Automatic inspection and alarm method and system for section insulator

Abstract

The invention discloses an automatic inspection alarm method and system for a section insulator, and solves the problems of high inspection cost, long time consumption, low efficiency and large potential safety hazard of the section insulator in the prior art. The method comprises the following steps: cutting a guide plate into a plurality of parts, and setting a preset position; routing inspection routes are set, and routing inspection strategies on different routing inspection routes are set; acquiring a video stream, performing video decoding, and detecting image quality; obtaining a current inspection strategy, and determining current preset bit information; and determining the currently detected part according to the preset bit information, performing image recognition, judging whether the detected part is abnormal or not, and giving an alarm when the detected part is abnormal. A video intelligent application technology and an equipment remote control technology are utilized to carry out real-time monitoring, intelligent identification and automatic inspection on core components of the section insulator, automatic alarm of each core component of the section insulator equipment is realized, an informatization platform is combined, the manual inspection pressure is reduced, the efficiency is high, the cost is low, and normal operation of a line can be guaranteed.

Description

Technical Field

[0001] The invention relates to the technical field of track safety, and in particular to an automatic inspection and alarm method and system for segmented insulators. Background Art

[0002] As a key equipment in subway lines, segmented insulators are easily affected by complex factors such as abnormal dynamic contact between the pantograph and the catenary, differences in the electrical environment on both sides, train running vibration, wear and other complex factors in their working environment. They often experience faults such as changes in arc extinguishing angle displacement, excessive wear of the guide plate, loose fasteners, etc., which in turn trigger power tripping and uneven current distribution, seriously affecting the continuity and safety of subway operations.

[0003] At present, the detection of segment insulators is mainly carried out through manual maintenance, with monthly inspections and special inspections. At least 6 people are involved in each inspection, and each inspection is time-consuming, inefficient, and has great safety hazards, which seriously affects the normal operation of the line. In view of the complexity and variability of the subway line environment and the huge number of segment insulator equipment, the inspection and maintenance of segment insulator equipment is relatively difficult. At present, in the field of segment insulator monitoring, automated intelligent technology protection measures have not yet been adopted, so it is impossible to timely and effectively avoid major failures and ensure driving safety. Summary of the invention

[0004] The purpose of the present invention is to solve the problems of high cost, long time consumption, low efficiency and great safety hazards in the inspection of segmented insulators in the prior art, and to provide a segmented insulator automatic inspection alarm method and system, which utilizes video intelligent application technology and equipment remote control technology to perform real-time monitoring, intelligent identification and automatic inspection of the core components of the segmented insulator, so as to realize automatic alarm of the core components of the segmented insulator equipment, and combine with the information platform to reduce the pressure of manual inspection, with high efficiency and low cost, and can ensure the normal operation of the line.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A segmented insulator automatic inspection and alarm method, characterized in that it comprises the following steps: S1: cutting the guide plate into multiple parts and setting preset positions; S2: Set inspection routes and set inspection strategies on different inspection routes; S3: Obtain the video stream and perform video decoding to detect image quality; S4: Obtain the current inspection strategy and determine the current preset position information; S5: Determine the current detection part according to the preset position information, perform image recognition, judge whether the detection part is abnormal, and alarm when it is abnormal.

[0006] The method of the present invention can automatically inspect and intelligently alarm multiple core components of subway section insulator equipment. By setting different inspection strategies, customized intelligent detection of multiple core components can be implemented, which can not only realize inspection automation and rapid fault alarm, but also reduce construction costs and labor costs. It is highly efficient and low-cost, and can ensure the normal operation of the line.

[0007] Preferably, the method of determining the current detection position based on the preset position information includes: if it is an arc extinguishing angle preset position, detecting whether the arc extinguishing angle displacement is abnormal through the video stream; if it is a guide plate preset position, detecting whether the guide plate wear is abnormal through the video stream; if it is a combined preset position, detecting whether the fasteners are loose and whether the arc extinguishing angle displacement is abnormal at the same time through the video stream.

[0008] Preferably, the inspection route includes a first inspection route and a second inspection route, the first inspection route includes a first guide plate preset position, a second guide plate preset position and a third guide plate preset position, and the inspection strategy is: stay at each preset position for t minutes and inspect once every n hours; the second inspection route includes an arc extinguishing angle preset position, a fastener preset position, and a fastener and arc extinguishing angle combination preset position, and the inspection strategy is: stay at each preset position for 2t minutes and inspect once every 4n hours.

