Method for detecting structural abnormalities of key components of an overhead contact line support device

Abstract

The application discloses a kind of contact network support device key component structure anomaly detection methods, comprising the following steps: application shooting template, the angle position of image acquisition is limited, ensure the standard of shooting image, and construct contact network support device image database;Current contact network support device image needs to be evaluated and collected;Contact network support device regional positioning;Contact network support device key component semantic profile extraction;The angle information of each component of contact network support device is calculated, compared with historical image angle information, and the abnormal result is output.The application, by semantic profile combined with traditional image processing method, successfully obtains the angle information of each component of contact network support device, solves the problem that manual cannot measure angle on site, at the same time, by comparing with historical image angle information, the abnormal state can be output, which greatly guarantees the normal state of contact network support device and ensures the safe operation of vehicle.

Description

A method for detecting structural anomalies in key components of overhead contact line support devices Technical Field

[0001] This invention belongs to the field of image processing technology, and specifically relates to a method for detecting structural abnormalities in key components of a contact wire support device. Background Technology

[0002] The overhead contact system is a special type of power transmission line that runs along a railway line, supplying power to electric locomotives. It consists of several parts: contact suspension, support devices, positioning devices, supports, and foundations. The contact suspension includes the contact wire, droppers, catenary wire, connecting parts, and insulators. The contact suspension is supported by support devices on supports, and its function is to transmit electrical energy from the traction substation to the electric locomotive. Support devices support the contact suspension and transfer its load to supports or other structures. These devices vary depending on the section of track, station, and large building in which the contact network is located. Support devices include cantilever arms, horizontal tie rods, suspension insulator strings, rod insulators, and other special support equipment for structures. Positioning devices include positioning tubes and positioners, whose function is to fix the position of the contact wire, ensuring it remains within the pantograph's travel path, preventing detachment from the pantograph, and transferring the horizontal load of the contact wire to the supports. The supports and foundations bear the entire load of the contact suspension, support devices, and positioning devices, and fix the contact suspension at the specified position and height. The existing overhead contact lines in China use prestressed reinforced concrete supports and steel columns. The foundation refers to the steel supports, meaning the steel supports are fixed to a reinforced concrete base. The foundation bears all the load transmitted from the supports and ensures their stability. The prestressed reinforced concrete supports and foundations are integrated, with their lower ends directly buried in the ground. Within a certain trajectory, the contact wires of the electrical contact network can also transmit the received pressure to the support system, thereby maintaining the stability of the entire electrified railway contact network system.

[0003] Due to the special nature of rail transit, the angles of each component of each overhead contact line support device are different during installation. When a problem occurs with the overhead contact line support device, and maintenance personnel have completed the repair, they may be unable to measure the angle of the repaired support device due to on-site limitations. This may result in an angle difference between the repaired support device and the original state, threatening the installation and transportation of vehicles. Summary of the Invention

[0004] To address the technical problems existing in the prior art, the purpose of this invention is to provide a method for detecting structural abnormalities in key components of a contact wire support device.

[0005] To achieve the above objectives and technical effects, the technical solution adopted by this invention is as follows:

[0006] A method for detecting structural abnormalities in key components of a catenary support device includes the following steps:

[0007] 1) Apply shooting templates to limit the angle and position of the captured images, ensure the standardization of the captured images, and build an image database for the contact wire support device;

[0008] 2) Collect images of the overhead contact line support device that needs to be evaluated, and obtain the original state images of the overhead contact line support device;

[0009] 3) Location of the overhead contact line support device area;

[0010] 4) Semantic contour extraction of key components of the overhead contact line support device;

[0011] 5) Calculate the angle information of each component of the overhead contact line support device, compare it with the angle information of historical images, and output the abnormal results.

[0012] Furthermore, in step 1), the images in the image database of the overhead contact line support device must contain information including: line, row, starting station, ending station, and pole number, so as to locate the position of the overhead contact line based on the information; the overhead contact line support and the flat cantilever arm must be within the template.

[0013] Furthermore, in step 2), the steps of acquiring images of the current overhead contact line support device to be evaluated and obtaining the original state image of the overhead contact line support device include:

[0014] Use imaging equipment to capture images of the overhead contact line support device that has been maintained. The captured images must meet the standards of step 1). At the same time, record the line, row, starting station, ending station, and pole number of the overhead contact line support device. Combine the information with the image database of the overhead contact line support device established in step 1) to obtain the original state image of the overhead contact line support device.

