Converter station personnel operation risk identification method and system based on machine vision

Abstract

The invention discloses a converter station personnel operation risk identification method and system based on machine vision. The method comprises the following steps: acquiring motion attitude images of an operator at different angles in a converter station scene as an image data set; the collected sample pictures are preprocessed after being screened, expanded and sorted, and a training set is formed; training a gait model for identity authentication of the converter station operating personnel by using the preprocessed training set image; manually calibrating the operation area or the unauthorized area according to the preset position screenshot of the camera during operation; the method comprises the following steps: automatically calculating a current operation area or an authorized area by presetting a preset position of a camera during operation of converter station personnel; when a person works, the identity of the person and the working area range are determined through images collected by a camera, and when the working risk is recognized, an alarm is automatically given, and a log is recorded and uploaded to a background system. According to the invention, the operation site of the converter station can be monitored in real time, various abnormal conditions can be quickly responded, and the safety supervision efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the field of machine vision, and in particular relates to a method and system for identifying operational risks of converter station personnel based on machine vision. Background Art

[0002] The current on-site operation management of UHV converter stations faces some challenges, especially in terms of obvious deficiencies in intelligent applications, which leads to excessive reliance on manual operations for the identification and control of operational risks. In this case, the inadequacy of intelligent means makes machine control far less common than manual control, increasing the reliance on human resources. In addition, key links such as the qualification review and site entry permit of operators still need to be manually confirmed, which not only consumes a lot of manpower costs, but also has unsatisfactory efficiency.

[0003] As for the demarcation of the operating personnel area, it needs to be re-demarcated every time the camera adjusts the angle or the preset position offset changes, which will also lead to a huge waste of human resources and is not conducive to the monitoring of the converter station operators and the identification of operational risks.

[0004] Furthermore, once the operators enter the site, their safety supervision is almost entirely dependent on the on-site supervision of safety managers. This dependence limits the efficiency of safety supervision and makes it difficult to ensure that safety hazards can be discovered and handled in a timely manner. Especially when complex operations such as large-scale power outages or annual maintenance are carried out at converter stations, the on-site safety supervision work becomes more arduous due to the heavy tasks and wide scope of work. The existing supervision system often cannot meet actual needs, and the timely discovery and handling of safety hazards has become a prominent problem. Therefore, it is urgent to introduce more advanced intelligent technologies and methods to improve the safety and efficiency of on-site operations. Summary of the invention

[0005] Purpose of the invention: In view of the problems existing in the prior art, the present invention proposes a method and system for identifying the operational risks of converter station personnel based on machine vision, which can monitor the converter station operation site in real time, respond quickly to various abnormal situations, and improve the efficiency of safety supervision.

[0006] Technical solution: The method for identifying the risk of converter station personnel operations based on machine vision described in the present invention comprises the following steps:

[0007] Collect images of workers’ motion postures at different angles in different scenes at the converter station as image datasets;

[0008] The sample images in the collected image data set are screened, expanded and sorted, and then preprocessed to form a preprocessed training set;

[0009] Using the preprocessed training set images to train a gait model for identification of converter station operators, a trained gait model is obtained;

[0010] In response to user operations, the operation area or unauthorized area is calibrated according to the screenshots of the camera preset positions during operation;

[0011] By presetting the camera preset position, the current operation area or authorized area is calculated according to the original demarcated area and projection matrix;

[0012] The identity of personnel and the scope of the working area are determined by collecting images through cameras, and the trained gait model is used to identify the operating risks of personnel in the converter station.

[0013] Furthermore, the uniform resolution of the images of motion postures at different angles is 1920×1080.

[0014] Furthermore, the preprocessing includes rotation, scaling, brightness adjustment, and removal of blurred, overexposed, and underexposed images.

