GIS disconnecting link state detection system based on machine learning

Abstract

The invention discloses a GIS disconnecting link state monitoring system based on machine learning. The GIS disconnecting link state monitoring system is composed of a detection network and a data network. The detection network comprises a plurality of cameras, a hard disk video recorder which is electrically connected with the cameras, a media switch which is in network connection with the hard disk video recorder, a data switch which is in network connection with the media switch, and a memory, a device operation platform and a GIS disconnecting link state monitoring device which are respectively in network connection with the data switch; the data network comprises a measurement and control device which is electrically connected with the GIS disconnecting link state monitoring device, a monitoring master station which is in network connection with the measurement and control device, a data center switch which is in network connection with the data switch, a data platform which is in network connection with the data center switch, and a work station which is in communication connection with the data platform. According to the method, the detection network is simple to construct, the factors, including vibration, creepage, ionizing radiation, transmission and the like, possibly influencing the precision are judged and eliminated in a target detection mode, and the detection accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of knife switch status monitoring, and in particular to a GIS knife switch status detection system based on machine learning. Background Art

[0002] GIS knife switch, also known as disconnector, is a key device in GIS system, used to establish or disconnect circuit connection in power system. GIS equipment uses sulfur hexafluoride (SF6) gas with excellent insulation and arc extinguishing performance as insulation and arc extinguishing medium, and seals all high-voltage electrical components in grounded metal cylinder, but the GIS knife switch equipment in UHV substation is very large and dangerous. Usually, maintenance workers first need to use climbing tools to climb the metal cylinder, then use endoscopes to insert observation holes to observe the internal state of GIS knife switch equipment, and finally perform maintenance work, which is risky and dangerous for GIS knife switches in operation and UHV. Therefore, it is very necessary to remotely determine the opening and closing state of GIS knife switch and track the movement process.

[0003] GIS knife switch is an important component of gas-insulated metal-enclosed switchgear (GIS), which is used to realize the circuit connection and disconnection of ultra-high voltage substations. The moving contact and the static contact are the core parts of the knife switch, which are used to realize the connection and disconnection of the circuit. The moving contact is connected to the operating mechanism and can move between the static contacts. When the moving contact contacts the static contact, the circuit is closed; otherwise, the circuit is disconnected. If it is impossible to confirm whether the three contacts are correctly disconnected or closed, it will directly affect the next operation. Existing detection devices usually use vibration or photoelectric sensors to confirm whether the three contacts are correctly disconnected or closed, but they cannot directly obtain the actual images of the moving contact and the static contact for maintenance personnel to view the status details. Summary of the invention

[0004] The purpose of the present invention is to provide a GIS knife switch status detection system based on machine learning, which can improve the accuracy of identifying the opening and closing status of the knife switch, capture the opening and closing timing and various parameters of the knife switch, realize unmanned maintenance, and check the knife switch status inside the GIS knife switch equipment.

[0005] In order to solve the above technical problems, the technical solution of the present invention is: a GIS switch state detection system based on machine learning, the GIS switch state detection system based on machine learning is composed of a detection network and a data network; The detection network consists of several cameras. A hard disk video recorder is electrically connected to the camera. Media switch, which is connected to the DVR network, A data switch connected to the media switch network, The storage device, device operation platform and GIS switch status monitoring device are respectively connected to the data switch network; The data network includes a measurement and control device, which is electrically connected to the GIS switch status monitoring device. The monitoring master station is connected to the measurement and control device network. A data center switch connected to a data switch network, The data platform is connected to the data center switch network. A workstation is communicatively connected with the data platform.

[0006] Preferably, there are three cameras, all of which are arranged in the GIS knife switch device.

[0007] Preferably, the side wall of the GIS knife switch device is provided with three observation holes, and three cameras are respectively plugged into the three observation holes.

[0008] Preferably, the GIS knife switch device is a three-phase knife switch, and three cameras respectively collect images of the three phase knife switches.