[0009] Preferably, an image is acquired, an analysis area of ​​interest is divided and the image is horizontally calibrated to outline the foreground range of the arc extinction angle, the positions of the vertical tangents on the left and right of the arc extinction angle are calculated based on the foreground range, the pixel spacing between the two tangents is calculated, and the pixel values ​​are converted into millimeter information. If the distance between the arc extinction angles is greater than a first threshold, the arc extinction angle is abnormal, and the vertical tangent is marked on the image, and the abnormal distance information is marked for storage and record.

[0010] Preferably, a rectangular window is taken in the x-direction according to the lower edge of the guide plate for sliding value acquisition, the grayscale mean in the y-direction is calculated, the mean curve is drawn, the curve trough spacing is calculated to obtain the guide plate thickness value, and the coordinate array of the upper and lower edges of the guide plate is output, and the pixel value is converted into millimeter information. If the distance between the upper and lower edges of the guide plate is less than the second threshold and the wear area is greater than the third threshold, the guide plate is abnormally worn, and the upper and lower edges of the guide plate and the area with continuous excessive wear are marked on the image for storage and record.

[0011] Preferably, capture an image, identify the fasteners on the segmented insulator through target detection, output the coordinate position of the fasteners, capture the image of the fasteners based on the coordinate position of the fasteners, identify the red marking line on the fasteners and the red marking line on the background plate, and if the angle of the two marking lines exceeds a fourth threshold, the fasteners are loose.

[0012] Preferably, when performing abnormality detection, if an abnormality is detected in a frame of image, one frame of image is extracted every second for identification. If abnormalities are found in m consecutive frames, an abnormality occurs in the detection part, an alarm message is sent, and screenshots of the video and alarm image m seconds before and after the alarm are taken.

[0013] As a preference, the abnormal alarm information is reconfirmed. If the second confirmation is an alarm, the train number and staff information passing the fault point are queried, the fault point information is sent to the staff, and corresponding emergency handling measures are executed.

[0014] An automatic inspection and alarm system for a segmented insulator, comprising: Front-end intelligent sensing equipment sets inspection routes and inspection strategies to obtain video data of core components of segmented insulators; The inspection dispatching unit identifies the inspection route and inspection strategy, automatically opens and closes the corresponding intelligent detection unit according to the current preset position, and decomposes the video data into independent image frames; An intelligent detection unit detects image quality, pre-processes the image frames and extracts key features, and performs intelligent prediction and abnormal alarm on the status of core components of the segmented insulator; The central control client receives the alarm information sent by the intelligent detection unit to issue an alarm prompt, and simultaneously plays the real-time video collected by the front-end intelligent sensing device.

[0015] Preferably, the intelligent detection unit comprises: The arc extinction angle intelligent detection unit uses the OPENCV image processing algorithm to process the image frames of the special area, calculate the distance between the arc extinction angles, and determine whether the arc extinction angle is abnormal; The guide plate intelligent detection unit grayscales and performs noise reduction processing on the image, calculates the guide plate thickness value, and determines whether the guide plate wear is abnormal; The fastener intelligent detection unit uses the target detection YOLO neural network and OPENCV image processing algorithm to determine whether the angle between the red line of the fastener and the red line of the background plate is greater than the threshold.

[0016] Therefore, the present invention has the following beneficial effects: using video intelligent application technology and equipment remote control technology to perform real-time monitoring, intelligent identification, and automatic inspection of the core components of the segmented insulator (arc extinguishing angle, guide plate, and disconnector), and realizing automatic alarm of the core components of the segmented insulator equipment. Combined with the information platform, it reduces the pressure of manual inspection, improves inspection efficiency, reduces costs, and ensures the normal operation of the line. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure is a flowchart of the overall steps of the automatic inspection and alarm method for segmented insulators in the first embodiment.

[0018] Figure 2 This is a schematic diagram of the architecture of the automatic inspection alarm of the segmented insulator in the second embodiment.

[0019] In the figure: 1. Front-end intelligent sensing device; 2. Computer room; 3. Central control center; 4. Spotlight; 5. First network switch; 6. Segment insulator; 7. Camera; 8. Convergence network switch; 9. Second network switch; 10. Algorithm server; 11. Data server; 12. Arc extinguishing angle intelligent detection unit; 13. Guide plate intelligent detection unit; 14. Fastener intelligent detection unit; 15. Third network switch; 16. Central control client. DETAILED DESCRIPTION

[0020] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments: Embodiment 1: This embodiment provides a segmented insulator automatic inspection alarm method, such as Figure 1 As shown, the operation process is as follows: step one, cutting the guide plate into multiple parts and setting preset positions; step two, setting inspection routes and setting inspection strategies on different inspection routes; step three, obtaining video streams and performing video decoding to detect image quality; step four, obtaining the current inspection strategy and determining the current preset position information; step five, determining the current inspection location according to the preset position information, performing image recognition, judging whether the inspection location is abnormal, and sounding an alarm when an abnormality occurs.

