Ground main ventilator room video inspection method and system

Abstract

The invention provides a ground main ventilator room video inspection method and system, and the method comprises the steps: collecting images according to a dynamic frame rate through intrinsic safety type high-definition camera units disposed in a fan impeller region, an instrument region and a pipeline region, and carrying out the dust noise reduction processing and vibration anti-shake correction of the images, and obtaining a preprocessed image; inputting the preprocessed image into a pre-trained ventilation part identification model to obtain the positions of an impeller, an instrument and a pipeline in the preprocessed image, and determining an impeller image, an instrument image and a pipeline image; performing fault analysis on the impeller image, the instrument image and the pipeline image to obtain a fault analysis result; and then a control instruction is generated according to the fault level and is sent to a fan control system through an industrial communication protocol so as to execute linkage control operation. According to the technical scheme provided by the invention, full-coverage, high-precision and real-time monitoring and intelligent linkage disposal of visual and state faults of the ventilation equipment are realized.

Description

A video inspection method and system for ground main ventilation fan room Technical Field

[0001] This application relates to the field of mine ventilation technology, and in particular to a video inspection method and system for the main ventilation room on the ground. Background Technology

[0002] The main ventilation room on the surface is the "heart" of the ventilation system in a coal mine. The continuous and stable operation of its fans, motors, lubrication systems, and auxiliary pipelines and instruments directly affects the safety of underground workers and the normal operation of production activities. Therefore, real-time, comprehensive, and reliable inspections of the equipment operating status in the room are crucial.

[0003] Currently, the industry mainly relies on the following three methods for inspecting ground-based main ventilation fan rooms, each with significant technical limitations: 1. Manual timed inspection mode: Inspection personnel follow fixed routes and cycles (usually 1-3 times per shift), carrying inspection instruments and logbooks, and use sensory experience such as "seeing, hearing, touching, and smelling" to check fan vibration, abnormal noise, bearing temperature, instrument readings, oil level, and leaks. This mode has inherent defects: First, it has poor timeliness, with a time window of several hours between the occurrence and discovery of a fault, making it unable to cope with emergencies; second, it is highly subjective, with inspection quality heavily dependent on personnel experience and sense of responsibility, easily leading to missed inspections and misjudgments; third, it poses safety risks, requiring personnel to approach high-speed rotating and high-voltage electrical equipment, and cannot achieve full-time coverage. 2. Sensor discrete monitoring system: Physical sensors such as temperature, vibration, pressure, and flow are installed at key points of the equipment, converting analog signals into digital signals and uploading them to the monitoring center. This system achieves continuous monitoring of some key parameters, but its limitations are as follows: First, there are blind spots in coverage, rendering it completely ineffective for visual faults such as impeller surface cracks, coating peeling, loose bolts, pipeline corrosion and oil leakage, and control cabinet indicator light status. Second, the system is fragmented; sensor signals and video monitoring systems are not interconnected, data cannot be fused, and comprehensive fault diagnosis is difficult. Third, there are cost and maintenance pressures; comprehensive sensor deployment is costly, and the sensors themselves also have issues with faults, calibration, and cable maintenance. 3. Traditional video monitoring systems: Ordinary or analog cameras are installed in the computer room for 24-hour uninterrupted recording and storage. This method only achieves "visualization" of the site, not "intelligentization." The problems are as follows: First, it's "monitoring without control." Video streams require manual monitoring on a backend screen or post-event review, resulting in a significant lag between fault occurrence and response, and is highly susceptible to oversight due to visual fatigue. Second, it lacks environmental adaptability. Dust-filled machine rooms in coal mines, uneven lighting, and strong equipment vibrations lead to poor image quality and render conventional image processing algorithms ineffective. Third, it lacks intelligent analysis, failing to automatically identify equipment status or link the identification results with the control system; it's merely a passive recording tool. In summary, the existing technology system presents an "information silo" state: manual inspection, sensor networks, and video surveillance operate independently, failing to create a synergistic effect. Its core pain points can be summarized as: delayed fault detection, missed visual defects, low recognition rates in harsh environments, and a disconnect between monitoring and control. This prevents the real-time, automatic, and accurate perception and handling of safety hazards in ventilation systems, making it difficult to meet the urgent needs of modern intelligent mines for "unmanned operation, intelligent inspection, and proactive safety." Therefore, there is an urgent need for a new inspection solution that can integrate multi-source information, adapt to harsh industrial environments, and achieve a closed loop from intelligent sensing to automatic control. Summary of the Invention

[0004] This application provides a video inspection method and system for ground main ventilation room, which at least solves the technical problem that safety hazards in the ventilation system cannot be detected and dealt with in real time, automatically and accurately.