[0009] Preferably, the GIS knife switch status monitoring device is provided with 8 groups of IO output interfaces from left to right, and the different opening and closing states of groups 1-6 of IO output interfaces represent different states of the GIS knife switch equipment, and the states of groups 7-8 of IO output interfaces represent the health state of the GIS knife switch status monitoring device.

[0010] Preferably, the GIS switch status monitoring device comprises a target detector module, an image processor module, a task manager module, an IO controller module, a data interface module and a data analysis module, and each module is controlled and operated through a program bus; The task manager module is used to schedule tasks and transfer data for other modules; The image processor module is used to optimize the images of GIS switch equipment acquired in real-time high-definition digital video streams; The target detector module is used to detect the image of the GIS switch device optimized by the image processor module and determine the relative position of the switch; The IO controller module is used to output the monitoring results of the GIS switch status monitoring device and the abnormal status of the GIS switch status monitoring device; The data interface module is used to connect to the data network, obtain the initial configuration of the GIS knife switch status monitoring device, report the error data of the abnormal state of the GIS knife switch status monitoring device and report the dynamic report of the knife switch switching; The data analysis module is used to perform data analysis on the detection results output by the target detection module and generate a dynamic report on the switching of the knife switch.

[0011] Preferably, the GIS switch status monitoring device also performs self-inspection and initial configuration before performing the monitoring task.

[0012] Preferably, when the GIS switch status monitoring device performs a monitoring task, first, the data and task scheduling table are read into the program bus, and then the task manager performs task scheduling and data transmission for other modules in turn through the program bus. After completing the monitoring task, the IO controller module, the data interface module and the data analysis module are respectively called to output the monitoring task result. The specific implementation steps are as follows: Step 1: Obtain the initial configuration of the GIS switch status monitoring device; If the initial state of the IO output interface configuration is None, proceed to step 2; If the initial state of the IO output interface configuration is yes, execute step 5; Step 2: Obtain any frame of image of the GIS switch device through a hard disk recorder and obtain image P1 after optimization using an image processor module; Step 3: Use the object detector module to detect the image P1 obtained in step 2 and output the detection result; If a knife switch is detected in the image P1 within the preset time threshold t1, step 4 is executed; If no knife switch is detected in the image P1 or the image P1 is empty within the preset time threshold t1, step 2 is executed: reacquiring the image P1; If the knife switch is still not detected in the image P1 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 4: Analyze the target detection result in step 3 using the data analysis module, and set the analysis result as the initial state S1 of the knife switch; Step 5: Based on the initial state S1 of the knife switch in step 4, the states of the 1-6 groups of IO output interfaces of the GIS knife switch state monitoring device are also synchronously set to S1; Step 6: Continuously obtain a frame image of the GIS switch device in the video stream through the hard disk recorder and obtain image P2 after optimization using the image processor module; If the image P2 in the video stream can be continuously obtained, execute step 7; If the image P2 in the video stream cannot be continuously obtained, the abnormal state of the GIS switch status monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 7: Continue to detect image P2 using the object detector module; If a knife switch is detected in the image P2 within the preset time threshold t1, step 8 is executed to wait for the external trigger mode Tr1 to be triggered, and the data of the image P2 is transferred to the new monitoring thread as the image P; If the knife switch is still not detected in the image P2 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 8: After the switching command of the knife switch is monitored by the data interface module, step 9 is executed, and the image P is detected by the target detector module; If the knife switch is detected to be displaced, execute step 9; If no displacement of the knife switch is detected, execute step 7 and continue to acquire image P; Step 9: According to the switching instruction of the knife switch, the external trigger mode Tr1 is triggered, and the trigger time is recorded as T1, then the knife switch state of any phase in the monitoring task is set to 1; when the external trigger mode is not triggered, the target detector module in step 8 detects that the knife switch is displaced, then the knife switch state of any phase in the monitoring task is also set to 1, which has the same effect as the external trigger mode, and realizes the switching function of the automatic monitoring knife switch; Step 10: Continuously obtain a frame image of the GIS switch device in the video stream through the hard disk recorder and obtain image P3 after optimization using the image processor module; Step 11: Determine whether the video stream in step 10 is interrupted; If the video stream is not interrupted, execute step 12; If the video stream is interrupted, the IO controller module is called to output the abnormal state of the GIS switch status monitoring device and report the error data through the data interface module; Step 12: Use the object detector module to continuously detect image P3 and output the detection results; Step 13: Get the current time T, and serialize the detection results in step 12 to obtain the analysis results; Step 14: Determine whether the difference between the current time T and the trigger time T1 reaches the expected switching time length N of the knife switch; If the expected switching time length N of the knife switch has been reached, a trigger signal is sent to the data analysis module, and a switching dynamic report R of the knife switch is generated according to the analysis result obtained in step 13; If the expected switching time length N of the knife switch has not been reached, execute step 12 to reacquire the image P3; Step 15: Summarize the switching dynamic reports R of the knife switches corresponding to each phase to obtain the total switching dynamic report of the knife switches , get the current switch state S2; Step 16: Output the current switch state S2 by calling the IO controller module, and report the switch switching dynamic total report through the data interface module , complete the status monitoring task of GIS switch equipment.