[0021] The automatic inspection and alarm method for segmented insulators provided in this embodiment utilizes video intelligent application technology and equipment remote control technology to perform real-time monitoring, intelligent identification, and automatic inspection of the core components of segmented insulators (arc extinguishing angles, guide plates, and disconnectors), thereby realizing automatic alarm of the core components of the segmented insulator equipment. Combined with an information platform, it reduces the pressure of manual inspections, improves inspection efficiency, reduces costs, and ensures the normal operation of the line.

[0022] The following further illustrates the technical solution and technical effects of the present invention through specific examples and specific application scenarios. The following examples are explanations of the present invention and the present invention is not limited to the following examples.

[0023] This embodiment provides an automatic inspection and alarm method for segmented insulators, which is mainly used to solve the problems of existing segmented insulator equipment failing to give timely alarms, manual inspections being costly, time-consuming, and inefficient, and managers being unable to grasp the status of key equipment in real time.

[0024] The specific process is: Step 1: Cut the guide plate into multiple parts and set the preset positions.

[0025] Since the guide plate is relatively long and cannot be displayed in the same screen, it is cut into multiple parts and multiple preset positions are set to facilitate image recognition.

[0026] In this embodiment, the preset positions include an arc extinguishing angle preset position, a first guide plate preset position, a second guide plate preset position, a third guide plate preset position, a fastener preset position, a fastener and arc extinguishing angle combination preset position, etc.

[0027] Step 2: Set up inspection routes and set inspection strategies on different inspection routes.

[0028] In this embodiment, two inspection routes are set up, wherein the first inspection route includes: the first guide plate preset position, the second guide plate preset position and the third guide plate preset position; the second inspection route includes: the arc extinguishing angle preset position, the fastener preset position, and the fastener and arc extinguishing angle combination preset position.

[0029] Then, the inspection strategy of the first inspection route is: stay at each preset position for t minutes and inspect once every n hours; the inspection strategy of the second inspection route is: stay at each preset position for 2t minutes and inspect once every 4n hours. In actual application, the inspection strategy can be fine-tuned according to the actual situation.

[0030] Step 3: Get the video stream and decode the video to detect the image quality.

[0031] If the image quality passes, execute step 4, otherwise continue to obtain the video stream and perform time-frequency decoding to detect the image quality until the image quality passes.

[0032] Step 4: Get the current inspection strategy and determine the current preset position information.

[0033] Step 5: Determine the current detection part according to the preset position information, perform image recognition, judge whether the detection part is abnormal, and alarm if it is abnormal.

[0034] If the current position is the arc extinguishing angle preset position, the video stream is used to detect whether the arc extinguishing angle displacement is abnormal; if the current position is the guide plate preset position, the video stream is used to detect whether the guide plate wear is abnormal; if the current position is the combined preset position, the video stream is used to simultaneously detect whether the fasteners are loose and whether the arc extinguishing angle displacement is abnormal.

[0035] Specific: (1) Detecting whether the arc extinction angle displacement is abnormal through video stream includes: The offset of the normal spacing of the arc extinction angle is ±2mm. If the algorithm detects that the offset exceeds ±5mm, an alarm needs to be issued immediately. This embodiment captures the camera image, divides the analysis area of ​​interest and performs horizontal calibration on the image, then binarizes the image, outlines the foreground range of the arc extinction angle, calculates the vertical tangent positions on the left and right of the arc extinction angle based on the foreground range, provides the calculation of the pixel spacing between the two tangents, and converts the pixel value into millimeters based on the configuration information corresponding to the pixel value and the millimeter. If the calculated distance between the arc extinction angles is greater than the first threshold change, the arc extinction angle is considered abnormal, and the vertical tangent is marked on the image, and the abnormal distance information is marked for storage and recording. In order to reduce false alarms, after the first frame of abnormality occurs, one frame of image is extracted every second, and multiple recognitions are performed until m frames are abnormal in succession, and the alarm information is immediately pushed to the platform, and a short video of m seconds before and after the alarm and the alarm image are intercepted and pushed to the platform at the same time.

