A smart monitoring system for power plant equipment based on AI video fusion

Abstract

This invention relates to the field of power plant equipment monitoring technology and discloses an intelligent monitoring system for power plant equipment based on AI video fusion. The invention employs a short-wave infrared sensor to penetrate dusty environments and collect video streams from the equipment. It reconstructs the dynamic vibration field across the entire surface of the equipment using phase amplification and optical flow algorithms. A Structural Resonance Integrity Index (SRII) is proposed to quantify the health status of the equipment based on the overall morphological differences of the vibration field. A dynamic interference mask is generated by combining personnel / vehicle positioning signals to shield against non-equipment motion interference in real time. When the SRII is abnormal, a laser vibrometer is intelligently triggered to perform high-precision calibration of the vibration peak area. This overcomes the technical bottlenecks of low accuracy and weak anti-interference capability in full-domain vibration monitoring under complex power plant environments, achieving early and accurate warning of equipment structural damage and significantly improving the intelligence level and operational reliability of power plant safety management.

Description

Technical Field

[0001] This invention relates to the field of power plant equipment monitoring technology, and in particular to an intelligent monitoring system for power plant equipment based on AI video fusion. Background Technology

[0002] In modern industrial production, especially in power plant operation, the stability and safety of equipment are paramount. Traditional equipment monitoring methods, such as manual inspections and contact sensors, have many limitations. Manual inspections are inefficient, unable to detect equipment anomalies in real time, and pose safety risks. While contact sensors can provide accurate local vibration data, they are complex to install, costly, and cannot monitor vibrations over the entire surface of the equipment, making them particularly unsuitable for large, high-temperature, or inaccessible equipment. Furthermore, the complex environment of power plants, with factors such as dust, high temperatures, and electromagnetic interference, often affects the accuracy of monitoring data.

[0003] While some non-contact monitoring methods, such as laser vibrometers or ordinary visible light cameras, have overcome the limitations of contact sensors to some extent, they still face challenges. Laser vibrometers typically only scan single points or limited areas, making it difficult to acquire full-area vibration information of equipment surfaces. Ordinary visible light cameras lack penetration in the dusty environment of power plants, resulting in limited image quality. They are also susceptible to interference from changes in ambient lighting and non-equipment movement such as personnel or vehicles, leading to inaccurate monitoring data and even false alarms. Furthermore, existing technologies often rely on single indicators to assess equipment operating status, lacking a comprehensive judgment of the equipment's structural integrity and making it difficult to effectively predict potential equipment failure risks. Therefore, how to achieve full-area, non-contact, high-precision, and interference-resistant intelligent monitoring of power plant equipment, and effectively assess the health status of equipment structures, is a pressing technical problem that needs to be solved in the field of industrial monitoring. Summary of the Invention

[0004] The technical problem to be solved by this invention is that existing power plant equipment monitoring technologies suffer from insufficient non-contact monitoring accuracy, poor anti-interference ability, inability to monitor the entire area, and lack of comprehensive structural health assessment. To address this, we propose an intelligent monitoring system for power plant equipment based on AI video fusion.

[0005] To achieve the above objectives, this application adopts the following technical solution: an intelligent monitoring system for power plant equipment based on AI video fusion, comprising:

[0006] Shortwave infrared video acquisition module: includes a shortwave infrared sensor, which is an indium gallium arsenide image sensor with a response wavelength range of 900-1700nm, and is equipped with an anti-electromagnetic interference shield to penetrate the video stream on the surface of the power plant dust environment acquisition equipment and generate a first optical signal;

[0007] The core measurement module, connected to the shortwave infrared video acquisition module, includes:

[0008] Vibration field reconstruction unit: Converts the video stream into a dynamic vibration displacement matrix using a phase motion amplification algorithm and dense optical flow method. Characterizes the global vibration pattern of the equipment surface;

[0009] Structural health analysis unit: Calculates the structural resonance integrity index. ,in The reference vibration mode matrix is ​​pre-calibrated and stored for the equipment in a healthy state. The stress coefficient is determined through material fatigue testing, with a value ranging from 0.5 to 1.2. Denotes the Frobenius norm of a matrix;