[0013] Compared with the prior art, the present invention has the following beneficial effects: (1) On-site implementation: The detection network of the present invention is simple to construct, and there is no need to disassemble any major parts, which will not cause negative impacts such as downtime; (2) Detection accuracy: The present invention uses target detection to basically eliminate all possible factors that may affect accuracy, including vibration, creepage, ionizing radiation, transmission, etc., which greatly improves the accuracy of the detection results. It can directly access the existing data system to provide marginal data, visualize the evaluation results, and provide a user-friendly detection report for manual secondary analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 It is a schematic diagram of the detection system of the present invention; Figure 2 It is a schematic diagram of the IO output interface of the GIS switch status monitoring device in the present invention; Figure 3 This is a schematic diagram of camera installation in the present invention; Among them, 1-GIS knife switch equipment, 2-observation hole, 3-camera; Figure 4 It is a module schematic diagram of the GIS switch status monitoring device in the present invention; Among them, 10-target detector module, 20-image processor module, 30-task manager module, 40-IO controller module, 50-data interface module, 60-data analysis module, 70-program bus; Figure 5 It is a monitoring task flow diagram of the GIS switch status monitoring device in the present invention; Figure 6 It is a schematic diagram of the self-check and initial configuration process of the GIS switch status monitoring device of the present invention. DETAILED DESCRIPTION

[0016] The following is a further detailed description of the present invention in conjunction with the accompanying drawings. The terminal technical solution of the embodiment of the present invention is clearly and completely described. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] like Figure 1 As shown, the present invention discloses a GIS switch state detection system based on machine learning, which is composed of a detection network and a data platform; The detection network includes three cameras 3, namely, phase A camera 3, phase B camera 3 and phase C camera 3.

[0018] The three cameras 3 are connected to the NVR hard disk video recorder via a highly shielded coaxial cable, which can improve the reliability and stability of image acquisition. The three cameras 3 provide three real-time high-definition video streams to the NVR hard disk video recorder respectively.

[0019] The A-phase camera 3, the B-phase camera 3, and the C-phase camera 3 are all arranged in the GIS knife switch device 1. Specifically, the GIS knife switch device 1 is a three-phase knife switch. The side wall of the GIS knife switch device 1 is provided with three observation holes 2. The A-phase camera 3, the B-phase camera 3, and the C-phase camera 3 are respectively inserted into the three observation holes 2 and rotated and fixed, respectively corresponding to collecting the images of the three phase knife switches of the GIS knife switch device 1 and transmitting them back to the NVR hard disk recorder, such as Figure 3 shown.