[0015] Furthermore, in step 3), the step of locating the contact wire support device area includes:

[0016] The original state image and current image of the catenary support device obtained in step 2) are processed. First, the area of ​​the catenary support device is located. The YOLOv5 target detection model is used to annotate the collected images of the catenary support device, and the rectangular boxes of the catenary support device area are marked. Based on the annotated image data and the annotation information, the catenary support device positioning model is trained and the coordinate information of the catenary support device is output.

[0017] Furthermore, in step 4), the semantic contour extraction steps for key components of the overhead contact line support device include:

[0018] Data annotation is performed on the image data of the overhead contact line support device, and key components of the overhead contact line are annotated. The OCRNet semantic segmentation model is used. First, a rough segmentation result is obtained using a general semantic segmentation model. At the same time, the features of each pixel can also be obtained from the backbone. Based on the semantic information and features of each pixel, the features of each category are obtained. Then, the similarity between the pixel features and the features of each category is calculated. Based on the similarity, the probability of each pixel belonging to each category is obtained. The representation of each region is weighted to obtain the enhanced feature representation of the current pixel. The overhead contact line support device segmentation model is trained by combining the labeled image and annotation information. The image of the overhead contact line support device obtained in step 3) is input into the overhead contact line support device segmentation model to obtain the corresponding semantic contour.

[0019] Furthermore, in step 5, the steps of calculating the angle information of each component of the overhead contact line support device, comparing it with the angle information of historical images, and outputting abnormal results include:

[0020] Based on the semantic contour obtained in step 4), the angle information of key components of the overhead contact line support device is calculated.

[0021] The angle information difference between the current image and the historical image is calculated separately. When the deviation is greater than the set threshold, it is considered that there is a problem with the current repair work, an alarm is reported, and the on-site maintenance personnel are reminded to carry out secondary construction to ensure that the contact wire support device is restored to its original state after repair.

[0022] Furthermore, the steps for calculating the angle information of key components of the overhead contact line support device include:

[0023] First, based on the semantic segmentation labels, obtain the masks corresponding to each key component of the overhead contact line support device, exclude the semantic information of other components, and then use the opencvminAreaRect function to obtain the minimum bounding rectangle of the semantic contour of each key component, and obtain the angle information of the rectangle.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0025] This invention discloses a method for detecting structural anomalies in key components of a catenary support device, comprising the following steps: applying a shooting template to limit the angle and position of the acquired images to ensure the standardization of the captured images, and constructing a catenary support device image database; acquiring the current image of the catenary support device to be evaluated, obtaining the original state image of the catenary support device; locating the catenary support device region; extracting semantic contours of key components of the catenary support device; calculating the angle information of each component of the catenary support device, comparing it with the angle information of historical images, and outputting the anomaly result. In the method for detecting structural anomalies in key components of a catenary support device provided by this invention, the application of a shooting template to limit the angle and position of the acquired images ensures the standardization of the captured images, guarantees the consistency of the angle and position of the captured images, reduces the deviation between historical images and the current image to be detected, and successfully obtains the angle information of each component of the catenary support device by combining semantic contours with traditional image processing methods, solving the problem that manual on-site angle measurement is impossible. Simultaneously, by comparing with the angle information of historical images, anomaly status can be output, greatly ensuring the normal state of the catenary support device and ensuring the safe operation of the vehicle. Attached Figure Description

[0026] Figure 1 is a schematic diagram of step 1) of a method for detecting structural abnormalities in key components of a contact wire support device according to the present invention.

[0027] Figure 2 is a schematic diagram of step 3) of a method for detecting structural abnormalities in key components of a contact wire support device according to the present invention;

[0028] Figure 3 is a schematic diagram of step 4) of a method for detecting structural abnormalities in key components of a contact wire support device according to the present invention.

[0029] Figures 4-5 are schematic diagrams of step 5) of a method for detecting structural abnormalities in key components of a contact wire support device according to the present invention. Detailed Implementation

[0030] The present invention will now be described in detail so that its advantages and features can be more easily understood by those skilled in the art, thereby providing a clearer and more explicit definition of the scope of protection of the present invention.

