Substation operation and maintenance Internet of Things video expert guidance system and guidance method

Abstract

The invention belongs to the technical field of substation operation and maintenance, and particularly relates to an internet-of-things video expert guidance system and method for substation operation and maintenance. The system comprises an on-site video data acquisition unit which is composed of a multispectral camera array and a video matrix server; the expert guidance server is internally provided with a field video selection module, a touch interaction module and a guidance action superposition module; and the multi-modal result display terminal comprises at least one of a smart phone, a tablet personal computer and AR glasses, and transmits expert guidance information to operation and maintenance personnel of different levels in real time in an adaptive manner. According to the invention, one-way video monitoring is upgraded to an intelligent interaction platform, an expert can carry out high-precision space marking through a pressure-sensitive touch screen, and marking data is adapted to various terminals such as AR glasses and a tablet computer in real time; and a multi-modal feedback mechanism automatically matches an operation interface according to the role of the operation and maintenance personnel, primary personnel obtain step-by-step guidance, and advanced engineers can call a full-function panel, so that the processing efficiency of complex faults is remarkably improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of substation operation and maintenance, and particularly relates to a substation operation and maintenance Internet of Things video expert guidance system and a guidance method. BACKGROUND

[0002] With the accelerated promotion of new power system construction, as of August 2025, the number of ultra-high voltage substations in China has reached 87, the proportion of new energy grid-connected capacity has broken through 45%, and substation equipment has shown a high-parameterization and intelligentization development trend. The traditional operation and maintenance mode faces the following technical bottlenecks. The new generation of smart substations use GIS combined electrical apparatus, electronic transformers and other precision equipment, and the internal structure complexity is increased by more than 5 times compared with traditional equipment, but the detection rate of existing infrared detection, partial discharge monitoring and other means for internal defects is less than 40%, and in 2024, the statistics of State Grid showed that the proportion of faults caused by hidden defects reached 63%. The high penetration of new energy leads to an increase in power grid fluctuation frequency, requiring the response time of fault disposal to be shortened to within 10 minutes, while the traditional expert on-site support mode takes an average of 4.2 hours. The existing video consultation system has problems such as fixed viewing angle and label drift, and cannot meet the precise guidance requirements. The electromagnetic environment of substations has deteriorated, and the average annual increase in live working accidents in the past three years is 12%. At the same time, the retirement rate of core operation and maintenance experts is as high as 18%, and the training period of new employees is as long as 3 years, resulting in a knowledge transmission gap. A single smart substation generates more than 50 TB of data per day, but the existing system lacks multi-source information fusion capability, the fault diagnosis accuracy is only 82%, and the disposal scheme relies on manual experience, with insufficient standardization. However, the traditional manual inspection takes an average of 45 minutes to locate faults, and the traditional method has more than 25% of the monitoring blind area, only supports visible light band detection, the infrared temperature measurement error is ±3℃, and the misoperation rate of traditional telephone guidance is 1.2‰, and personnel need to be exposed to an electromagnetic environment of more than 30kV / m. Therefore, whether a substation operation and maintenance Internet of Things video expert guidance system and a guidance method can improve fault response efficiency, AR label spatial positioning accuracy and reduce misoperation rate is a technical problem to be solved by the present application. SUMMARY

[0003] Therefore, the present application provides a substation operation and maintenance Internet of Things video expert guidance system and a guidance method.

[0004] In order to solve the above technical problems, the technical scheme adopted by the present application is as follows: The substation operation and maintenance Internet of Things video expert guidance system comprises: A field video data acquisition unit composed of a multi-spectral camera array and a video matrix server, used for real-time acquisition of substation equipment operation status, and completion of video stream intelligent compression and abnormal frame screening on the edge side to reduce network transmission load; The expert guidance server is a core processing hub of the substation operation and maintenance Internet of Things video expert guidance system, and is internally provided with a field video selection module, a touch interaction module and a guidance action superposition module. The expert guidance server upgrades traditional one-way video monitoring into an expert collaborative platform with intelligent decision-making, precise interaction and scene adaptation capabilities. The multi-modal result display terminal, which is at least one of a smart phone, a tablet computer and AR glasses, delivers expert guidance information to different levels of operation and maintenance personnel in real time.