[0010] Multi-source interference suppression module: Connects to UWB personnel positioning system and RFID vehicle positioning system to generate dynamic binary mask matrix. The area with a mask value of 1 corresponds to non-device motion interference;

[0011] Closed-loop control module: when Less than the preset threshold At that time, the laser Doppler vibration meter was triggered to... Perform point scan calibration in the area with the largest mid-amplitude values ​​and increase the video sampling rate to 100fps;

[0012] Monitoring result output module: Used to receive output data from the core measurement module and the multi-source interference suppression module, and display the equipment monitoring results.

[0013] Preferably, the vibration field reconstruction unit performs:

[0014] Blackbody radiation compensation: Correcting radiated noise in video streams from the surface of high-temperature equipment. ;in, This is the video stream intensity value after radiated noise compensation. The raw video stream acquired by the shortwave infrared sensor at the location Location and Time The intensity value at that location, For material emissivity, The second radiation constant, The temperature field acquired by the infrared camera represents the temperature field at location. Location and Time The absolute temperature at that location, This refers to the operating wavelength of the shortwave infrared sensor.

[0015] Phase motion amplification: Perform complex manipulatory pyramid decomposition on the compensated video to extract the phase components. The vibration of the target frequency band is amplified by time bandpass filtering.

[0016] Preferred multi-source interference suppression module:

[0017] UWB positioning coordinates Mapped to image pixel region ;

[0018] Forced setting in dense optical flow calculation ;in For the final optical flow vector at position Location and Time The value at that location.

[0019] Preferably, after the closed-loop control module is triggered:

[0020] Control the shortwave infrared sensor for the area Optical zoom with a focal length ≥200mm;

[0021] The scanning path of the laser Doppler vibrometer is adjusted synchronously to cover the zoom area, achieving a sampling accuracy of 0.1 μm.

[0022] Preferred options also include:

[0023] Beidou Time and Space Synchronization Unit: Receives PPS pulse timing signals triggered by access control / vehicle events with an accuracy of ±0.3ms, aligns with video stream and UWB / RFID data timestamps, and has a spatial positioning error of ≤0.3m.

[0024] Preferably, the shortwave infrared sensor includes:

[0025] Polarizing filter switching mechanism: When the ambient light intensity is >100,000 Lux, it automatically switches to polarizing imaging mode to suppress specular reflection noise on the metal surface;

[0026] Active cooling device: Maintains sensor temperature ≤45℃ to avoid thermal noise drift.

[0027] Preferably, the monitoring result output module displays:

[0028] Dynamic vibration displacement matrix The pseudo-color cloud map maps the vibration amplitude using color levels;

[0029] The SRII index curve over time, with the threshold θ and historical extreme values ​​marked;

[0030] Dynamic interference mask The boundary outline.

[0031] Preferred core measurement module:

[0032] In response to device start / stop events sent by the access control system and passing events sent by the vehicle system, the corresponding baseline matrix is ​​dynamically loaded. :

[0033] When the equipment is shut down, a static thermal deformation compensation matrix is ​​applied.

[0034] A typical vibration interference template matrix is ​​applied when a vehicle passes by.

[0035] The technical effects and advantages of this invention are as follows:

[0036] This invention utilizes short-wave infrared optical sensing to penetrate dusty environments, combines phase motion amplification with optical flow methods to reconstruct the global dynamic vibration field, and innovatively defines the SRII index to quantify structural health status. This solves the problem that traditional monitoring methods cannot achieve global, accurate, and interference-resistant measurements in the complex environment of power plants. Simultaneously, it uses UWB / RFID positioning to generate dynamic masks to shield against interference from personnel / vehicles in real time, and triggers laser calibration of abnormal areas through closed-loop control, achieving early warning of equipment structural damage and significantly improving the intelligence level and operational reliability of power plant safety management. Attached Figure Description

[0037] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts:

[0038] Figure 1 This is a block diagram of the module structure of the present invention;

[0039] Figure 2 This is a logical structure diagram of the present invention. Detailed Implementation

[0040] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0041] Example 1

[0042] Reference Figures 1-2 As shown, the present invention provides an intelligent monitoring system for power plant equipment based on AI video fusion, which mainly includes a short-wave infrared video acquisition module, a core measurement module, a multi-source interference suppression module, and a closed-loop control module.