[0031] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form to prepare for the more detailed descriptions that follow.

[0032] Example 1

[0033] As shown in Figures 1-5, a method for detecting structural abnormalities in key components of a catenary support device includes the following steps:

[0034] 1) Apply shooting templates to limit the angle and position of the captured images, ensuring the standardization of the captured images, and construct an image database for the overhead contact line support device:

[0035] The angles of each component of the overhead contact line support device are different during installation, making it impossible to use a uniform standard to judge each overhead contact line support device. Therefore, the images of the overhead contact line support device captured by the present invention using mobile phones or other imaging devices need to simultaneously include information such as line, row, starting station, ending station, and pole number, and then all information is entered into the overhead contact line support device image database.

[0036] In this step, in order to ensure the consistency of the shooting angle and position, a position restriction needs to be added when shooting the contact network support device. It is stipulated that the contact network support column and flat bracket arm to be shot must be within the template (as shown in the red box in Figure 1) to reduce the impact of shooting position differences on angle calculation, reduce the deviation between historical images and the current image to be detected, and finally obtain the required contact network support device image.

[0037] 2) Acquire images of the overhead contact line support device currently being evaluated, and obtain the original state images of the overhead contact line support device:

[0038] Use mobile phones or other camera devices to capture images of the overhead contact line support device that has been maintained. The captured images must meet the standards of step 1). At the same time, record the line, row, starting station, ending station, and pole number of the overhead contact line support device. Combine this information with the image database of the overhead contact line support device established in step 1) to obtain the original state image of the overhead contact line support device.

[0039] 3) Location of the overhead contact line support device area:

[0040] The original state image and current image of the overhead contact system obtained in step 2) are processed. First, the area of ​​the overhead contact system support device is located. In this stage, the YOLOv5 target detection model is used. First, the collected images of the overhead contact system support device are labeled with rectangular boxes to mark the area of ​​the overhead contact system support device. Based on the labeled image data and the labeling information, the positioning model of the overhead contact system support device is trained and the coordinate information of the overhead contact system support device is output, as shown in Figure 2. The area shown in the red box in Figure 2 is the area of ​​the overhead contact system support device. The positioning of the area of ​​the overhead contact system support device is realized through the target detection model, and the corresponding position coordinates are obtained.

[0041] 4) Semantic contour extraction of key components of the overhead contact line support device:

[0042] Data annotation is performed on the image data of the overhead contact system support device, annotating areas such as overhead contact system supports, flat cantilever arms, inclined arms, positioning tubes, locators, and diagonal braces. An OCRNet semantic segmentation model is used. First, a general semantic segmentation model is used to obtain a rough segmentation result. Simultaneously, features of each pixel can be obtained from the backbone. Based on the semantic information and features of each pixel, features for each category can be obtained. Then, the similarity between pixel features and features of each category can be calculated. Based on this similarity, the probability of each pixel belonging to each category can be obtained. Further weighting of the representation of each region yields an enhanced feature representation for the current pixel. The overhead contact system support device segmentation model is trained using the labeled image and annotation information. The image of the overhead contact system support device obtained in step 3) is input into the segmentation model to obtain the corresponding semantic contours, as shown in Figure 3. In Figure 3, different colored masks represent different categories: red represents supports, green represents flat cantilever arms, yellow represents inclined arms, blue represents crossbars, cyan represents locators, and pink represents diagonal braces. The segmentation model achieves semantic segmentation of different key components, thereby obtaining the contours of the key components.

[0043] 5) Calculate the angle information of each component of the overhead contact line support device, compare it with the angle information in historical images, and output the alarm result:

[0044] Based on the semantic contours obtained in step 4), the angle information of key components such as the catenary support device pillars, flat brackets, inclined brackets, positioners, and diagonal braces is calculated. The specific calculation method is as follows:

[0045] First, the corresponding mask is obtained based on the semantic segmentation labels. Taking the pillar as an example, the obtained semantic image is shown in Figure 4. The red area represents the pillar outline, excluding the semantic information of other components. Then, the minimum bounding rectangle of the pillar's semantic outline is obtained using OpenCV's `minAreaRect` function, and the angle information of the rectangle is obtained. This process is repeated for each component, resulting in the angle information shown in Figure 5: the pillar angle is 90°, the horizontal arm angle is 3°, the diagonal arm angle is 150°, the diagonal brace angle is 53°, the locator angle is 177°, and the crossbar angle is 2°. This step successfully obtained the angles of each component by combining semantic contours with traditional image processing methods.