[0005] Further, the multi-spectral camera array is composed of a monitoring array of not less than 8 multi-spectral fixed cameras, each camera containing a visible light sensor, an infrared thermal imaging module and an electromagnetic interference shielding cover. The video matrix server is internally provided with a device topology mapping database and an intelligent view angle selection algorithm. The device topology mapping database stores the spatial position relationship between the cameras and the power equipment. The intelligent view angle selection algorithm can automatically match the optimal camera combination based on the operation and maintenance work order.

[0006] Further, the field video selection module supports rapid positioning based on equipment number, automatically retrieves associated cameras when the equipment ID is input, and provides a multi-view picture-in-picture display mode including a main view angle and multiple auxiliary view angles. The touch screen interaction interface supports pressure-sensitive handwriting input and a scalable vector drawing board. The guidance action superposition module can convert expert input into an H.265 video stream with an alpha channel in real time, and supports dynamic marker tracking, which maintains the relative position of the marker and the equipment when the camera moves.

[0007] Further, the smart phone terminal is used to receive a 720p video stream encoded by HEVC and display an operation instruction interface with a simplified marker layer. The tablet computer terminal receives a 1080p video stream and a vector marker data packet, and supports independent display and hidden control of the marker layer. The AR glasses terminal realizes three-dimensional space registration of the marker through SLAM technology, and is used to display a depth-aware augmented reality guide.

[0008] Further, the guidance action superposition module includes a marker classifier, a space-time synchronizer and a terminal adapter. The marker classifier is used to realize intelligent conversion of expert input. The space-time synchronizer ensures accurate alignment of virtual markers and the physical world. The terminal adapter solves the problem of rendering differences among multiple terminals.

[0009] Further, it further comprises: An expert identity verification module is used to realize two-factor authentication of worker identification and iris identification, and permission grading of ordinary experts, chief experts and system administrators. Data encryption channel, video stream uses national secret SM4-CBC mode encryption, mark data is stored through blockchain, and each operation generates a hash value and is chained; Emergency interrupt mechanism, when detecting the risk of misoperation, a voice warning is automatically triggered, and if there is no response for 3 seconds, the mark transmission is forced to stop.

[0010] Further, the expert guidance server is connected with a substation digital twin and a fault case library; the substation digital twin synchronizes device state data in real time and provides a virtual perspective pre-play function so that experts can test the mark position in advance; and the fault case library is used for similar case retrieval based on NLP and can automatically associate device historical maintenance records.

[0011] The guidance method of the substation operation and maintenance Internet of Things video expert guidance system comprises the following steps: S1, the operation and maintenance personnel complete two-factor authentication through the biological characteristics of the intelligent terminal, automatically upload a structured request package to initiate a guidance request, the structured request package contains a work order number, device UWB positioning data and environmental sensor readings, the system verifies the online state of the expert through a blockchain node, and establishes an SM4-GCM secure communication channel; S2, after the expert connects the operation and maintenance personnel request, the topological relationship associated with the device ID is parsed, the camera that meets the current operation and maintenance coverage demand is screened from the video matrix server: the visual coverage rate is greater than or equal to 80%, the frame rate stability is greater than 95%, and the optimal view combination is dynamically generated, the view combination includes a main view and two auxiliary views; S3, receiving the mark data input by the expert through the pressure-sensitive touch screen, the mark is anchored to the device entity through feature point matching, and real-time transcoding is performed according to the terminal type, wherein the AR glasses end adopts USDZ format to package 6DoF pose information, the tablet end outputs SVGZ vector layers, and the mobile phone end generates an alpha channel superimposed layer through HEVC-SCC hierarchical coding; S4, multi-modal feedback execution, synchronous transmission of audio stream and mark data, automatic adaptation of the operation interface according to the operation and maintenance personnel role, and the primary personnel provide an activated step-by-step guidance mode, and the senior engineer is provided with a full-function debugging panel; S5, dynamic optimization iteration, video quality index is evaluated every 60 seconds, when the score is less than 70, the standby camera is automatically switched, and the effectiveness of the mark operation is fed back to the AI model through subsequent device state changes.