[0043] Shortwave Infrared Video Acquisition Module: This module is the system's data input, acquiring video streams from the surface of power plant equipment via a shortwave infrared sensor. Its core component is an indium gallium arsenide (IGaAs) image sensor with a response wavelength range of 900-1700nm. IGaAs sensors exhibit high sensitivity within this wavelength range, effectively penetrating common media in power plant environments such as dust, water vapor, and smoke to obtain clear images of the equipment surface, overcoming the insufficient penetration of visible light cameras in harsh environments. To ensure the stability of data acquisition, this module is also equipped with an anti-electromagnetic interference shield, effectively suppressing interference from the complex electromagnetic environment of the power plant on the sensor signal and ensuring the quality of the video stream. The acquired video stream serves as the first optical signal, transmitted to the core measurement module for subsequent processing.

[0044] In a preferred embodiment, the shortwave infrared sensor further includes a polarization filter switching mechanism and an active cooling device. The polarization filter switching mechanism automatically switches to polarization imaging mode when the ambient light intensity is greater than 100,000 Lux, thereby suppressing specular reflection noise from metal surfaces and further improving image quality. This is particularly suitable for scenarios in power plants where there are a large number of metal devices and strong light sources. The active cooling device maintains the sensor temperature below or equal to 45°C to avoid thermal noise drift caused by excessive sensor temperature, ensuring the stability and accuracy of image data.

[0045] Core Measurement Module: This module is the core processing unit of the system. It is connected to the shortwave infrared video acquisition module and is responsible for converting the acquired video stream into quantifiable vibration data and performing structural health assessment. The core measurement module includes a vibration field reconstruction unit and a structural health analysis unit.

[0046] Vibration Field Reconstruction Unit: This unit is crucial for achieving full-area vibration monitoring of the equipment surface. It converts the video stream input from the short-wave infrared video acquisition module into a dynamic vibration displacement matrix by combining a phase motion amplification algorithm and a dense optical flow method. This matrix represents the vibration displacement information of the m rows and n columns of pixels on the surface of the device at time t, thus realizing the accurate representation of the vibration pattern of the entire surface of the device.

[0047] In a preferred embodiment, the vibration field reconstruction unit performs blackbody radiation compensation before performing phase motion amplification; the blackbody radiation compensation unit corrects the radiation noise of the video stream on the surface of the high-temperature equipment using the following correction formula: ;in, The original video stream intensity. This is the video stream intensity value after radiated noise compensation. The raw video stream acquired by the shortwave infrared sensor at the location Location and Time The intensity value at that location, For material emissivity, The second radiation constant, The temperature field acquired by the infrared camera represents the temperature field at location. Location and Time The absolute temperature at that location, This is the operating wavelength of the short-wave infrared sensor; this compensation effectively eliminates the interference of high-temperature equipment's own radiation on the video signal, improving the accuracy of vibration measurement; subsequently, the phase motion amplification unit performs complex manipulatory pyramid decomposition on the compensated video to extract the phase components. The vibration in the target frequency band is amplified through time bandpass filtering; specifically, phase motion amplification includes:

[0048] Perform complex manipulable pyramid decomposition on video frames to extract scale. ,direction and phase ,right Perform time bandpass filtering with a passband of 5-500Hz to separate vibration components. ;

[0049] Will Multiply by magnification factor , This allows for the reconstruction of motion-magnified video; in this way, even minute, imperceptible vibrations can be effectively amplified and identified.