[0036] (2) Checking whether the guide plate wear is abnormal through video streaming includes: The thickness of the guide plate is 10mm under normal circumstances. It will continue to wear during use, and the thickness of the guide plate will gradually decrease. If the algorithm detects that the thickness is less than 6mm (wear is more than 4mm), an alarm needs to be issued immediately. This embodiment adopts traditional OPENCV image processing technology. After capturing the camera image and dividing the analysis area of ​​interest, the image is grayed, filtered, and denoised. According to the lower edge of the guide plate, a rectangular window is taken in the x-axis direction for sliding value, and the grayscale mean in the y-axis direction is calculated. The curve is drawn, and the thickness value of the guide plate is obtained by calculating the curve trough spacing, and the pixel coordinate array of the upper and lower edges of the guide plate is output. According to the configuration information corresponding to the pixel value and the millimeter, the pixel value is converted to millimeter. If the calculated distance between the upper and lower edges of the guide plate is less than the second threshold, and the wear area is greater than the third threshold, it is considered that the guide plate is abnormally worn too much, and the upper and lower edges of the guide plate and the area of ​​continuous excessive wear are marked on the image for storage and recording. In order to reduce false alarms, after the first frame of abnormality appears, one frame of image is extracted per second and multiple recognitions are performed until m frames are abnormal continuously. Then the alarm information is immediately pushed to the platform, and a short video of m seconds before and after the alarm and the alarm image are captured and pushed to the platform at the same time.

[0037] (3) Detecting whether fasteners are loose through video streaming simultaneously includes: Under normal circumstances, the red line on the fastener and the red line on the background are on the same straight line. The fastener will gradually loosen due to vibration and other reasons. If the algorithm detects that the angle between the red line on the fastener and the red line on the background is greater than the fourth threshold, it is considered that the fastener is loose and an alarm needs to be issued immediately. The target detection YOLO neural network and OPENCV image processing technology are used. Capture camera image I kFirst, identify the fasteners on the segmented insulator through target detection, and output the fastener coordinate position [f 1k , f 2k , …, f ik , …f nk ], where f 1k A fastener identified by the fastener deep learning model, including the coordinates of the identification box and the width and height. 1k Image C of the fastener in the mid-coordinate intercept image k , C k The image is processed by OPENCV technology to identify the fasteners and the red marking lines on the background plate. If the angle of the two red marking lines exceeds the fourth threshold, the fastener is considered to be loose, and the location area and identification line of the fastener are marked on the image for storage and record. In order to reduce false alarms, after the first frame of abnormality appears, one frame of image is extracted per second and recognized multiple times until m consecutive frames are abnormal. The alarm information is immediately pushed to the platform, and a short video of m seconds before and after the alarm and the alarm image are intercepted and pushed to the platform at the same time. The fastener intelligent detection unit is installed on the algorithm server.

[0038] When an abnormal situation is successfully detected, the automatic alarm mechanism and voice prompt function are immediately activated, an alarm message is sent, and a real-time video is played synchronously so that the on-duty personnel can obtain the on-site situation immediately. After receiving the voice prompt, the on-duty personnel can quickly check the detailed alarm images and video materials, and directly access the real-time video stream for secondary confirmation from multiple angles. Once the equipment is confirmed to have failed, the information of the emergency repair personnel near the fault location is automatically provided. The on-duty personnel can quickly make a phone call to notify the emergency repair personnel to rush to the scene for emergency repair. At the same time, the train number and the driver's contact information that will pass the fault point can also be automatically queried. The on-duty personnel can quickly notify the driver of the corresponding train number, take emergency measures, and notify relevant management personnel to ensure driving safety and the timeliness of fault handling. Fault records and emergency repair records are archived, and fault data is deeply analyzed to continuously optimize maintenance strategies and provide more accurate and efficient decision support.

[0039] Embodiment 2: This embodiment provides a segmented insulator automatic inspection and alarm system, which is brought into a specific application scenario and is used to implement the segmented insulator automatic inspection and alarm method in the first embodiment.

[0040] Specifically, an automatic inspection and alarm system for segmented insulators includes: a front-end intelligent sensing device 1, a machine room 2 and a central control center 3. The front-end intelligent sensing device is connected to the machine room, and the machine room is connected to the central control center. The front-end intelligent sensing device is installed on the on-site subway line, and the machine room is installed at the vehicle base. Each device is connected via an RJ-45 Ethernet interface.

[0041] The automatic inspection and alarm system for segmented insulators provided in this embodiment is further described in detail below.