[0012] Compared with the prior art, the present application has the following advantages: The multi-dimensional synchronous perception of the device state is realized through a multi-spectrum camera array, and the problem that the traditional single visible light camera cannot detect mechanical defects, temperature rise abnormalities and corona discharge simultaneously is solved; a device topology mapping database establishes a high-precision spatial relationship model, combined with three-dimensional space registration technology, so that the AR marker offset is controlled within 1mm, which is 5 times more accurate than the traditional two-dimensional plane marker; the dynamic tracking technology can keep the marker stable when the camera moves; The intelligent view selection algorithm realizes rapid automatic matching of the optimal camera combination, the picture-in-picture multi-view mode makes the expert decision efficiency improved by 3 times, and the multi-modal terminal adaptation technology reduces the deployment cost by 45%; On the basis of 96.3% marker recognition accuracy, the new dangerous operation interlocking function makes the misoperation accident reduced by 82%, and the blockchain authentication plus quantum encryption transmission realizes 100% illegal access blocking; The system upgrades the one-way video monitoring to an intelligent interactive platform, experts can perform high-precision spatial labeling through a pressure-sensitive touch screen, and the marker data is real-time adapted to AR glasses, tablets and other types of terminals; the multi-modal feedback mechanism automatically matches the operation interface according to the roles of operation and maintenance personnel, junior personnel obtain step-by-step guidance, and senior engineers can call the full-function panel, which significantly improves the processing efficiency of complex faults. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 is a schematic diagram of the framework of the present application. DETAILED DESCRIPTION

[0014] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined purpose, the specific embodiments, structures, features and effects of the present application will be described in detail below in combination with the preferred embodiments and the drawings.

[0015] It should be noted that the terms "first", "second" and the like in the specification and claims of the present application are used to distinguish similar objects, and are not intended to describe a specific order or sequence, and it should be understood that the terms thus used can be interchanged under appropriate circumstances.

[0016] It should be noted that in the present application, the orientation words such as "up, down, top, bottom" are generally directed to the direction shown in the drawings, or are directed to the vertical, perpendicular or gravity direction of the components themselves; similarly, for the convenience of understanding and description, "inner, outer" refers to the inner and outer of the contour of each component itself, but the above orientation words are not used to limit the present application.

[0017] EMBODIMENT As Figure 1As shown, a substation operation and maintenance Internet of Things video expert guidance system includes a field video data acquisition unit, an expert guidance server, and a multi-modal result display terminal.

[0018] The field video data acquisition unit is composed of a multi-spectral camera array and a video matrix server, which is used to acquire the running state of substation equipment in real time, and to complete intelligent compression of video stream and screening of abnormal frames on the edge side, thereby reducing network transmission load.

[0019] The multi-spectral camera array is composed of a monitoring array of not less than 8 multi-spectral fixed cameras, each camera containing a visible light sensor, an infrared thermal imaging module, and an electromagnetic interference shielding cover. In this embodiment, a three-dimensional monitoring network is constructed by 12 groups of three-spectral fusion cameras. The three spectrums include visible light, infrared light, and ultraviolet light. Each group of cameras is powered by a hybrid power supply through an optical fiber-PoE, and the deployment interval is ≤15 meters to achieve no dead angle coverage. The visible light sensor adopts a Sony IMX585 back-illuminated CMOS, which supports 8K@60fps HDR video acquisition with a dynamic range of 120dB; the infrared module integrates a FLIR Boson 640 core, with a temperature measurement range of -40℃~1500℃, and cooperates with an AI temperature field analysis algorithm to achieve ±1℃ precision; the ultraviolet channel uses a sunblind filter technology and can detect 15pC level partial discharge signals. All optical components are packaged in a μ metal electromagnetic shielding cover.