[0050] Structural health analysis unit: This unit receives the dynamic vibration displacement matrix output by the vibration field reconstruction unit. And calculate the structural resonance integrity index SRII; the formula for calculating SRII is: ;in, This is the health status baseline matrix, representing the vibration characteristics of the equipment under normal health conditions; The stress coefficient, calibrated through material fatigue testing, ranges from 0.5 to 1.2 and is used to adjust the degree of influence of vibration deviation on SRII. The SRII index is the Frobenius norm, used to measure the "size" or "energy" of a matrix; it quantifies the deviation of the current vibration state of equipment from a healthy baseline, thereby assessing the structural integrity of the equipment; the closer the SRII value is to 1, the healthier the equipment structure; the smaller the SRII value, the more likely the equipment has structural damage or abnormalities.

[0051] In a preferred embodiment, the core measurement module also dynamically loads the corresponding reference matrix in response to device start / stop events sent by the access control system and passing events sent by the vehicle system. Specifically, when the equipment stops, a static thermal deformation compensation matrix is ​​loaded as... To eliminate the impact of deformation caused by thermal expansion and contraction of equipment on vibration assessment; when a vehicle passes by, a typical vibration disturbance template matrix is ​​loaded as... This is to eliminate the influence of vehicle vibration on equipment vibration when calculating SRII, thus ensuring the accuracy of SRII.

[0052] Multi-source interference suppression module: This module aims to eliminate interference from non-equipment movement in the power plant environment on vibration monitoring, such as personnel walking or vehicles passing by. It connects to an ultra-wideband (UWB) personnel positioning system and an RFID vehicle positioning system to acquire real-time location information of personnel and vehicles. Based on this positioning data, the module generates a dynamic binary mask matrix. ;exist In the diagram, areas with a mask value of 1 correspond to non-device motion interference areas, i.e., areas where personnel or vehicles are located, while areas with a mask value of 0 correspond to the equipment itself.

[0053] In a preferred embodiment, the multi-source interference suppression module locates the UWB coordinates. Mapped to image pixel region In dense optical flow calculations, the optical flow vector is forcibly set. for: ; For the final optical flow vector at position Location and Time The value at this location; this means that within the identified interference region, the optical flow vector is forced to zero, thereby effectively suppressing the interference of motion in these regions on the vibration field reconstruction, ensuring... It only reflects the vibration of the equipment itself.

[0054] Closed-loop control module: This module is the intelligent feedback mechanism of this system; it continuously monitors the SRII index output by the structural health analysis unit; when the SRII value is lower than a preset threshold... When this occurs, it indicates a possible abnormality or potential malfunction in the equipment; at this time, the closed-loop control module will immediately trigger the laser Doppler vibration meter to... Point scan calibration is performed in the area with the largest mid-amplitude values ​​to obtain more accurate vibration data for that area; at the same time, the system automatically increases the video sampling rate of the short-wave infrared video acquisition module to 100fps to capture finer vibration details.

[0055] In a preferred embodiment, after the closed-loop control module is triggered, it controls the short-wave infrared sensor to monitor the area. Optical zoom with a focal length ≥200mm is used to make the area with the largest amplitude occupy a larger proportion in the image, thereby improving its detail resolution. At the same time, the scanning path of the laser Doppler vibrometer is adjusted synchronously to accurately cover the zoomed area and ensure that the sampling accuracy reaches 0.1μm. This linkage mechanism enables focused, high-precision, and high-frequency monitoring of abnormal areas, providing more reliable data support for fault diagnosis.

[0056] Beidou Spatiotemporal Synchronization Unit: In a preferred embodiment, the system further includes a Beidou spatiotemporal synchronization unit; this unit receives PPS pulse timing signals triggered by the access control system or vehicle system with an accuracy of ±0.3ms, aligns the timestamps of the video stream and UWB / RFID data, and ensures high temporal synchronization of all sensor data; at the same time, its spatial positioning error is less than or equal to 0.3m, which provides a guarantee for the accurate mapping of UWB / RFID positioning data and video pixels, and further improves the accuracy of multi-source data fusion.