[0042] Among them, the front-end intelligent sensing device includes a spotlight 4, a first network switch 5, a segment isolator 6 and a camera 7. The camera is connected to the converged network switch 8 in the computer room through the first network switch.

[0043] The segment insulator is an insulating device installed on the contact network of a subway line for electrical segmentation.

[0044] Spotlights are devices that illuminate around poorly lit segment insulators in subway tunnels to enhance video image quality, which can improve the accuracy of the algorithm.

[0045] In the embodiment, the camera adopts an 8-megapixel starlight-level infrared network high-definition high-speed smart ball camera, and adopts a panoramic 2 / 3" progressive Scan CMOS sensor; it has color: 0.002Lux @ (F1.5, AGC ON); black and white: 0.0002Lux @ (F1.5, AGC ON); 0 Lux with IR low illumination performance; it supports setting the cruise route through the preset position, and performs automatic cruise to realize automatic inspection of the monitoring area. It supports automatic conversion of day and night, and has electronic image stabilization, 3D noise reduction, strong light suppression, 16x digital zoom, 25x optical zoom and other performances, and supports IP67; 6000V Lightning protection, surge protection, and surge protection meet the four-level protection standards of GB / T17626.2 / 3 / 4 / 5 / 6. The software integrates open API, ONVIF, ISAPI, third-party management platform access, GB / T28181 protocol, supports EZVIZ access, ISUP and other interfaces, and can be more flexibly applied to subway line environments, passing vehicles and long installation distances. The high-definition high-speed smart ball camera is connected to network switch 1, which is connected to the aggregation network switch.

[0046] The computer room includes a second network switch 9, an algorithm server 10, a data server 11 and an intelligent detection unit. The second network switch is connected to the aggregation network switch. The second network switch is also connected to the algorithm server and the data server. The algorithm server is connected to the intelligent detection unit.

[0047] The AI ​​server is equipped with the deep neural network reasoning hardware acceleration unit and image processing hardware acceleration unit required for AI intelligent analysis. It runs intelligent algorithms based on deep neural networks and image processing technologies, and provides them to programs such as the arc extinguishing angle intelligent detection unit, the guide plate intelligent detection unit, the fastener intelligent detection unit, and the video decoding unit for operation.

[0048] The data server is used to install the system software, inspection and dispatching unit and alarm gateway software. The management software has the functions of equipment management, real-time video preview, alarm display processing, driver and other personnel management, train number management, and notification issuance. The inspection and dispatching unit has the functions of identifying inspection routes, inspection strategies, current preset positions, and automatically opening and closing intelligent detection units. The alarm gateway software has the functions of receiving and storing alarm messages. It has the function of storing alarm images and alarm videos. The data server is connected to the second network switch, which is connected to the aggregation network switch.

[0049] In the embodiment, there are three intelligent detection units, namely, an arc extinguishing angle intelligent detection unit 12, a guide plate intelligent detection unit 13 and a fastener intelligent detection unit 14, and all three intelligent detection units are connected to the algorithm server.

[0050] Specific: The arc extinction angle intelligent detection unit is used to intelligently detect the excessive displacement of the arc extinction angle through the video stream. The arc extinction angle intelligent detection unit adopts the traditional OPENCV image processing technology. By capturing the camera image, dividing the analysis area of ​​interest and calibrating the image horizontally, the image is binarized, the foreground range of the arc extinction angle is outlined, and the vertical tangent position on the left and right of the arc extinction angle is calculated according to the foreground range. The pixel spacing between the two tangents is provided, and the pixel value is converted to millimeters according to the configuration information corresponding to the pixel value and the millimeter. If the calculated distance between the arc extinction angles is greater than ±5mm, the arc extinction angle is considered abnormal, and the vertical tangent is marked on the image, and the abnormal distance information is marked for storage and recording. In order to reduce false alarms, after the first frame of abnormality occurs, one frame of image is extracted per second, and multiple identifications are performed until 10 consecutive frames are abnormal, and the alarm information is immediately pushed to the platform, and a short video of 10 seconds before and after the alarm and the alarm image are intercepted and pushed to the platform at the same time. The arc extinction angle intelligent detection unit is installed on the algorithm server.