[0015] Furthermore, the implementation process of using the preprocessed training set images to train the gait model for identity authentication of converter station operators is as follows:

[0016] StrongSORT is used as a personnel tracking model to detect the personnel operation video collected by the converter station, and the personnel ID information and coordinate information of each frame image are obtained. The image containing a single person is cropped according to the personnel coordinate information; the personnel coordinate information includes the upper left corner, width, and height information of the personnel detection box;

[0017] A single person image is used as the input of the portrait segmentation model PP-HumanSeg to obtain the human silhouette contour information; a gait model is trained based on the human silhouette contour information.

[0018] Furthermore, the process of calibrating the working area or unauthorized area according to the camera preset position screenshot during the operation is as follows:

[0019] By adjusting the camera angle to aim at the work scene of the personnel, the current screenshot is saved and the current angle and focal length of the camera are saved to the preset position number parameters; when the situation of the camera at this position is checked according to user needs, the camera can be quickly moved to this position by simply calling the preset position number;

[0020] The marked working area or unauthorized area is a polygonal irregular area, and the open source Labelme software is used to save the area coordinates as a json file.

[0021] Furthermore, the process of calculating the current operating area or the authorized area according to the original demarcated area and the projection matrix is ​​as follows:

[0022] Use the Disk feature point extraction model to extract feature points from the original image preset position screenshot A1 and the currently called preset position screenshot A2, use the feature point matching method to retain the feature points corresponding to A1 and A2, and calculate the projection transformation matrix of A1 and A2 based on the corresponding feature points; according to the operation area or unauthorized area D1 marked in A1, the operation area or unauthorized area D2 of the current preset position screenshot can be calculated through projection transformation matrix mapping;

[0023] The projection transformation matrix calculation steps are:

[0024] Set the original image preset position screenshot point coordinates (x i ,y i ), call the preset position to take a screenshot of the corresponding point (x j ,y j ), the projection transformation matrix is:

[0025]

[0026] Construct matrices B and M so that B·M=0

[0027]

[0028]

[0029] Among them, n is the number of point pairs, (x i1 ,y i1 ) is the coordinate of the first point of the preset screenshot, (x j1 ,y j1 ) is the coordinate of the first point corresponding to the preset position screenshot, (x in ,y in ) is the coordinate of the nth point of the preset screenshot, (x jn ,y jn ) is the coordinate of the nth point corresponding to the preset position screenshot; the equation B·M=0 is solved using least squares to obtain the transformation matrix parameters.

[0030] Furthermore, the operation risk identification includes the operation risk identification of operators entering unauthorized areas and the operation risk identification of operation supervisors leaving the work area for a long time.

[0031] Furthermore, the risk identification process of the operator entering the unauthorized area is as follows:

[0032] Use the trained gait recognition model to identify whether the person in the current image is an operator. If not, issue an alarm directly. If the current person is an operator, determine the current operator's position and the unauthorized area for area detection. The specific formula is:

[0033]

[0034] Among them, Shuman represents the intersection of the area of ​​the operator and the unauthorized area, Sarea represents the area of ​​the operator area, and Sratio represents the ratio of the intersection of the area of ​​the operator and the unauthorized area to the area of ​​the operator area; when the operator has not entered the authorized area, Shuman is equal to 0, and the value of Sratio is equal to 0; when the operator has completely entered the authorized area, the value of Sratio is equal to 1; considering the calculation error, Sratio is set to 0.95 when the operator enters the unauthorized area for operation risk detection; when Sratio is less than 0.95, it is considered that the operator has not entered the unauthorized area; when Sratio is greater than or equal to 0.95, it is considered that the operator has entered the unauthorized area.

[0035] Furthermore, the process of identifying the operation risk of the operation supervisor being away from the work area for a long time is as follows:

[0036] Use the trained gait recognition model to identify whether the person in the current image is a supervisor or not.

[0037] The occupational supervisor identifies the operational risk according to the operator; for an operation supervisor, when the supervisor leaves the work area, the Sratio value is calculated. When Sratio is less than 0.95, the supervisor's leaving the work area is timed. When the absence is greater than 0.5 hours, an operation risk alarm is issued for the supervisor's long-term absence from the work area.