[0057] Monitoring result output module: In a preferred embodiment, the system further includes a monitoring result output module for intuitively displaying the monitoring results; this module can display:

[0058] Dynamic vibration displacement matrix The pseudo-color cloud map maps vibration amplitude with color levels, allowing operators to intuitively see the vibration intensity distribution in different areas of the equipment surface.

[0059] The SRII index changes over time, with preset thresholds marked. Historical extreme values ​​are available to help operators understand the changing trends and abnormal conditions of the equipment's structural health.

[0060] Dynamic interference mask The boundary contours clearly indicate the non-device motion interference areas that have been identified and suppressed by the system, enhancing the transparency and reliability of the system.

[0061] Example 2

[0062] This invention also provides an intelligent monitoring method for power plant equipment based on AI video fusion. This method can be implemented by the above-mentioned system and includes the following steps:

[0063] Step 1: Acquire video stream from the surface of the device using a shortwave infrared sensor.

[0064] This step utilizes the indium gallium arsenide (IGaAs) image sensor in the short-wave infrared video acquisition module to acquire video streams from the surface of power plant equipment at a frame rate of 30fps and a resolution of 1920×1080. The characteristics of the short-wave infrared band ensure that the video stream can penetrate dust and smoke in the power plant environment to obtain clear images of the equipment. At the same time, an anti-electromagnetic interference shield ensures the purity of the video signal. When the ambient light intensity is high, the polarization filter switching mechanism automatically switches to polarization imaging mode to suppress specular reflection noise. The active cooling device maintains the sensor temperature stability and avoids thermal noise.

[0065] Step 2: Perform blackbody radiation compensation and phase motion amplification to generate a dynamic vibration displacement matrix. .

[0066] This step is completed in the vibration field reconstruction unit of the core measurement module. First, blackbody radiation compensation is performed on the acquired video stream, especially for high-temperature equipment surfaces. Radiation noise is corrected using a formula to obtain the compensated video stream. Then, phase motion amplification is performed on the compensated video stream. Specifically, complex manipulable pyramid decomposition is performed on the video frames to extract scale. ,direction and phase Next, regarding Time bandpass filtering was performed, with a passband of 5-500Hz, to separate the vibration components. Finally, Multiply by magnification factor , The motion-enlarged video was reconstructed, and the dynamic vibration displacement matrix characterizing the global vibration pattern of the device surface was finally generated by calculating using the dense optical flow method. .

[0067] Step 3: Generate a dynamic mask based on the UWB / RFID signal to suppress non-device motion vectors.

[0068] This step is completed in the multi-source interference suppression module; the system connects to the UWB personnel positioning system and the RFID vehicle positioning system to obtain the precise location information of personnel and vehicles in real time; and the UWB positioning coordinates are... Mapped to image pixels This area is the non-device motion interference area; in subsequent dense optical flow calculations, it is forcibly set to... optical flow vector within the region It is 0, that is In this way, the interference of personnel and vehicle movement on equipment vibration measurement is effectively eliminated, ensuring the accuracy of the vibration displacement matrix.

[0069] Step 4: Calculate the SRII index. If SRII < And it continues for 3 cycles, triggering a laser calibration scan.

[0070] This step is completed collaboratively within the structural health analysis unit and closed-loop control module of the core measurement module; the system calculates based on the dynamic vibration displacement matrix. and the preset health status benchmark matrix Calculate the structural resonance integrity index SRII: The system continuously monitors the SRII index, and when the SRII value falls below a preset threshold... Furthermore, if this state persists for three monitoring cycles, the closed-loop control module will trigger the laser Doppler vibration meter to... The area with the largest amplitude A point scan calibration is performed to obtain more accurate vibration data for the area; at the same time, the video sampling rate of the short-wave infrared video acquisition module is increased to 100fps to capture finer vibration details; when the calibration scan is triggered, the short-wave infrared sensor will also optically zoom the area with a focal length ≥200mm, and the scanning path of the laser Doppler vibrometer will be adjusted synchronously to cover the zoomed area, with a sampling accuracy of 0.1μm.