[0051] The guide plate intelligent detection unit is used to intelligently detect whether the guide plate is excessively worn through the video stream. The guide plate intelligent detection unit adopts the traditional OPENCV image processing technology. After grabbing the camera image and dividing the analysis area of ​​interest, the image is grayed, filtered, and denoised. According to the lower edge of the guide plate, a rectangular window is taken in the x direction for sliding value, and the grayscale mean in the y direction is calculated. The curve is drawn, and the thickness value of the guide plate is obtained by calculating the trough spacing of the curve. The coordinate array of the upper and lower edges of the guide plate is output, and the pixel value is converted to millimeters according to the configuration information corresponding to the pixel value and millimeter. If the calculated distance between the upper and lower edges of the guide plate is less than 6mm, and the wear area is greater than 1.5cm, it is considered that the guide plate is abnormally worn too much. The upper and lower edges of the guide plate and the area with continuous excessive wear are marked on the image for storage and record. In order to reduce false alarms, after the first abnormal frame appears, one frame of image is extracted per second and recognized multiple times until 10 consecutive frames are abnormal. The alarm information is immediately pushed to the platform, and a short video of 10 seconds before and after the alarm and the alarm image are intercepted and pushed to the platform at the same time. The intelligent detection unit for deflector wear is installed on the algorithm server.

[0052] The fastener intelligent detection unit is used to detect fastener looseness through video stream intelligently. The fastener intelligent detection unit uses two technologies: target detection YOLO neural network and OPENCV image processing technology. Capture camera image I k First, identify the fasteners on the segmented insulator through target detection, and output the fastener coordinate position [f 1k , f 2k , …, f ik , …f nk ], where f 1k A fastener identified by the fastener deep learning model, including the coordinates of the identification box and the width and height. 1k Image C of the fastener in the mid-coordinate intercept image k , C k The image is processed by OPENCV technology to identify the fasteners and the red marking lines on the background plate. If the angle between the two red marking lines exceeds 30°, the fastener is considered loose, and the location area and identification line of the fastener are marked on the image for storage and record. In order to reduce false alarms, after the first abnormal frame appears, one frame of image is extracted per second and recognized multiple times until 10 consecutive frames are abnormal. The alarm information is immediately pushed to the platform, and a short video of 10 seconds before and after the alarm and the alarm image are captured and pushed to the platform at the same time. The fastener intelligent detection unit is installed on the algorithm server.

[0053] The central control center includes a third network switch 15 and a central control client 16. The third network switch is connected to the aggregation network switch, and the central control client is connected to the third network switch.

[0054] The central control client includes a client computer, keyboard, mouse, and three screens. It adopts a one-machine three-screen structure and is used to run the alarm receiving terminal. It has the functions of alarm message voice reminder, alarm detail viewing, alarm image and alarm video viewing, alarm processing opinions and alarm information archiving and alarm statistics, and video preview and analysis functions such as real-time camera video preview, video playback, and video download. The central control client is connected to the third network switch, which is connected to the aggregation network switch.

[0055] When working, front-end intelligent sensing devices are installed on the subway lines facing the segmented insulator equipment, and different preset positions, inspection routes and inspection strategies are set. The algorithm server in the computer room obtains the inspection strategy and current preset position of the ball camera in real time, starts and stops the intelligent monitoring units of different core components, and pulls the video stream of the ball camera in real time. It uses efficient video decoding technology to decompose the video stream into independent image frames, and inputs the image into the corresponding core component intelligent detection unit for preprocessing and extraction of key features, thereby realizing intelligent prediction and abnormal alarm of the status of the core components of the segmented insulator (arc extinguishing angle, guide plate and fasteners). When the intelligent detection unit successfully detects an abnormal situation, the system will immediately activate the automatic alarm mechanism and voice prompt function. On the management client platform of the central control center, the alarm information will pop up quickly and play the real-time video synchronously, so that the on-duty personnel can obtain the on-site situation immediately.

[0056] The specific process of performing automatic inspection and alarm of segmented insulators using the automatic inspection and alarm system of segmented insulators provided in this embodiment includes: Step (1): Prepare the work site environment and conditions.

[0057] (1.1) Install spotlights next to the segment insulator location on the on-site subway line, install a super star light projector on the opposite tunnel, and connect the super star light ball machine to the first network switch.

[0058] (1.2) Set preset positions for the ball camera. According to the image quality requirements of the intelligent detection unit, set the arc extinguishing angle preset position, the first guide plate preset position, the second guide plate preset position, the third guide plate preset position, the fastener preset position, and the fastener and arc extinguishing angle combination preset position. Because the guide plate is too long to be displayed in the same screen, it is cut into multiple parts and multiple preset positions are set.

[0059] Step (2): Set up inspection routes and inspection strategies.

[0060] (2.1) Setting the inspection route of the ball camera: set the first guide plate preset position, the second guide plate preset position and the third guide plate preset position as a first inspection route; set the arc extinguishing angle preset position, the fastener preset position, and the combination of the fastener and the arc extinguishing angle as a second inspection route.