[0038] The present invention provides a machine vision-based converter station personnel operation risk identification system, comprising:

[0039] Data acquisition module: collects images of workers’ motion postures at different angles in different scenes of the converter station as image data sets;

[0040] Data processing module: pre-process the collected sample images after screening, expansion and sorting to form a training set;

[0041] Gait model training module: uses the preprocessed training set images to train the gait model used for identity identification of converter station operators;

[0042] Manual calibration module for the operating area or unauthorized area: manually calibrate the operating area or unauthorized area based on the camera preset position screenshot during the operation;

[0043] Working area or authorized area calculation module: When the converter station personnel are working, they preset the camera positions and calculate the current working area or authorized area according to the original designated area and projection matrix;

[0044] Operation risk identification module: When personnel are working, the identity of the personnel and the scope of the working area are determined by collecting images through the camera, and the trained gait model is used to identify the operation risks of the converter station personnel.

[0045] A storage medium described in the present invention stores a computer program on the storage medium, and when the computer program is executed by at least one processor, the steps of the method for identifying operational risks of converter station personnel based on machine vision as described above are implemented.

[0046] An electronic device according to the present invention comprises a memory and a processor, wherein:

[0047] A memory for storing computer programs that can be run on the processor;

[0048] The processor is used to execute the steps of the above-mentioned method for identifying the operation risks of converter station personnel based on machine vision when running the computer program.

[0049] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects: the present invention recognizes personnel identities through gait, thereby solving the problem of adding additional equipment to identify personnel identities in the converter station; by automatically calculating the working area or the authorized or unauthorized area, the problem of manually dividing the area for each operation is solved; by adopting machine vision technology, the identification errors caused by subjective judgment and human errors at the converter station site can be significantly reduced, which not only improves the accuracy of risk identification at the converter station operation, but also brings revolutionary changes to the safety management of the working site. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a flow chart of a method for identifying risks of converter station personnel operations based on machine vision;

[0051] Figure 2 Calculate a flow chart for the area of ​​operation or the area of ​​authorization. DETAILED DESCRIPTION

[0052] In order to further describe the technical features and effects of the present invention, the present invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0053] like Figure 1 As shown, the present invention proposes a method for identifying the risk of converter station personnel operations based on machine vision, which comprises the following steps:

[0054] Step S1: Collecting images of workers' motion postures at different angles in a converter station scene as an image data set.

[0055] The collected images include images of workers walking in the converter station at different times, under different lighting conditions, and under different camera angles; the unified resolution of the collected images is 1920×1080.

[0056] Step S2: Preprocessing is performed on sample images in the collected image data set after screening, expansion and arrangement to form a preprocessed training set.

[0057] Preprocessing includes the use of image enhancement techniques, including rotation, scaling, brightness adjustment, etc., to increase the diversity of the dataset. It is also necessary to remove blurred, overexposed or underexposed images to avoid adding noise.

[0058] Step S3: using the preprocessed training set images to train a gait model for identity authentication of converter station operators to obtain a trained gait model.

[0059] Firstly, StrongSORT is used as the personnel tracking model to detect the personnel operation video collected in the converter station, and the personnel ID information and coordinate information of each frame image (including the upper left corner, width, and height information of the personnel detection box) are obtained. The image containing a single person is cropped according to the personnel coordinate information.

[0060] Secondly, the single person image information obtained in the previous step is used as the input of the portrait segmentation model PP-HumanSeg to obtain the human silhouette contour information.

[0061] Finally, the gait model is trained based on the human silhouette contour information obtained in the previous step.

[0062] Step S4: In response to user operation, manually calibrate the operating area or unauthorized area according to the screenshot of the camera preset position during the operation.

[0063] "Operation area" refers to the place where various operations, maintenance and overhaul work are carried out in the converter station; in these areas, workers can carry out daily work such as installation, commissioning, operation and maintenance of power equipment. "Unauthorized area" refers to the area in the converter station where unauthorized personnel are not allowed to enter; these areas impose strict restrictions on non-relevant personnel to ensure equipment safety and avoid major misoperation.