[0005] The first aspect of this application proposes a video inspection method for a ground-based main ventilation fan room. The method includes: acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area; performing dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; inputting the pre-processed images into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images; determining the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; performing fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; determining the fault type and classifying the fault level according to the analysis results; and then generating control commands according to the fault level and sending them to the fan control system via an industrial communication protocol to execute linkage control operations.

[0006] Preferably, the deployment method of the intrinsically safe high-definition camera group includes: deploying a camera with a telephoto lens dynamic module in the fan impeller area; deploying a camera with a fixed-focus lens and an infrared fill light module in the instrument area; and deploying a camera with a wide-angle lens in the pipeline area.

[0007] Furthermore, the step of performing dust noise reduction and vibration stabilization correction on the image to obtain a preprocessed image includes: using an improved BM3D algorithm to perform multi-frame collaborative filtering on the acquired image to remove dust particle noise; calculating the motion vector between adjacent frames of the image after removing dust particle noise based on SIFT feature point matching and performing reverse mapping to achieve vibration stabilization correction.

[0008] Preferably, the ventilation component recognition model is trained using the YOLOv8 algorithm with an added attention mechanism module.

[0009] Furthermore, the fault analysis of the impeller image, instrument image, and pipeline image to obtain fault analysis results includes: calculating the blade motion trajectory based on the impeller image using optical flow field and extracting vibration frequency features through Fourier transform; determining an impeller abnormality when the difference between the vibration frequency and a preset vibration frequency threshold is greater than 5Hz; identifying the bearing temperature corresponding to the instrument image using Hough line detection and OCR character recognition methods; determining bearing wear of the fan when the bearing temperature is greater than a preset bearing temperature threshold; and identifying the oil leakage area using an improved background difference method with a set adaptive threshold based on the pipeline image; determining an oil leakage when the ratio of the oil leakage area to the background area of ​​the casing is greater than the set adaptive threshold.

[0010] Furthermore, the step of determining the fault type and classifying the fault level based on the analysis results includes: when an impeller abnormality occurs, determining the fault level of the fan as Level I emergency fault; when bearing wear occurs, determining the fault level of the fan as Level II important fault; and when oil leakage occurs, determining the fault level of the fan as Level III general fault.

[0011] Furthermore, the generation of control instructions based on the fault level includes: when the fault level is Level I emergency fault, generating instructions to immediately shut down and start the standby fan, and sending alarm information to maintenance personnel; when the fault level is Level II important fault, generating a load reduction operation instruction; and when the fault level is Level III general fault, generating an instruction to send a warning information.

[0012] The second aspect of this application proposes a video inspection system for a ground-based main ventilation fan room, comprising: a data acquisition module, used to acquire images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area, and to perform dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; a recognition module, used to input the pre-processed images into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images, and to determine the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; a fault analysis module, used to perform fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; and a control module, used to determine the fault type and classify the fault level according to the analysis results, and then generate control commands according to the fault level and send them to the fan control system through an industrial communication protocol to execute linkage control operations.

[0013] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the method described in the first aspect embodiment.

[0014] A fourth aspect of this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described in the first aspect.

[0015] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects: This application proposes a video inspection method and system for a ground main ventilation fan room. The method includes: acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area; performing dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; inputting the pre-processed images into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images; determining the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; performing fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; determining the fault type and classifying the fault level according to the analysis results; and then generating control commands according to the fault level and sending them to the fan control system via an industrial communication protocol to execute linkage control operations. The technical solution proposed in this application achieves full coverage, high precision, real-time monitoring, and intelligent linkage handling of visual and status faults of ventilation equipment, significantly improving the safety and reliability of mine ventilation systems.

[0016] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0017] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 is a flowchart of a video inspection method for a ground main ventilation fan room according to an embodiment of this application; Figure 2 is a structural diagram of a video inspection system for a ground main ventilation fan room according to an embodiment of this application. Detailed Implementation

[0018] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0019] This application proposes a video inspection method and system for a ground-based main ventilation fan room. The method includes: acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area; performing dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; inputting the pre-processed images into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images; determining the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; performing fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; determining the fault type and classifying the fault level according to the analysis results; and then generating control commands according to the fault level and sending them to the fan control system via an industrial communication protocol to execute linkage control operations. The technical solution proposed in this application achieves full coverage, high precision, real-time monitoring, and intelligent linkage handling of visual and status faults of ventilation equipment, significantly improving the safety and reliability of mine ventilation systems.