[0020] The NVR hard disk video recorder is connected to the video stream interface of the media switch through an RJ45 network cable. The NVR hard disk video recorder provides the media switch with a transcoded real-time high-definition digital video stream (RTSP video stream).

[0021] The media switch is connected to the data switch via an RJ45 network cable to isolate subnets and exchange data with the data switch.

[0022] The memory, device operating platform and GIS knife switch status monitoring device are also connected to the data switch network through RJ45 network cables respectively; the GIS knife switch status monitoring device first obtains the real-time high-definition digital video stream through the data switch, then monitors the knife switch status of the GIS knife switch device 1, and finally stores the monitoring results of the knife switch status in the memory; the GIS knife switch status monitoring device can be initially configured and checked through the device operating platform.

[0023] When the NVR hard disk recorder, storage device, device operating platform and GIS knife switch status monitoring device are in the same local area network, data can be directly exchanged with each other. The GIS knife switch status monitoring device can directly obtain real-time high-definition digital video stream through the RTSP video port of the NVR hard disk recorder to complete the status monitoring task of the GIS knife switch equipment 1.

[0024] The data network includes a measurement and control device, which is connected to the IO output interface of the GIS switch status monitoring device through a jumper; Figure 2 As shown, the GIS knife switch status monitoring device has 8 groups of IO output interfaces, and the monitoring results of the GIS knife switch status monitoring device are presented in real time by the status of the 8 groups of IO output interfaces. The status of the IO output interface is the on-off of the IO output interface, and the on-off of the IO output interface is captured by the lower connected measurement and control device. The status of the 1-6 groups of IO output interfaces on the left side presents the current on-off status of the GIS knife switch device 1. The combination of different states of the 1-6 groups of IO output interfaces on the left side represents the different states of the current knife switch. The status of the 7-8 groups of IO output interfaces on the right side presents the health status of the GIS knife switch status monitoring device. The abnormal health status includes sensor failure and device abnormality. The measurement and control device collects the data of the health status of the GIS knife switch status monitoring device and the data of the on-off status of the GIS knife switch device 1 through the status of the 8 groups of IO output interfaces, and reports the data to the monitoring main station. The monitoring main station is connected to the measurement and control device through an RJ45 network cable.

[0025] The data center switch is connected to the data switch via an RJ45 network cable, which can expand the network scale and realize data sharing and interaction.

[0026] The data platform is connected to the data center switch via an RJ45 network cable for data transmission.

[0027] The workstation calls the exposed interface of the data platform through a network connection to query data.

[0028] like Figure 4 As shown, the GIS switch state monitoring device includes a target detector module 10, an image processor module 20, a task manager module 30, an IO controller module 40, a data interface module 50 and a data analysis module 60, and each module is controlled and operated through a program bus 70; The task manager module 30 is used to perform task scheduling and data transmission for other modules; The image processor module 20 is used to optimize the image of the GIS switch device 1 obtained in the real-time high-definition digital video stream; The target detector module 10 is used to detect the image of the GIS knife switch device 1 optimized by the image processor module 20 to determine the relative position of the knife switch; The IO controller module 40 is used to output the monitoring result of the GIS switch status monitoring device and the abnormal status of the GIS switch status monitoring device; The data interface module 50 is used to connect to the data network, obtain the initial configuration of the GIS switch status monitoring device, report the error data of the abnormal state of the GIS switch status monitoring device and report the switching dynamic report of the switch; The data analysis module 60 is used to perform data analysis on the detection results output by the target detection module and generate a dynamic report on the switching of the knife switch.