[0064] By adjusting the camera angle to aim at the work scene of the personnel, save the current screenshot and save the camera's current angle, focal length and other parameters to the preset position number parameters. When you need to check the situation of the camera at that position, you only need to call the preset position number to quickly move the camera to this position.

[0065] The marked working area or unauthorized area is a polygonal irregular area, and the open source Labelme software is used to save the area coordinates as a json file.

[0066] Step S5: Automatically calculate the current operation area or authorized area by presetting the camera preset position.

[0067] "Authorized area" refers to the area within the flow station where people with specific permissions are allowed to enter and work. Specific permissions include monitoring, operating power equipment or performing specific power tasks.

[0068] like Figure 2 As shown, the Disk feature point extraction model is used to extract feature points of the original preset position screenshot A1 and the currently called preset position screenshot A2, and the feature point matching method is used to retain the feature points corresponding to A1 and A2, and the projection transformation matrix of A1 and A2 is calculated according to the corresponding feature points; the operating area or unauthorized area D1 marked in A1 can be used to calculate the operating area or unauthorized area D2 of the current preset position screenshot through projection transformation matrix mapping.

[0069] The steps for calculating the projection transformation matrix are:

[0070] Set the original image preset position screenshot point coordinates (x i ,y i ), call the preset position to take a screenshot of the corresponding point (x j ,y j ), the projection transformation matrix is:

[0071]

[0072] Construct matrices B and M so that B·M=0

[0073]

[0074]

[0075] Among them, n is the number of point pairs, (x i1 ,y i1 ) is the coordinate of the first point of the preset screenshot, (x j1 ,y j1 ) is the coordinate of the first point corresponding to the preset position screenshot, (x in ,y in ) is the coordinate of the nth point of the preset screenshot, (x jn ,y jn ) is the coordinate of the nth point corresponding to calling the preset position screenshot.

[0076] The equation B·M=0 is solved using least squares to obtain the transformation matrix parameters.

[0077] Step S6: The identity of the personnel and the scope of the working area are determined by collecting images through the camera. If any operating risk is detected, an alarm is automatically issued, a log is recorded and uploaded to the backend system.

[0078] Operational risks include two types: operators entering unauthorized areas and operation supervisors leaving the work area for a long time.

[0079] The specific steps for detecting the risk of operators entering unauthorized areas are: using the gait recognition model trained in step S3 to identify whether the person in the current image is an operator, if not, directly issue an alarm. If the current person is an operator, determine the current operator's position and the unauthorized area calculated in step S5 for area detection. The specific formula is:

[0080]

[0081] Among them, Shuman represents the intersection of the area of ​​the operator and the unauthorized area, Sarea represents the area of ​​the operator area, and Sratio represents the ratio of the intersection of the area of ​​the operator and the unauthorized area to the area of ​​the operator area. When the operator has not entered the authorized area, Shuman is equal to 0, and Sratio is equal to 0; when the operator has completely entered the authorized area, Shuman is equal to Sarea, and the ratio of Sratio is equal to 1; considering the calculation error, Sratio is set to 0.95 when the operator enters the unauthorized area for operation risk detection; when Sratio is less than 0.95, it is considered that the operator has not entered the unauthorized area; when Sratio is greater than or equal to 0.95, it is considered that the operator has entered the unauthorized area.

[0082] The specific steps to reduce the risk of the work supervisor leaving the work area for a long time are:

[0083] Use the gait recognition model trained in step S3 to identify whether the person in the current screen is a work supervisor. If the person is not a work supervisor, the following calculation logic will not be executed. If the person is a work supervisor, the following calculation logic will be executed: When the supervisor leaves the work area, the Sratio value is calculated. When Sratio is less than 0.95, the supervisor starts to leave the work area. The departure time threshold is set. When the threshold exceeds 0.5 hours, the supervisor will be alerted for long-term work risk.