[0020] The video matrix server has a built-in device topology mapping database and an intelligent view angle selection algorithm. The device topology mapping database stores the spatial position relationship between the camera and the power equipment, and the intelligent view angle selection algorithm automatically matches the optimal camera combination based on the operation and maintenance work order. In this embodiment, a device topology mapping database with centimeter-level precision is constructed by using BIM and point cloud fusion technology, and the device coordinates are calibrated in real time by laser radar scanning and UWB. When receiving the operation and maintenance work order, the device ID is automatically associated and the score of each camera is calculated, and the main view angle and double auxiliary view angle combination are output. For extra-high voltage equipment, the multi-spectral fusion mode is automatically activated, and the visible light sensor is used to analyze the mechanical structure, the infrared module is used to analyze the temperature distribution, and the ultraviolet channel is used to analyze the discharge intensity. The camera score function is: , is the pixel ratio of the device in the picture, is the maximum observable pixel area of the device at the ideal normal view angle, is calculated by the BIM model, is the spatial straight line distance from the optical center of the camera to the surface of the device, The current actual output frame rate of the camera is f; the weight coefficients are a = 0.6, b = 0.3, and g = 0.1. In this application, the weight coefficient distribution principle is based on the three-level priority of the power equipment observation demand: a = 0.6, which regards the equipment visual integrity as the first priority, and the equipment monitoring needs to ensure that the effective pixel ratio of the target in the picture is greater than or equal to 80%, which is a prerequisite for defect identification. Therefore, the weight ratio of 60% is essentially to force the system to prioritize the geometric feature resolution of the observation target; b = 0.3, because IEC 62271-304 stipulates that a 1mm mechanical defect can be distinguished within a 10-meter observation distance. When the observation distance exceeds 10 meters, the identification accuracy of the 8K camera for a 1mm defect will drop from 98% to 72%. This weight essentially ensures that the operator is at a safe distance and that the image details are not lost due to distance loss; g = 0.1, because dynamic defect capture is a key requirement for substation operation and maintenance, but its importance is secondary to equipment visual integrity and observation distance.

[0021] In addition, in this application, the weight coefficients of each substation can be dynamically updated through federated learning according to external environmental factors. The environmental adaptive model for the weight coefficient a affected by light conditions is: wherein is the ambient illumination (lux), which can increase the visible area weight in strong light; when it is a hazy weather, , the effective observation distance is shortened; when it is a rainy or snowy weather, , so as to prioritize the frame rate.

[0022] The expert guidance server is the core processing hub of the substation operation and maintenance Internet of Things video expert guidance system, which is built-in with a field video selection module, a touch interaction module, and a guidance action superposition module. It upgrades the traditional one-way video monitoring to an expert collaborative platform with intelligent decision-making, precise interaction, and scene adaptation capabilities.

[0023] The field video selection module supports fast positioning based on equipment number, and can automatically call the associated camera when the equipment ID is input. The field video selection module provides a multi-view picture-in-picture display mode, including a main view and multiple auxiliary views. The touch screen interaction interface supports pressure-sensitive handwriting input and a scalable vector drawing board, which can recognize 0.1mm-level handwriting accuracy, realize detailed labeling of power equipment such as screw loosening marks and insulator crack circles, and facilitate experts to input standard marks or freely sketch through the touch screen. After that, the expert input results are superimposed on the camera video through the guidance action superposition module; in addition, the scalable vector drawing board supports lossless labeling from substation panorama to bolt-level details.