[0029] The GIS switch status monitoring device also includes an embedded memory (EMMC), in which all functional modules and configuration files are deployed. The configuration files include a main configuration file and a backup configuration file, which are synchronized to the device operating platform through the network for storage. Figure 6 As shown, the GIS knife switch status monitoring device needs to perform self-check and initial configuration before executing the monitoring task to complete the task of parameter acquisition and health check. The status monitoring task of the GIS knife switch device 1 will be executed only after the parameters are correctly acquired and the self-check is passed. The specific implementation steps are as follows: Step 1: Start the OTA upgrade module and configuration module. The OTA upgrade module opens the upgrade service port to receive the program firmware of the device operating platform and update the execution code binary file and program library file of the status monitoring task of the GIS knife switch equipment in the EMMC in real time. The configuration module obtains and verifies the main configuration file in the GIS knife switch status monitoring device. The main configuration file contains the RTSP video stream address, the expected switching time length N of the knife switch, the external trigger mode Tr1, the time threshold t1, the initial state of the IO output interface configuration, the IP address, the service port, the model file, the target detection threshold and the heartbeat keep-alive period.

[0030] The main configuration file is stored in the EMMC of the GIS switch status monitoring device by a 128-bit hash value. Before obtaining the main configuration file, it is necessary to verify the hash value stored in the EMMC with the real-time calculated hash value of the application configuration file to determine whether the main configuration file is damaged.

[0031] If the main configuration file is damaged, go to step 2; If the main configuration file is not damaged, proceed to step 4.

[0032] Step 2: Obtain and verify the backup configuration file in the GIS switch status monitoring device; If the backup configuration file is damaged, go to step 3; If the backup configuration file is not damaged, go to step 4.

[0033] Step 3: Obtain and verify the backup configuration file in the device operating platform; If the backup configuration file in the device operation platform is damaged, the abnormal status of the GIS knife switch status monitoring device will be output through the status of the 7-8 group IO output interface on the right side of the GIS knife switch status monitoring device and reported to the device operation platform; If the backup configuration file in the device operating platform is not damaged, proceed to step 4.

[0034] Step 4: Load the configuration file that has not been damaged in steps 1-3 into the program memory.

[0035] Step 5: Use the heartbeat keep-alive cycle in the configuration file loaded in step 4 as an input parameter to start the keep-alive module. The keep-alive module is used to monitor the life cycle of the GIS knife switch status monitoring device. The device operating platform can detect the online activity of the GIS knife switch status monitoring device through the keep-alive module, and continuously obtain the real-time heartbeat data during its operation.

[0036] Step 6: According to the configuration file loaded in step 4, the GIS switch status monitoring device performs a health self-check, and after obtaining the health self-check result, execute step 7. The health self-check includes: IP address check, service port check, whether the RTSP video stream address is accessible, and whether the IO output interface is normal.

[0037] Step 7: Analyze the results of the health self-examination in step 6; If the self-test passes, go to step 8; If the self-test fails, go to step 9.

[0038] Step 8: Use the RTSP video stream address, the expected switching time length N of the knife switch, the external trigger mode Tr1, the time threshold t1, and the initial state of the IO output interface configuration in the configuration file loaded in step 4 as input parameters to start the monitoring task of the GIS knife switch status monitoring device; Step 9: Set the 7-8 groups of IO output interfaces on the right side of the GIS knife switch status monitoring device to an abnormal state, and report the abnormal state of the GIS knife switch status monitoring device to the device operation platform.

[0039] The self-check and initial configuration of the GIS switch status monitoring device is a pre-task before the GIS switch status monitoring device performs the monitoring task, which can improve the accuracy of the parameters and static files required for the monitoring task and improve the overall stability of the GIS switch status monitoring device. In addition, the configuration module and OTA upgrade module are started during the self-check and initial configuration process, so that the device can switch the configuration according to different application scenarios and modify and debug the configuration in real time.