[0020] The following description, with reference to the accompanying drawings, illustrates a video inspection method and system for a ground-based main ventilation fan room, according to an embodiment of this application.

[0021] Figure 1 is a flowchart of a video inspection method for a ground-based main ventilation fan room according to an embodiment of this application. As shown in Figure 1, the method includes: Step 1: acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area, and performing dust noise reduction and vibration anti-shake correction on the images to obtain a pre-processed image; In this embodiment, the deployment of the intrinsically safe high-definition camera group includes: deploying a camera with a telephoto lens dynamic module in the fan impeller area; deploying a camera with a fixed-focus lens and an infrared fill light module in the instrument area; and deploying a camera with a wide-angle lens in the pipeline area.

[0022] It should be noted that 6-8 intrinsically safe high-definition cameras will be deployed, with the following configurations: ① The fan impeller area uses a telephoto lens dynamic module; ② The instrument area uses a fixed-focus lens + infrared supplementary lighting module; ③ The pipeline area uses a fisheye lens (180° field of view). All equipment has an IP66 protection rating and supports an operating temperature range of -30℃ to 70℃.

[0023] It adopts an industrial-grade fiber optic ring network with a bandwidth of 1000Mbps, a transmission latency of ≤50ms, and has redundant backup function.

[0024] Deploy an edge computing server (configured with an NVIDIA A100 GPU) with a built-in video analytics engine and fault decision system.

[0025] Furthermore, the step of performing dust noise reduction and vibration stabilization correction on the image to obtain a preprocessed image includes: using an improved BM3D algorithm to perform multi-frame collaborative filtering on the acquired image to remove dust particle noise; calculating the motion vector between adjacent frames of the image after removing dust particle noise based on SIFT feature point matching and performing reverse mapping to achieve vibration stabilization correction.

[0026] It should be noted that in step 1-1, the camera acquires images at a dynamic frame rate and transmits them at an efficiency of 3 frames per second under normal operating conditions.

[0027] Steps 1-2: Dust noise reduction is performed using an improved BM3D algorithm. Dust particle noise is removed through multi-frame collaborative filtering, and the peak signal-to-noise ratio (PSNR) is increased to over 35dB.

[0028] Steps 1-3: Vibration stabilization correction: Based on SIFT feature point matching, the motion vector between adjacent frames is calculated, and the image is stabilized through reverse mapping, with a stabilization accuracy of 0.5 pixels.

[0029] Step 2: Input the preprocessed image into the pre-trained ventilation component recognition model to obtain the positions of the impeller, instrument and pipe in the preprocessed image, and determine the impeller image, instrument image and pipe image based on the positions of the impeller, instrument and pipe in the preprocessed image; In this embodiment of the disclosure, the ventilation component recognition model is trained based on the YOLOv8 algorithm with added attention mechanism module.

[0030] It should be noted that the component detection is based on the improved YOLOv8 algorithm, with the addition of an attention mechanism module, to perform real-time positioning of key components such as the fan impeller, casing, pipelines, and pressure gauges.

[0031] Step 3: Perform fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; In this embodiment, step 3 specifically includes: calculating the blade motion trajectory based on the impeller image using optical flow field and extracting vibration frequency features through Fourier transform; when the difference between the vibration frequency and a preset vibration frequency threshold is greater than 5Hz, it is determined that the impeller is abnormal; based on the instrument image and using Hough line detection and OCR character recognition methods, identify the bearing temperature corresponding to the instrument image; when the bearing temperature is greater than a preset bearing temperature threshold, it is determined that the fan bearing is worn; based on the pipeline image and using an improved background difference method with a set adaptive threshold, identify the area of ​​the oil leakage area; when the ratio of the oil leakage area to the background area of ​​the casing is greater than the set adaptive threshold, it is determined that an oil leakage has occurred.

[0032] It should be noted that the state analysis and impeller fault identification are as follows: the blade motion trajectory is calculated by optical flow field, and the vibration frequency characteristics are extracted by Fourier transform. When the frequency deviation is >5Hz, it is judged as abnormal (detection time <200ms).

[0033] Instrument reading recognition: Combining Hough line detection and OCR character recognition, the reading error of pointer instruments is ≤0.5kPa, and the recognition accuracy of digital instruments is ≥99.3%.