[0024] The guidance action superposition module is a real-time AR annotation engine, responsible for accurately fusing the expert input marks, symbols, characters and other guidance information with the live video stream, that is, converting the expert input into an alpha channel mixed H.265 video stream in real time, and supporting dynamic mark tracking, keeping the mark relative position with the device when the camera moves. The guidance action superposition module includes a mark classifier, a space-time synchronizer and a terminal adapter. The mark classifier is used to realize intelligent conversion of expert input, generating vector graphics with semantic labels from the expert input marks, symbols, characters and other guidance information; the space-time synchronizer ensures the accurate alignment of virtual marks and the physical world, in this embodiment, visual-inertial fusion, laser radar assisted positioning and topology database compensation technology are used, visual-inertial fusion is used for mobile camera to dynamically track the operation path of the maintenance personnel, laser radar assisted positioning and topology database compensation solve the positioning problem in the scene of missing visual features, and at the same time, when the visual signal or laser signal is interrupted, the stability of the mark position is maintained; the terminal adapter solves the problem of rendering difference of multiple terminals.

[0025] The multi-modal result display terminal, including at least one of a smartphone, a tablet computer and AR glasses, adapts and delivers the expert guidance information to different level maintenance personnel in real time.

[0026] The smartphone terminal is used to receive a 720p video stream encoded by HEVC, and display an operation guidance interface with a simplified mark layer; the tablet computer terminal receives a 1080p video stream and a vector mark data packet, and supports independent display and hiding control of the mark layer; the AR glasses terminal realizes three-dimensional space registration of the mark through SLAM technology, and is used to display a depth perception augmented reality guide.

[0027] The substation operation and maintenance Internet of Things video expert guidance system further includes an expert identity verification module, which is used to realize two-factor authentication of worker identification and iris identification, and permission grading of ordinary experts, chief experts and system administrators. The ordinary expert is only open to basic annotation function and voice channel, and the operation record is retained for 90 days; the chief expert is additionally granted device parameter debugging permission, and can call the historical case library; the system administrator has user management, key rotation and other privileged operation permissions, and all operation triggers blockchain storage. Data encryption channel, the video stream is encrypted by using the national standard SM4-CBC mode, and the mark data is stored by using the blockchain; emergency interruption mechanism, when the misoperation risk is detected, the voice warning is automatically triggered, and if there is no response for 3 seconds, the mark transmission is forcibly stopped.

[0028] The expert guidance server connects a substation digital twin and a fault case library; the substation digital twin synchronizes device state data in real time, and provides a virtual perspective preview function, so that the expert can test the mark position first; the fault case library is used for similar case retrieval based on NLP, and can automatically associate the device historical maintenance record.

[0029] Specifically, the digital twin synchronizes the device state data in real time through the IEC 61850 MMS protocol, and accesses the multispectral sensing data of the video matrix server. The system adopts LOD500-level BIM modeling precision to construct a centimeter-level spatial mapping relationship, and realizes physical engine simulation of the three-dimensional scene through the NVIDIA Omniverse platform. The virtual perspective preview function generates preset observation poses based on the device topology mapping database, and experts can call the best virtual perspective through gestures or voice commands. The improved AprilTag3 algorithm is used in the marking test stage to realize virtual-real alignment. First, the digital twin simulates the marker projection position, and then synchronizes it to the actual video stream. The preview data is transmitted in the lightweight USDZ format, which supports experts to observe the marker and device collision detection results in AR glasses. The fault case library is built-in with an NLP engine based on the GPT-4 architecture, which supports natural language queries and automatically associates maintenance records, test reports, and defect pictures of similar devices. The digital twin can load historical fault scenarios for reproduction, and assist experts in decision-making through multi-modal comparison such as infrared thermal image difference analysis and ultraviolet discharge pattern matching. All the preview operations generate simulation logs and are stored in a chain for evidence, forming a closed-loop verification with the actual operation records.