[0061] (2.2) Set the inspection strategy for the first inspection route: stay at each preset position for 5 minutes and inspect once an hour.

[0062] (2.3) Set the inspection strategy for the second inspection route: stay at each preset position for 10 minutes and inspect once every 4 hours.

[0063] The dwell time and inspection frequency can be set according to actual conditions. If there are many preset positions, the dwell time can be shortened.

[0064] Step (3): Turn on the intelligent detection unit for detection.

[0065] After the preparations for step (1) and step (2) are completed, the inspection scheduling unit on the data server and the three intelligent detection units on the algorithm server are turned on. The inspection scheduling unit obtains the inspection route and inspection strategy of the ball camera in real time, obtains the current preset position information, and automatically turns on and off the intelligent detection unit according to the preset position information.

[0066] Specifically: if the current preset position is the first guide plate preset position, the guide plate intelligent detection unit is turned on; if it is a combined preset position, the fastener intelligent detection unit and the arc extinction angle intelligent detection unit are turned on. However, in actual applications, if the image quality can meet the requirements of the three intelligent detection units at the same time, the three intelligent detection units can also be turned on at the same time.

[0067] After the intelligent detection unit obtains the image, it first detects the image quality. If the image quality passes, it will perform intelligent recognition. According to the alarm conditions of the intelligent detection unit, if one of the intelligent detection unit conditions is met, the alarm information will be immediately pushed to the inspection and dispatching unit. After receiving the alarm information, the inspection and dispatching unit will immediately push the message to the central control center client and trigger a voice reminder.

[0068] Step (4): The inspection dispatching unit executes each inspection line cyclically according to step (3) to realize automatic inspection of the segmented insulator.

[0069] Step (5): After the alarm and voice reminder pop up on the central control center client, the on-duty personnel immediately check the alarm details and real-time video and reconfirm the alarm information.

[0070] Step (6): If the second confirmation is a false alarm, immediately cancel the alarm and mark it as a false alarm; if the second confirmation is an alarm, immediately query the system for the nearest repair personnel and phone number, and call the repair personnel through the system. At the same time, query the system for the number of vehicles and driver information passing the fault point, immediately notify the driver of the fault point location information, take emergency treatment measures, temporarily stop the car, etc.; and keep relevant management personnel, real-time emergency plans and notifications, etc.

[0071] Step (7): After the emergency repair personnel successfully repair the fault, they will notify the driver on the system to resume normal operation of the train and notify relevant management personnel to cancel the emergency plan. The on-duty personnel need to cancel the fault alarm on the system and register the fault handling measures.

[0072] Step (8): The system will automatically generate and store a full-process record of the fault from its occurrence to its handling.

[0073] Compared with the prior art, the beneficial technical effects of the automatic inspection and alarm system for segmented insulators provided in this embodiment are: through the collaboration of multiple technologies such as front-end intelligent devices, algorithm servers, intelligent control, and automatic inspection, automatic inspection and intelligent alarm can be realized for multiple core components of subway segmented insulator equipment. By installing a camera on site, setting different inspection strategies and intelligent detection units, customized intelligent detection of multiple core components can be implemented, which can not only realize inspection automation and rapid fault alarm, but also reduce construction costs and labor costs. Through the information platform, equipment failure data can be stored for a long time, and through in-depth mining and analysis of historical data, equipment life can be predicted and maintenance strategies can be optimized, so as to promote the entire subway system to develop in a more intelligent and automated direction. It not only helps to improve the overall efficiency and safe operation of subway operations, but also provides strong technical support and demonstration effects for promoting the comprehensive construction and development of smart cities.

[0074] The above-described embodiment is only a preferred solution of the present invention and does not limit the present invention in any form. There are other variations and modifications without exceeding the technical solution described in the claims.

Claims

1. A segmented insulator automatic inspection and alarm method, characterized in that: include: S1: cutting the guide plate into multiple parts and setting preset positions; S2: Set inspection routes and set inspection strategies on different inspection routes; S3: Obtain the video stream and perform video decoding to detect image quality; S4: Obtain the current inspection strategy and determine the current preset position information; S5: Determine the current detection part according to the preset position information, perform image recognition, judge whether the detection part is abnormal, and alarm when it is abnormal.