[0034] Leakage detection: An improved background difference method is used, and an adaptive threshold is set. When the oil leakage area of ​​the casing / the background area of ​​the casing is greater than the set threshold, it is determined that an oil leak has occurred.

[0035] Step 4: Determine the fault type and classify the fault level based on the analysis results, and then generate control commands according to the fault level and send them to the wind turbine control system through the industrial communication protocol to execute the linkage control operation.

[0036] In this embodiment of the disclosure, the step of determining the fault type and classifying the fault level based on the analysis results includes: when an impeller abnormality occurs, determining the fault level of the fan as a Level I emergency fault; when bearing wear occurs, determining the fault level of the fan as a Level II major fault; and when oil leakage occurs, determining the fault level of the fan as a Level III general fault.

[0037] The generation of control commands based on the fault level includes: when the fault level is Level I emergency fault, generating commands to immediately shut down and start the standby fan, and sending alarm information to maintenance personnel; when the fault level is Level II important fault, generating a load reduction operation command; and when the fault level is Level III general fault, generating a command to send early warning information.

[0038] It should be noted that the fault level is divided into three levels according to the scope of the fault's impact: Level I (emergency): such as impeller breakage, triggering immediate shutdown; Level II (important): such as bearing temperature exceeding the standard, triggering reduced load operation; Level III (general): such as minor pipeline leakage, only issuing a warning.

[0039] Control command generation: The system automatically generates control commands based on the fault level and transmits them to the wind turbine PLC control system via the Modbus protocol. The command response time is <1 second.

[0040] Emergency Response: When a Level I fault is detected, the backup fan is automatically started, and an alarm message containing the fault location and image is sent to the maintenance personnel via the 4G module.

[0041] In summary, the video inspection method for the main ventilation room on the ground proposed in this embodiment achieves full coverage, high precision, real-time monitoring and intelligent linkage of visual and status faults of ventilation equipment, significantly improving the safety and reliability of the mine ventilation system.

[0042] Figure 2 is a structural diagram of a ground-based main ventilation fan room video inspection system according to an embodiment of this application. As shown in Figure 2, the system includes: a data acquisition module 100, used to acquire images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the fan impeller area, instrument area, and pipeline area, and to perform dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; wherein, the deployment method of the intrinsically safe high-definition camera group includes: deploying a camera with a telephoto lens dynamic module in the fan impeller area; deploying a camera with a fixed-focus lens and an infrared fill light module in the instrument area; and deploying a camera with a wide-angle lens in the pipeline area.

[0043] The recognition module 200 is used to input the preprocessed image into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instrument and pipe in the preprocessed image, and to determine the impeller image, instrument image and pipe image based on the positions of the impeller, instrument and pipe in the preprocessed image; wherein, the ventilation component recognition model is trained based on the YOLOv8 algorithm with an added attention mechanism module.

[0044] The fault analysis module 300 is used to perform fault analysis on the impeller image, instrument image and pipeline image to obtain fault analysis results; the control module 400 is used to determine the fault type and classify the fault level according to the analysis results, and then generate control commands according to the fault level and send them to the fan control system through the industrial communication protocol to execute linkage control operations.

[0045] In this embodiment of the disclosure, the acquisition module 100 is further configured to: perform multi-frame collaborative filtering on the acquired image using an improved BM3D algorithm to remove dust particle noise; calculate the motion vector between adjacent frames of the image after removing dust particle noise based on SIFT feature point matching and perform reverse mapping to achieve vibration anti-shake correction.

[0046] In this embodiment of the disclosure, the fault analysis module 300 is further configured to: calculate the blade motion trajectory based on the impeller image using optical flow field, and extract vibration frequency features through Fourier transform; determine impeller abnormality when the difference between the vibration frequency and a preset vibration frequency threshold is greater than 5Hz; identify the bearing temperature corresponding to the instrument image based on the instrument image using Hough line detection and OCR character recognition methods; determine bearing wear of the fan when the bearing temperature is greater than a preset bearing temperature threshold; identify the oil leakage area based on the pipeline image using an improved background difference method with a set adaptive threshold; determine oil leakage when the ratio of the oil leakage area to the background area of ​​the casing is greater than the set adaptive threshold.

[0047] In this embodiment of the disclosure, the control module 400 is further configured to: determine the fault level of the fan as Level I emergency fault when an impeller abnormality occurs; determine the fault level of the fan as Level II important fault when bearing wear occurs; and determine the fault level of the fan as Level III general fault when oil leakage occurs.