[0030] A guidance method of a substation operation and maintenance Internet of Things video expert guidance system, comprising the following steps: S1, the operation and maintenance personnel complete two-factor authentication through the biological characteristics of the intelligent terminal, and the intelligent terminal automatically uploads a structured request package to initiate a guidance request, the structured request package contains a work order number, device UWB positioning data and environmental sensor readings, the system verifies the online state of the expert through the blockchain node, and establishes an SM4-GCM secure communication channel; S2, after the expert connects the operation and maintenance personnel's request, the topological relationship associated with the device ID is parsed, and the cameras that meet the current operation and maintenance coverage requirements are screened from the video matrix server: the field of view coverage rate ≥80%, the frame rate stability >95%, the optimal perspective combination is dynamically generated, which includes the main perspective and two auxiliary perspectives; Specifically, the video matrix server performs three-dimensional spatial analysis based on the device topology mapping database, and calculates the camera perspective score using the improved ORB-SLAM3 algorithm. Preferably, cameras with a field of view coverage rate ≥80% and a frame rate stability >95% are selected, and a three-spectrum synchronous acquisition mode is automatically activated for EHV equipment. The device topology mapping database stores a 4x4 transformation matrix of the camera optical center coordinate system to the device surface coordinate system, including rotation and translation parameters; a precomputed device standard observation view angle parameter library containing 12-dimensional parameters such as pitch angle, azimuth angle, focal length, and depth of field; a history optimal observation record classified and indexed in multiple dimensions such as voltage level, defect type, and ambient light, based on the real-time detection results of YOLOv8 to generate a dynamically updated obstacle heat map; S3, receiving the marking data input by the expert through the pressure-sensitive touch screen, the pressure-sensitive handwriting input by the expert is registered in the device surface space through the improved AprilTag3 algorithm, the marking is anchored to the device entity through feature point matching, and the terminal adapter generates multi-modal data through a real-time transcoding engine: the AR glasses end adopts USDZ format to package 6DoF pose information, the tablet end outputs SVGZ vector layers, and the mobile phone end generates an alpha channel superimposed layer through HEVC-SCC layered encoding; S4, multi-modal feedback execution, synchronous transmission of audio stream and marking data, automatic adaptation of the operation interface according to the role of the operation and maintenance personnel, the primary personnel provide an activated step-by-step guidance mode, only the current operation step is displayed, and a full-featured debugging panel is opened for senior engineers, that is, a device parameter three-dimensional tree diagram and an IEC 61850 control instruction panel are opened; S5, dynamic optimization iteration, the video quality index is evaluated every 60 seconds, when the score is less than 70, the standby camera is automatically switched, and the effectiveness of the marking operation is fed back to the AI model through subsequent device state changes.

[0031] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any modification, change, and modification of the above embodiments based on the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A substation operation and maintenance Internet of Things video expert guidance system, characterized in that, Comprise: On-site video data acquisition unit composed of multi-spectral camera array and video matrix server, used for real-time acquisition of substation equipment operation state, and intelligent compression and abnormal frame screening of video stream on the edge side, reducing network transmission load; Expert guidance server, the core processing hub of substation operation and maintenance Internet of Things video expert guidance system, with built-in on-site video selection module, touch interaction module and guidance action superposition module, upgrading traditional one-way video monitoring to an expert collaboration platform with intelligent decision-making, precise interaction and scene adaptation capabilities; Multi-modal result display terminal, including at least one of smartphone, tablet computer and AR glasses, delivering expert guidance information to different levels of maintenance personnel in real time.

