uses the division of the above functional modules as an example to implement its functions. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and their specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0081] In an exemplary embodiment, a computer device is also provided, which includes a processor and a memory, wherein a computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the method for identifying the intersection position of the girth weld and the spiral weld as described above.

[0082] In an exemplary embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, the method for identifying the intersection position of the girth weld and the spiral weld as described above is implemented.

[0083] In an exemplary embodiment, a computer program product is also provided. When the computer program product is run on a computer device, the computer device is enabled to execute the above-mentioned method for identifying the intersection position of the girth weld and the spiral weld.

[0084] The above description is only an exemplary embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying the intersection position of a girth weld and a spiral weld, characterized in that: The method comprises: Acquire the original detection signal image corresponding to the target oil and gas pipeline; Based on at least one girth weld in the target oil and gas pipeline, intercepting standard detection signal images corresponding to the at least one girth weld from the original detection signal image according to target parameters; For a target girth weld among the at least one girth weld, calling an intersection recognition model, processing a standard detection signal image corresponding to the target girth weld, and obtaining an intersection recognition result corresponding to the target girth weld; The intersection recognition result corresponding to the target girth weld is used to indicate the intersection between the target girth weld and the spiral weld in the target oil and gas pipeline obtained by the intersection recognition model; The training process of the intersection recognition model is as follows: Select n girth welds from the at least one girth weld, where n is a positive integer; Performing intersection marking processing on the standard detection signal images corresponding to the n girth welds respectively, to obtain intersection marking results corresponding to the n girth welds respectively; Constructing n training samples according to the standard detection signal images corresponding to the n girth welds and the intersection annotation results corresponding to the n girth welds; Training the intersection recognition model based on the n training samples to obtain a trained intersection recognition model; Wherein, the intersection marking result is used to indicate the intersection between the girth weld and the spiral weld obtained through the intersection marking process; The step of training the intersection recognition model based on the n training samples to obtain a trained intersection recognition model includes: The intersection recognition model is called to process the standard detection signal images corresponding to the n girth welds respectively with the first step length to obtain the first intersection recognition results corresponding to the n girth welds; Determine first identification errors corresponding to the n girth welds respectively based on first intersection recognition results corresponding to the n girth welds respectively and intersection marking results corresponding to the n girth welds respectively; According to the first recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model are adjusted to obtain a preliminarily trained intersection recognition model; Calling the intersection recognition model that has completed preliminary training, and processing the image areas with the smallest first recognition errors in the standard detection signal images corresponding to the n girth welds respectively with a second step length, to obtain second intersection recognition results corresponding to the n girth welds respectively, wherein the second step length is smaller than the first step length; Determine second identification errors corresponding to the n girth welds respectively based on second intersection recognition results corresponding to the n girth welds respectively and intersection marking results corresponding to the n girth welds respectively; According to the second recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model that has completed preliminary training are adjusted to obtain a trained intersection recognition model.

2. The method according to claim 1, characterized in that The intersection recognition result corresponding to the target girth weld includes: intersection existence data and intersection position data, wherein: The intersection point existence data is used to indicate whether there is an intersection point between the target girth weld and the spiral weld; The intersection position data is used to indicate the intersection position between the target girth weld and the spiral weld, and / or the direction parameter of the spiral weld relative to the target girth weld when there is an intersection between the target girth weld and the spiral weld.

3. The method according to claim 2, characterized in that The spiral weld includes an upstream spiral weld and a downstream spiral weld; In the case where there is an intersection between the target girth weld and the upstream spiral weld, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the upstream spiral weld; the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the upstream spiral weld intersects the target girth weld in a positive direction, and the upstream spiral weld intersects the target girth weld in a reverse direction; In the case where there is an intersection between the target girth weld and the downstream spiral weld, the intersection position between the target girth weld and the spiral weld includes the intersection position between the target girth weld and the downstream spiral weld; the direction parameter of the spiral weld relative to the target girth weld includes any one of the following: the downstream spiral weld intersects the target girth weld in the positive direction, and the downstream spiral weld intersects the target girth weld in the reverse direction.

4. The method according to claim 2, characterized in that: The intersection point position data includes the intersection point coordinate position; After calling the intersection recognition model and processing the standard detection signal image corresponding to the target girth weld to obtain the intersection recognition result corresponding to the target girth weld, the method further includes: Obtaining a target coordinate position corresponding to a target clock position in a standard detection signal image corresponding to the target girth weld; Based on the intersection coordinate position and the target coordinate position, an intersection clock position corresponding to the intersection coordinate position is determined.

5. The method according to claim 4, characterized in that The determining, based on the intersection coordinate position and the target coordinate position, the intersection clock position corresponding to the intersection position coordinate comprises: Determining a coordinate position difference between the intersection coordinate position and the target coordinate position; Obtaining the image size of the standard detection signal image corresponding to the target girth weld; determining a clock position difference between the intersection clock position and the target clock position based on the coordinate position difference and the image size; The intersection clock position is determined based on the clock position difference and the target clock position.

6. A device for identifying the intersection position of a girth weld and a spiral weld, characterized in that: The device comprises: An image acquisition module, used to acquire the original detection signal image corresponding to the target oil and gas pipeline; An image standardization module, for intercepting a standard detection signal image corresponding to the at least one girth weld in the target oil and gas pipeline from the original detection signal image according to target parameters; An intersection recognition module is used to call an intersection recognition model for a target girth weld among the at least one girth weld, process a standard detection signal image corresponding to the target girth weld, and obtain an intersection recognition result corresponding to the target girth weld; The intersection recognition result corresponding to the target girth weld is used to indicate the intersection between the target girth weld and the spiral weld in the target oil and gas pipeline obtained by the intersection recognition model; The device also includes: A sample selection module, used for selecting n girth welds from the at least one girth weld, wherein n is a positive integer; An intersection marking module is used to perform intersection marking processing on the standard detection signal images corresponding to the n girth welds respectively, to obtain intersection marking results corresponding to the n girth welds respectively; A sample construction module, used to construct n training samples according to the standard detection signal images corresponding to the n girth welds and the intersection annotation results corresponding to the n girth welds; A model training module, used for training the intersection recognition model based on the n training samples to obtain a trained intersection recognition model; Wherein, the intersection marking result is used to indicate the intersection between the girth weld and the spiral weld obtained through the intersection marking process; The model training module is used to: The intersection recognition model is called to process the standard detection signal images corresponding to the n girth welds respectively with the first step length to obtain the first intersection recognition results corresponding to the n girth welds; Determine first identification errors corresponding to the n girth welds respectively based on first intersection recognition results corresponding to the n girth welds respectively and intersection marking results corresponding to the n girth welds respectively; According to the first recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model are adjusted to obtain a preliminarily trained intersection recognition model; Calling the intersection recognition model that has completed preliminary training, and processing the image areas with the smallest first recognition errors in the standard detection signal images corresponding to the n girth welds respectively with a second step length, to obtain second intersection recognition results corresponding to the n girth welds respectively, wherein the second step length is smaller than the first step length; Determine second identification errors corresponding to the n girth welds respectively based on second intersection recognition results corresponding to the n girth welds respectively and intersection marking results corresponding to the n girth welds respectively; According to the second recognition errors respectively corresponding to the n ring welds, the parameters of the intersection recognition model that has completed preliminary training are adjusted to obtain a trained intersection recognition model.

7. A computer device, characterized in that: The computer device includes a processor and a memory, wherein a computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the method for identifying the intersection position of the girth weld and the spiral weld as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying the intersection position of a girth weld and a spiral weld as described in any one of claims 1 to 5 is implemented.

Images