[0046] The angle information difference between the current image and the historical image is calculated separately. When the deviation is greater than the set threshold, it is considered that there is a problem with the current repair work, an alarm is reported, and the on-site maintenance personnel are reminded to carry out secondary construction to ensure that the contact wire support device is restored to its original state after repair.

[0047] Any parts or structures not specifically described in this invention can be made using existing technologies or products, and will not be elaborated upon here.

[0048] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for detecting structural abnormalities in key components of a catenary support device, characterized in that, Includes the following steps: 1) Apply shooting templates to limit the angle and position of the captured images, ensuring the standardization of the captured images, and construct an image database of the overhead contact line support device; 2) Collect images of the overhead contact line support device to be evaluated, and obtain the original state images of the overhead contact line support device; 3) Locate the area of ​​the overhead contact line support device; 4) Extract the semantic contours of key components of the overhead contact line support device; 5) Calculate the angle information of each component of the overhead contact line support device, compare it with the angle information of historical images, and output the abnormal results; In step 1), the images in the contact network support device image database must contain the following information: line, row, starting station, ending station, and pole number, which can be used to locate the contact network position; the contact network support and cantilever arm must be within the template; In step 2), the steps of collecting images of the contact network support device to be evaluated and obtaining the original state image of the contact network support device include: using a camera to collect images of the contact network support device that has been maintained, requiring that the collected images meet the standards of step 1), and simultaneously recording the line, row, starting station, ending station, and pole number of the contact network support device, and combining the information retrieval of the contact network support device image database established in step 1) to obtain the original state image of the contact network support device; In step 5, the steps of calculating the angle information of each component of the contact network support device, comparing it with the angle information of historical images, and outputting abnormal results include: based on the semantic contour obtained in step 4), calculating the angle information of key components of the contact network support device; calculating the difference in angle information between the current image and the historical image respectively, and when the deviation is greater than the set threshold, it is considered that there is a problem with the current repair work, an alarm is reported, and the on-site maintenance personnel are reminded to carry out secondary construction to ensure that the repaired contact network support device is restored to its original state.

2. The method for detecting structural abnormalities in key components of a contact wire support device according to claim 1, characterized in that, In step 3), the steps for locating the catenary support device area include: processing the original state image and current image of the catenary support device obtained in step 2); firstly, locating the catenary support device area; using the YOLOv5 target detection model to annotate the collected catenary support device image; annotating the rectangular box of the catenary support device area; training the catenary support device positioning model based on the annotated image data and annotation information; and outputting the coordinate information of the catenary support device.

3. The method for detecting structural abnormalities in key components of a contact wire support device according to claim 1, characterized in that, Step 4) involves extracting the semantic contours of key components of the overhead contact line support device. This includes: labeling the image data of the overhead contact line support device, labeling the key components of the overhead contact line, and using the OCRNet semantic segmentation model. First, a rough segmentation result is obtained using a general semantic segmentation model. Simultaneously, the features of each pixel can be obtained from the backbone. Based on the semantic information and features of each pixel, the features of each category are obtained. Then, the similarity between the pixel features and the features of each category is calculated. Based on the similarity, the probability of each pixel belonging to each category is obtained. The representation of each region is weighted to obtain the enhanced feature representation of the current pixel. The overhead contact line support device segmentation model is trained by combining the labeled image and annotation information. The image of the overhead contact line support device obtained in step 3) is input into the overhead contact line support device segmentation model to obtain the corresponding semantic contours.

4. The method for detecting structural abnormalities in key components of a contact wire support device according to claim 1, characterized in that, The steps for calculating the angle information of key components of the overhead contact line support device include: first, obtaining the mask corresponding to each key component of the overhead contact line support device based on the semantic segmentation label, excluding the semantic information of other components, and then using the opencv minAreaRect function to obtain the minimum bounding rectangle of the semantic contour of each key component, and obtaining the angle information of the rectangle.

Images