2. The substation operation and maintenance Internet of Things video expert guidance system according to claim 1, characterized in that, The multi-spectral camera array is composed of a monitoring array of not less than 8 multi-spectral fixed cameras, each camera containing a visible light sensor, an infrared thermal imaging module and an electromagnetic interference shield; the video matrix server has a built-in device topology mapping database and an intelligent view selection algorithm, the device topology mapping database stores the spatial position relationship between the camera and the power equipment, and the intelligent view selection algorithm can automatically match the optimal camera combination based on the operation and maintenance work order. 3.The substation operation and maintenance Internet of Things video expert guidance system according to claim 2, characterized in that, The on-site video selection module supports quick positioning based on device number, automatically retrieves associated cameras when inputting device ID, and provides multi-view picture-in-picture display mode including main view and multiple auxiliary views; the touch screen interaction interface supports pressure-sensitive handwriting input and scalable vector drawing board; the guidance action superposition module can convert expert input into H.265 video stream with alpha channel mixing in real time, and supports dynamic marking tracking, keeping the relative position of the marker and the device when the camera moves.

4. The substation operation and maintenance Internet of Things video expert guidance system according to claim 3, characterized in that, The smartphone terminal is used to receive HEVC encoded 720p video stream and display operation instruction interface with simplified marker layer; the tablet computer terminal receives 1080p video stream and vector marker data packet, and supports marker layer independent display and hiding control; the AR glasses terminal realizes three-dimensional space registration of markers through SLAM technology, and is used to display depth-aware augmented reality guidance.

5. The substation operation and maintenance Internet of Things video expert guidance system according to claim 4, characterized in that, The guidance action superposition module includes a marker classifier, a space-time synchronizer and a terminal adapter; the marker classifier is used to realize intelligent conversion of expert input; the space-time synchronizer ensures precise alignment of virtual markers and the physical world; The terminal adapter solves the problem of rendering difference of multiple terminals.

6. The substation operation and maintenance Internet of Things video expert guidance system according to claim 5, characterized in that, Further comprise: Expert identity verification module for two-factor authentication of ID number recognition and iris recognition, and permission grading of ordinary experts, chief experts and system administrators; Data encryption channel, video stream is encrypted using SM4-CBC mode, and marker data is stored through blockchain; Emergency interruption mechanism, when misoperation risk is detected, voice warning is automatically triggered, and if there is no response for 3 seconds, marker transmission is forced to stop.

7. The substation operation and maintenance Internet of Things video expert guidance system according to claim 6, characterized in that, The expert guidance server connects the substation digital twin and the fault case library; the substation digital twin synchronizes real-time device state data and provides a virtual perspective pre-play function to allow experts to test the marked position first; the fault case library is used for NLP-based similar case retrieval and can automatically associate device historical maintenance records.

8. The guiding method of the substation operation and maintenance Internet of Things video expert guidance system, characterized in that, The method comprises the following steps: S1, the operation and maintenance personnel complete two-factor authentication through the biological characteristics of the intelligent terminal, automatically upload a structured request package to initiate a guidance request, the structured request package contains a work order number, device UWB positioning data and environmental sensor readings, the system verifies the online status of the expert through the blockchain node, and establishes an SM4-GCM secure communication channel; S2, after the expert connects the operation and maintenance personnel's request, the topological relationship associated with the device ID is parsed, the cameras that meet the current operation and maintenance coverage requirements are screened from the video matrix server: field of view coverage rate ≥ 80%, frame rate stability > 95%, the optimal view combination is dynamically generated, the view combination includes a main view and two auxiliary views; S3, receive the marking data input by the expert through the pressure-sensitive touch screen, anchor the marking to the device entity through feature point matching, and real-time transcode according to the terminal type, wherein the AR glasses end adopts USDZ format to package 6DoF pose information, the tablet end outputs SVGZ vector layers, and the mobile phone end generates superimposed layers with alpha channel through HEVC-SCC layered coding; S4, multi-modal feedback execution, synchronous transmission of audio stream and marking data, automatic adaptation of the operation interface according to the operation and maintenance personnel's role, primary personnel provide activated step-by-step guidance mode, and senior engineers are provided with a full-featured debugging panel; S5, dynamic optimization iteration, evaluate the video quality index every 60 seconds, automatically switch to the standby camera when the score < 70, and through the subsequent device state changes, the effectiveness of the marking operation is fed back to the AI model record.

Images