[0040] like Figure 5As shown, when the GIS switch status monitoring device performs the monitoring task, first, the data and task scheduling table are read into the program bus 70, and then the task manager module 30 performs task scheduling and data transmission for other modules in turn through the program bus 70. After completing the monitoring task, the IO controller module 40, the data interface module 50 and the data analysis module 60 are respectively called to output the monitoring task result. The specific implementation steps are as follows: Step 1: The task manager module 30 calls the data interface module 50 to obtain the initial configuration of the GIS switch status monitoring device through the data center switch or the configuration file of the GIS switch status monitoring device itself; including the RTSP video stream address, the expected switching time length N of the switch, the external trigger mode Tr1, the time threshold t1, the network configuration information of the GIS switch status monitoring device itself, and the initial state of the IO output interface configuration; If the initial state of the IO output interface configuration is None, proceed to step 2; If the initial state of the IO output interface configuration is yes, execute step 5; Step 2: Obtain any frame of the GIS switch device 1 through the RTSP video stream specified by the NVR hard disk recorder and obtain the image P1 after optimization using the image processor module 20 Step 3: Detect the image P1 acquired in step 2 using the target detector module 10, and output the detection result to the data analysis module 60; If a knife switch is detected in the image P1 within the preset time threshold t1, step 4 is executed; If no knife switch is detected in the image P1 or the image P1 is empty within the preset time threshold t1, step 2 is executed: reacquiring the image P1; If the knife switch is still not detected in the image P1 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module 40 and the error data is reported through the data interface module 50; Step 4: Analyze the target detection result in step 3 using the data analysis module 60, and set the analysis result as the initial state S1 of the GIS switch device 1; Step 5: Based on the initial state S1 of the knife switch in step 4, the states of the 1-6 groups of IO output interfaces on the left side of the GIS knife switch state monitoring device are also synchronously set to S1; Step 6: Continuously obtain a frame image of the GIS switch device 1 in the RTSP video stream through the NVR hard disk recorder and obtain the image P2 after optimization using the image processor module 20; If the image P2 in the RTSP video stream can be continuously obtained, execute step 7; If the image P2 in the RTSP video stream cannot be continuously obtained, it indicates that the video stream is in a disconnected state or an abnormal situation exists, then the abnormal state of the GIS switch state monitoring device is output by calling the IO controller module 40 and the error data is reported through the data interface module 50; Step 7: Continuously detect the image P2 using the object detector module 10; If a knife switch is detected in the image P2 within the preset time threshold t1, step 8 is executed to wait for the external trigger mode Tr1 to be triggered, and the data of the image P2 is transferred to the new monitoring thread as the image P; If the knife switch is still not detected in the image P2 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module 40 and the error data is reported through the data interface module 50; Step 8: After the switching command of the knife switch is monitored by the data interface module 50, step 9 is executed, and the target detector module 10 is used to detect the image P simultaneously; If the knife switch is detected to be displaced, execute step 9; If no displacement of the knife switch is detected, execute step 7 and continue to acquire image P; Step 9: trigger the external trigger mode Tr1 according to the switching instruction of the knife switch, and record the trigger time as T1, then the knife switch S* state of any phase in the monitoring task is 1; when the external trigger mode is not triggered, the target detector module 10 in step 8 detects that the knife switch is displaced, then the knife switch S* state of any phase in the monitoring task is also 1, which has the same effect as the external trigger mode, and realizes the switching function of the automatic monitoring knife switch; the S*= SA||SB||SC, that is, the knife switch state of the A / B / C phase; Step 10: continuously obtain a frame image of the GIS switch device 1 in the RTSP video stream through the NVR hard disk recorder and obtain the image P3 after optimization using the image processor module 20; Step 11: Determine whether the RTSP video stream in step 10 is disconnected; If the RTSP video stream is not interrupted, execute step 12; If the RTSP video stream is interrupted, the abnormal state of the GIS switch status monitoring device is output by calling the IO controller module 40 and the error data is reported through the data interface module 50; Step 12: Continuously detect the image P3 using the object detector module 10 and output the detection result to the data analysis module 60; Step 13: Obtain the current time T, and use the data analysis module 60 to serialize the detection results in step 12 to obtain the analysis results; Step 14: Determine whether the difference between the current time T and the trigger time T1 reaches the expected switching time length N of the knife switch; If the expected switching time length N of the knife switch has been reached, a trigger signal is sent to the data analysis module 60, and a switching dynamic report R of the knife switch is generated according to the analysis result obtained in step 13; If the expected switching time length N of the knife switch has not been reached, execute step 12 to reacquire the image P3; Step 15: Summarize the switching dynamic reports R of the knife switches corresponding to each phase to obtain the total switching dynamic report of the knife switches , get the current switch state S2; Step 16: Output the current switch state S2 by calling the IO controller module, and report the switch switching dynamic total report through the data interface module , complete the status monitoring task of GIS switch equipment.