[0048] In this embodiment of the disclosure, the control module 400 is further configured to: generate instructions to immediately shut down and start the standby fan, and send alarm information to maintenance personnel when the fault level is Level I emergency fault; generate load reduction operation instructions when the fault level is Level II important fault; and generate instructions to send early warning information when the fault level is Level III general fault.

[0049] In summary, the video inspection system for the main ventilation room on the ground proposed in this embodiment achieves full coverage, high precision, real-time monitoring and intelligent linkage of visual and status faults of ventilation equipment, significantly improving the safety and reliability of the mine ventilation system.

[0050] Example 3 In order to implement the above embodiments, this disclosure also proposes an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the method described in Example 1.

[0051] Example 4 In order to implement the above embodiments, this disclosure also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in Example 1.

[0052] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0053] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0054] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A video inspection method for a ground-based main ventilation fan room, characterized in that, The method includes: acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the impeller area, instrument area, and pipeline area of ​​the fan; performing dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; inputting the pre-processed images into a pre-trained ventilation component recognition model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images, and determining the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; performing fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; determining the fault type and classifying the fault level according to the analysis results; and then generating control commands according to the fault level and sending them to the fan control system via an industrial communication protocol to execute linkage control operations.

2. The method as described in claim 1, characterized in that, The deployment methods of the intrinsically safe high-definition camera group include: deploying cameras with telephoto lens dynamic modules in the fan impeller area; deploying cameras with fixed-focus lenses and infrared fill light modules in the instrument area; and deploying cameras with wide-angle lenses in the pipeline area.

3. The method as described in claim 2, characterized in that, The process of performing dust noise reduction and vibration stabilization on the image to obtain a preprocessed image includes: using an improved BM3D algorithm to perform multi-frame collaborative filtering on the acquired image to remove dust particle noise; calculating the motion vectors between adjacent frames of the image after removing dust particle noise based on SIFT feature point matching and performing reverse mapping to achieve vibration stabilization.

4. The method as described in claim 1, characterized in that, The ventilation component recognition model is trained using the YOLOv8 algorithm with an added attention mechanism module.

5. The method as described in claim 3, characterized in that, The fault analysis of the impeller image, instrument image, and pipeline image to obtain fault analysis results includes: calculating the blade motion trajectory based on the impeller image using optical flow field and extracting vibration frequency features through Fourier transform; determining an impeller abnormality when the difference between the vibration frequency and a preset vibration frequency threshold is greater than 5Hz; identifying the bearing temperature corresponding to the instrument image using Hough line detection and OCR character recognition methods; determining bearing wear of the fan when the bearing temperature is greater than a preset bearing temperature threshold; and identifying the oil leakage area using an improved background difference method with a set adaptive threshold based on the pipeline image; determining an oil leakage when the ratio of the oil leakage area to the background area of ​​the casing is greater than the set adaptive threshold.

6. The method as described in claim 5, characterized in that, The step of determining the fault type and classifying the fault level based on the analysis results includes: when an impeller abnormality occurs, the fault level corresponding to the fan is determined to be Level I emergency fault; when bearing wear occurs, the fault level corresponding to the fan is determined to be Level II important fault; when oil leakage occurs, the fault level corresponding to the fan is determined to be Level III general fault.

7. The method as described in claim 6, characterized in that, The generation of control commands based on the fault level includes: when the fault level is Level I emergency fault, generating commands to immediately shut down and start the standby fan, and sending alarm information to maintenance personnel; when the fault level is Level II important fault, generating a load reduction operation command; and when the fault level is Level III general fault, generating a command to send early warning information.

8. A video inspection system for a ground-based main ventilation fan room, characterized in that, The system includes: an acquisition module for acquiring images at a dynamic frame rate using an intrinsically safe high-definition camera group deployed in the impeller area, instrument area, and pipeline area of ​​the fan, and performing dust noise reduction and vibration anti-shake correction on the images to obtain pre-processed images; an identification module for inputting the pre-processed images into a pre-trained ventilation component identification model to obtain the positions of the impeller, instruments, and pipelines in the pre-processed images, and determining the impeller image, instrument image, and pipeline image based on the positions of the impeller, instruments, and pipelines in the pre-processed images; a fault analysis module for performing fault analysis on the impeller image, instrument image, and pipeline image to obtain fault analysis results; and a control module for determining the fault type and classifying the fault level according to the analysis results, and then generating control commands according to the fault level and sending them to the fan control system through an industrial communication protocol to execute linkage control operations.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images