[0084] The present invention also provides a machine vision-based converter station personnel operation risk identification system, comprising:

[0085] Data acquisition module: collects images of workers’ motion postures at different angles in different scenes of the converter station as image data sets;

[0086] Data processing module: pre-process the collected sample images after screening, expansion and sorting to form a training set;

[0087] Gait model training module: uses the preprocessed training set images to train the gait model used for identity identification of converter station operators;

[0088] Manual calibration module for the operating area or unauthorized area: manually calibrate the operating area or unauthorized area based on the camera preset position screenshot during the operation;

[0089] Working area or authorized area calculation module: When the converter station personnel are working, they preset the camera positions and calculate the current working area or authorized area according to the original designated area and projection matrix;

[0090] Operation risk identification module: When personnel are working, the identity of the personnel and the scope of the working area are determined by collecting images through the camera, and the trained gait model is used to identify the operation risks of the converter station personnel.

[0091] The present invention also provides a storage medium having a computer program stored thereon. When the computer program is executed by at least one processor, the steps of the method for identifying operational risks of converter station personnel based on machine vision as described above are implemented.

[0092] The present invention also provides an electronic device, comprising a memory and a processor, wherein: the memory is used to store a computer program that can be run on the processor; the processor is used to execute the steps of the method for identifying operational risks of converter station personnel based on machine vision as described above when running the computer program.

[0093] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0094] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0095] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0097] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A method for identifying operating risks of converter station personnel based on machine vision, characterized in that: The following steps are involved: Collect images of workers’ motion postures at different angles in different scenes at the converter station as image datasets; The sample images in the collected image data set are screened, expanded and sorted, and then preprocessed to form a preprocessed training set; Using the preprocessed training set images to train a gait model for identification of converter station operators, a trained gait model is obtained; In response to user operations, the operation area or unauthorized area is calibrated according to the screenshots of the camera preset positions during operation; By presetting the camera preset position, the current operation area or authorized area is calculated according to the original demarcated area and projection matrix; The identity of personnel and the scope of the working area are determined by collecting images through cameras, and the trained gait model is used to identify the operating risks of personnel in the converter station.

2. The method for identifying risk of converter station personnel operations based on machine vision according to claim 1, characterized in that: The uniform resolution of the images of motion postures at different angles is 1920×1080.

3. The method for identifying the risk of converter station personnel operations based on machine vision according to claim 1, characterized in that: The preprocessing includes rotation, scaling, brightness adjustment, and removal of blurred, overexposed, and underexposed images.

4. The method for identifying risks of converter station personnel operations based on machine vision according to claim 1, characterized in that: The implementation process of using the preprocessed training set images to train the gait model for identity identification of converter station operators is as follows: StrongSORT is used as a personnel tracking model to detect the personnel operation video collected by the converter station, and the personnel ID information and coordinate information of each frame image are obtained. The image containing a single person is cropped according to the personnel coordinate information; the personnel coordinate information includes the upper left corner, width, and height information of the personnel detection box; A single person image is used as the input of the portrait segmentation model PP-HumanSeg to obtain the human silhouette contour information; a gait model is trained based on the human silhouette contour information.

5. The method for identifying risk of converter station personnel operations based on machine vision according to claim 1, characterized in that: The process of calibrating the working area or unauthorized area according to the camera preset position screenshot during operation is as follows: By adjusting the camera angle to aim at the work scene of the personnel, the current screenshot is saved and the current angle and focal length of the camera are saved to the preset position number parameters; when the situation of the camera at this position is checked according to user needs, the camera can be quickly moved to this position by simply calling the preset position number; The marked operating area or unauthorized area is a polygonal irregular area, and the open source Labelme software is used to save the area coordinates as a json file.