[0041] The parts not involved in the present invention are the same as the prior art or are implemented by using the prior art.

[0042] The above contents are further detailed descriptions of the present invention in combination with specific implementation methods, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field of the present invention, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the scope of protection of the present invention.

Claims

1. A GIS switch status detection system based on machine learning, characterized by: The GIS switch status detection system based on machine learning consists of a detection network and a data network; The detection network consists of several cameras. A hard disk video recorder is electrically connected to the camera. Media switch, which is connected to the DVR network, A data switch connected to the media switch network, The storage device, device operation platform and GIS switch status monitoring device are respectively connected to the data switch network; The data network includes a measurement and control device, which is electrically connected to the GIS switch status monitoring device. The monitoring master station is connected to the measurement and control device network. A data center switch connected to a data switch network, The data platform is connected to the data center switch network. A workstation is communicatively connected with the data platform.

2. The GIS switch status detection system based on machine learning according to claim 1 is characterized in that: There are three cameras, all of which are arranged in the GIS knife switch equipment.

3. The GIS switch status detection system based on machine learning according to claim 2 is characterized in that: The side wall of the GIS knife switch device is provided with three observation holes, and three cameras are respectively plugged into the three observation holes.

4. The GIS switch status detection system based on machine learning according to claim 2 is characterized in that: The GIS switch device is a three-phase switch, and three cameras respectively collect images of the three phase switches.

5. The GIS switch status monitoring system based on machine learning according to claim 1 is characterized in that: The GIS switch status monitoring device is provided with 8 groups of IO output interfaces from left to right. The different opening and closing states of groups 1-6 of IO output interfaces represent different states of the GIS switch equipment, and the states of groups 7-8 of IO output interfaces represent the health state of the GIS switch status monitoring device.

6. The GIS switch status detection system based on machine learning according to claim 1 is characterized in that: The GIS switch status monitoring device comprises a target detector module, an image processor module, a task manager module, an IO controller module, a data interface module and a data analysis module, and each module is controlled and operated through a program bus; The task manager module is used to schedule tasks and transfer data for other modules; The image processor module is used to optimize the images of GIS switch equipment acquired in real-time high-definition digital video streams; The target detector module is used to detect the image of the GIS knife switch device optimized by the image processor module, and determine whether there is a knife switch in the image; The IO controller module is used to output the monitoring results of the GIS switch status monitoring device and the abnormal status of the GIS switch status monitoring device; The data interface module is used to connect to the data network, obtain the initial configuration of the GIS knife switch status monitoring device, report the error data of the abnormal state of the GIS knife switch status monitoring device and report the dynamic report of the knife switch switching; The data analysis module is used to perform data analysis on the detection results output by the target detection module and generate a dynamic report on the switching of the knife switch.

7. The GIS switch status detection system based on machine learning according to claim 6 is characterized by: The GIS switch status monitoring device also needs to perform self-checking and initial configuration before performing the monitoring task.