2. The automatic inspection and alarm method for segmented insulators according to claim 1 is characterized in that: The method of determining the current detection position based on the preset position information includes: if it is an arc extinguishing angle preset position, detecting whether the arc extinguishing angle displacement is abnormal through the video stream; if it is a guide plate preset position, detecting whether the guide plate wear is abnormal through the video stream; if it is a combined preset position, detecting whether the fasteners are loose and whether the arc extinguishing angle displacement is abnormal at the same time through the video stream.

3. A segmented insulator automatic inspection and alarm method according to claim 1 or 2, characterized in that: The inspection route includes a first inspection route and a second inspection route. The first inspection route includes a first guide plate preset position, a second guide plate preset position and a third guide plate preset position, and the inspection strategy is: stay at each preset position for t minutes and inspect once every n hours; the second inspection route includes an arc extinguishing angle preset position, a fastener preset position, and a fastener and arc extinguishing angle combination preset position, and the inspection strategy is: stay at each preset position for 2t minutes and inspect once every 4n hours.

4. The automatic inspection and alarm method for segmented insulators according to claim 2 is characterized in that: Acquire an image, divide the analysis area of ​​interest and perform horizontal calibration on the image, outline the foreground range of the arc extinction angle, calculate the vertical tangent positions on the left and right of the arc extinction angle based on the foreground range, calculate the pixel spacing between the two tangents, and convert the pixel values ​​into millimeter information. If the distance between the arc extinction angles is greater than the first threshold, the arc extinction angle is abnormal, mark the vertical tangent on the image, and mark the abnormal distance information for storage and recording.

5. A segmented insulator automatic inspection and alarm method according to claim 2 or 4, characterized in that: According to the x-direction of the lower edge of the guide plate, a rectangular window is taken for sliding value, the grayscale mean in the y-direction is calculated, the mean curve is drawn, the trough spacing of the curve is calculated to obtain the guide plate thickness value, and the coordinate array of the upper and lower edges of the guide plate is output, and the pixel value is converted into millimeter information. If the distance between the upper and lower edges of the guide plate is less than the second threshold and the wear area is greater than the third threshold, the guide plate is abnormally worn, and the upper and lower edges of the guide plate and the area with continuous excessive wear are marked on the image for storage and record.

6. A segmented insulator automatic inspection and alarm method according to claim 2 or 4, characterized in that: Capture the image, identify the fasteners on the segmented insulator through target detection, output the coordinate position of the fastener, intercept the image of the fastener according to the coordinate position of the fastener, identify the red marking line on the fastener and the red marking line on the background plate, if the angle between the two marking lines exceeds the fourth threshold, the fastener is loose.

7. A segmented insulator automatic inspection and alarm method according to claim 1, 2 or 4, characterized in that: When performing anomaly detection, if an abnormality is detected in a frame of image, one frame of image is extracted every second for identification. If abnormalities are found in m consecutive frames, an abnormality occurs in the detection part, an alarm message is sent, and screenshots of the video and alarm image m seconds before and after the alarm are taken.

8. A segmented insulator automatic inspection and alarm method according to claim 1, 2 or 4, characterized in that: The abnormal alarm information is reconfirmed. If the second confirmation is an alarm, the train number and staff information passing the fault point are queried, the fault point information is sent to the staff, and the corresponding emergency handling measures are implemented.

9. An automatic inspection and alarm system for segmented insulators, using an automatic inspection and alarm method for segmented insulators according to any one of claims 1 to 8, characterized in that: include: Front-end intelligent sensing equipment sets inspection routes and inspection strategies to obtain video data of core components of segmented insulators; The inspection dispatch unit identifies the inspection route and inspection strategy, automatically opens and closes the corresponding intelligent detection unit according to the current preset position, and decomposes the video data into independent image frames; Intelligent detection unit detects image quality, extracts key features from image frames, and makes intelligent predictions and abnormal alarms on the status of core components of segmented insulators; The central control client issues an alarm and simultaneously plays the real-time video collected by the front-end intelligent sensing device.

10. The automatic inspection and alarm system for segmented insulators according to claim 9 is characterized in that: The intelligent detection unit comprises: The arc extinction angle intelligent detection unit uses the OPENCV image processing algorithm to process the image frames of the special area, calculate the distance between the arc extinction angles, and determine whether the arc extinction angle is abnormal; The guide plate intelligent detection unit grayscales and performs noise reduction processing on the image, calculates the guide plate thickness value, and determines whether the guide plate wear is abnormal; The fastener intelligent detection unit uses the target detection YOLO neural network and OPENCV image processing algorithm to determine whether the angle between the red line of the fastener and the red line of the background plate is greater than the threshold.

Images