6. The method for identifying risk of converter station personnel operations based on machine vision according to claim 1, characterized in that: The process of calculating the current operating area or authorized area based on the original demarcated area and the projection matrix is ​​as follows: Use the Disk feature point extraction model to extract feature points from the original image preset position screenshot A1 and the currently called preset position screenshot A2, use the feature point matching method to retain the feature points corresponding to A1 and A2, and calculate the projection transformation matrix of A1 and A2 based on the corresponding feature points; according to the operation area or unauthorized area D1 marked in A1, the operation area or unauthorized area D2 of the current preset position screenshot can be calculated through projection transformation matrix mapping; The projection transformation matrix calculation steps are: Set the original image preset position screenshot point coordinates (x i ,y i ), call the preset position to take a screenshot of the corresponding point (x j ,y j ), the projection transformation matrix is: Construct matrices B and M so that B·M=0 Where n is the number of point pairs, (x i1 ,y i1 ) is the coordinate of the first point of the preset screenshot, (x j1 ,y j1 ) is the coordinate of the first point corresponding to the preset position screenshot, (x in ,y in ) is the coordinate of the nth point of the preset screenshot, (x jn ,y jn ) is the coordinate of the nth point corresponding to the preset position screenshot; the equation B·M=0 is solved using least squares to obtain the transformation matrix parameters.

7. The method for identifying risks of converter station personnel operations based on machine vision according to claim 1, characterized in that: The operation risk identification includes the operation risk identification of operators entering unauthorized areas and the operation risk identification of operation supervisors leaving the work area for a long time.

8. The method for identifying risk of converter station personnel operations based on machine vision according to claim 7, characterized in that: The risk identification process of operators entering unauthorized areas is as follows: Use the trained gait recognition model to identify whether the person in the current image is an operator. If not, issue an alarm directly. If the current person is an operator, determine the current operator's position and the unauthorized area for area detection. The specific formula is: Among them, Shuman represents the intersection of the area of ​​the operator and the unauthorized area, Sarea represents the area of ​​the operator area, and Sratio represents the ratio of the intersection of the area of ​​the operator and the unauthorized area to the area of ​​the operator area; when the operator has not entered the authorized area, Shuman is equal to 0, and the value of Sratio is equal to 0; when the operator has completely entered the authorized area, the value of Sratio is equal to 1; considering the calculation error, Sratio is set to 0.95 when the operator enters the unauthorized area for operation risk detection; when Sratio is less than 0.95, it is considered that the operator has not entered the unauthorized area; when Sratio is greater than or equal to 0.95, it is considered that the operator has entered the unauthorized area.

9. The method for identifying risk of converter station personnel operations based on machine vision according to claim 7, characterized in that: The process of identifying the risk of the work supervisor leaving the work area for a long time is as follows: Use the trained gait recognition model to identify whether the person in the current picture is a work supervisor. If the person is not a work supervisor, the work risk is identified according to the work personnel. If the person is a work supervisor, when the supervisor leaves the work area, the Sratio value is calculated. When Sratio is less than 0.95, the guardian's leaving the work area is counted. If the absence is more than 0.5 hours, an operation risk alarm is issued for the work supervisor to leave the work area for a long time.

10. A machine vision-based converter station personnel operation risk identification system using the method according to any one of claims 1 to 9, characterized in that: include: Data acquisition module: collects images of workers’ motion postures at different angles in different scenes of the converter station as image data sets; Data processing module: pre-process the collected sample images after screening, expansion and sorting to form a training set; Gait model training module: uses the preprocessed training set images to train the gait model used for identity identification of converter station operators; Manual calibration module for the operating area or unauthorized area: manually calibrate the operating area or unauthorized area based on the camera preset position screenshot during the operation; Working area or authorized area calculation module: When the converter station personnel are working, they preset the camera positions and calculate the current working area or authorized area according to the original designated area and projection matrix; Operation risk identification module: When personnel are working, the identity of the personnel and the scope of the working area are determined by collecting images through the camera, and the trained gait model is used to identify the operation risks of the converter station personnel.

11. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by at least one processor, the steps of the method for identifying risks of converter station personnel operations based on machine vision are implemented as described in any one of claims 1 to 10.

12. An electronic device, characterized in that: comprising a memory and a processor, wherein: A memory for storing computer programs that can be run on the processor; A processor is used to execute the steps of the method for identifying risks of converter station personnel operations based on machine vision as described in any one of claims 1 to 10 when running the computer program.

Images