8. The GIS switch status detection system based on machine learning according to claim 6 is characterized by: When the GIS switch status monitoring device performs the monitoring task, first, the data and task scheduling table are read into the program bus, and then the task manager performs task scheduling and data transmission for other modules in turn through the program bus. After completing the monitoring task, the IO controller module, the data interface module and the data analysis module are respectively called to output the monitoring task results. The specific implementation steps are as follows: Step 1: Obtain the initial configuration of the GIS switch status monitoring device; If the initial state of the IO output interface configuration is None, proceed to step 2; If the initial state of the IO output interface configuration is yes, execute step 5; Step 2: Obtain any frame of image of the GIS switch device through a hard disk recorder and obtain image P1 after optimization using an image processor module; Step 3: Use the object detector module to detect the image P1 obtained in step 2 and output the detection result; If a knife switch is detected in the image P1 within the preset time threshold t1, step 4 is executed; If no knife switch is detected in the image P1 or the image P1 is empty within the preset time threshold t1, step 2 is executed: reacquiring the image P1; If the knife switch is still not detected in the image P1 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 4: Analyze the target detection result in step 3 using the data analysis module, and set the analysis result as the initial state S1 of the knife switch; Step 5: Based on the initial state S1 of the knife switch in step 4, the states of the 1-6 groups of IO output interfaces of the GIS knife switch state monitoring device are also synchronously set to S1; Step 6: Continuously obtain a frame image of the GIS switch device in the video stream through the hard disk recorder and obtain image P2 after optimization using the image processor module; If the image P2 in the video stream can be continuously obtained, execute step 7; If the image P2 in the video stream cannot be continuously obtained, the abnormal state of the GIS switch status monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 7: Continue to detect image P2 using the object detector module; If a knife switch is detected in the image P2 within the preset time threshold t1, step 8 is executed to wait for the external trigger mode Tr1 to be triggered, and the data of the image P2 is transferred to the new monitoring thread as the image P; If the knife switch is still not detected in the image P2 when the time threshold t1 is reached, the abnormal state of the GIS knife switch state monitoring device is output by calling the IO controller module and the error data is reported through the data interface module; Step 8: After the switching command of the knife switch is monitored by the data interface module, step 9 is executed, and the image P is detected by the target detector module; If the knife switch is detected to be displaced, execute step 9; If no displacement of the knife switch is detected, execute step 7 and continue to acquire image P; Step 9: According to the switching instruction of the knife switch, the external trigger mode Tr1 is triggered, and the trigger time is recorded as T1, then the knife switch state of any phase in the monitoring task is set to 1; when the external trigger mode is not triggered, the target detector module in step 8 detects that the knife switch is displaced, then the knife switch state of any phase in the monitoring task is also set to 1, which has the same effect as the external trigger mode, and realizes the switching function of the automatic monitoring knife switch; Step 10: Continuously obtain a frame image of the GIS switch device in the video stream through the hard disk recorder and obtain image P3 after optimization using the image processor module; Step 11: Determine whether the video stream in step 10 is interrupted; If the video stream is not interrupted, execute step 12; If the video stream is interrupted, the IO controller module is called to output the abnormal state of the GIS switch status monitoring device and report the error data through the data interface module; Step 12: Use the object detector module to continuously detect image P3 and output the detection results; Step 13: Get the current time T, and serialize the detection results in step 12 to obtain the analysis results; Step 14: Determine whether the difference between the current time T and the trigger time T1 reaches the expected switching time length N of the knife switch; If the expected switching time length N of the knife switch has been reached, a trigger signal is sent to the data analysis module, and a switching dynamic report R of the knife switch is generated according to the analysis result obtained in step 13; If the expected switching time length N of the knife switch has not been reached, execute step 12 to reacquire the image P3; Step 15: Summarize the switching dynamic reports R of the knife switches corresponding to each phase to obtain the total switching dynamic report of the knife switches , get the current switch state S2; Step 16: Output the current switch state S2 by calling the IO controller module, and report the switch switching dynamic total report through the data interface module , complete the status monitoring task of GIS switch equipment.

Images