[0071] Step 5: Output vibration contour plot, SRII curve and mask boundary;

[0072] This step is completed in the monitoring result output module; the system presents the processed monitoring results to the operator in an intuitive way, including:

[0073] Dynamic vibration displacement matrix The pseudo-color cloud map uses different shades of color to represent the magnitude of vibration, making it easy to quickly identify abnormal areas.

[0074] The SRII index curve over time clearly shows the trend of the equipment's structural health status and marks the threshold values. Historical extreme values ​​help determine the health status of equipment;

[0075] Dynamic interference mask The boundary outline is marked with a red dashed line, clearly indicating the non-device motion interference areas that have been identified and suppressed by the system, thus enhancing the interpretability of the system.

[0076] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. An AI video fusion-based intelligent monitoring system for power plant equipment, characterized in that, Comprise: Short-wave infrared video acquisition module: containing short-wave infrared sensor, the short-wave infrared sensor uses indium gallium arsenide image sensor with response wavelength range of 900-1700nm, and is provided with anti-electromagnetic interference shield cover, for collecting equipment surface video stream through power plant dust environment, and generating first optical signal; Core measurement module connected with the short-wave infrared video acquisition module, comprising: Vibration field reconstruction unit: convert video stream into dynamic vibration displacement matrix by phase motion amplification algorithm and dense optical flow method , representing the global vibration pattern of the device surface; Structural health analysis unit: calculate structural resonance integrity index: wherein is a reference vibration modal matrix pre-calibrated and stored by the device in a healthy state, is a stress coefficient calibrated by material fatigue test, with a value range of 0.5-1.2, denotes the Frobenius norm of a matrix; Multi-source interference suppression module: access ultra-wideband (UWB) personnel positioning system and radio frequency identification (RFID) vehicle positioning system to generate a dynamic binary mask matrix wherein the region with mask value 1 corresponds to non-device motion interference; Closed-loop control module: when Less than the preset threshold At that time, the laser Doppler vibration meter was triggered to... Perform point scan calibration in the area with the largest mid-amplitude values ​​and increase the video sampling rate to 100fps; Monitoring result output module: for receiving the output data of core measurement module and multi-source interference suppression module, and displaying equipment monitoring result; The vibration field reconstruction unit executes: Blackbody radiation compensation: radiation noise correction on high temperature equipment surface video stream: ; wherein, is the intensity value of the video stream after radiation noise compensation, is the intensity value of the original video stream collected by the short-wave infrared sensor at position and time , is the material emissivity, is the second radiation constant, is the temperature field collected by the infrared camera, indicating the absolute temperature at position and time , is the working wavelength of the short-wave infrared sensor; Phase motion amplification: complex steerable pyramid decomposition is performed on the compensated video, extracting the phase component , amplifying the target frequency band vibration through time band-pass filtering; The multi-source interference suppression module: UWB positioning coordinates mapping to image pixel regions ; Forced setting in dense optical flow computation ; wherein is the value of the final used optical flow vector at position and time .

2. The power plant equipment intelligent monitoring system based on AI video fusion according to claim 1, characterized in that, After the closed loop control module is triggered: Controlling short-wave infrared sensors to regions Optical zoom, focal length > 200 mm; Synchronous adjustment of the scanning path of the laser Doppler vibrometer, covering the zoom area, and the sampling accuracy reaches 0.1 μm. 3.The power plant equipment intelligent monitoring system based on AI video fusion of claim 1, wherein, Also comprising: Beidou space-time synchronization unit: receiving PPS pulse time signal triggered by access control / vehicle event, accuracy ±0.3ms, aligning video stream, UWB / RFID data timestamp, spatial positioning error≤0.3m.

4. The power plant equipment intelligent monitoring system based on AI video fusion according to claim 1, characterized in that, The short-wave infrared sensor comprises: Polarization filter switching mechanism: when the ambient light intensity is greater than 100,000Lux, automatically switch to polarization imaging mode to suppress metal surface specular reflection noise; Active cooling device: maintain the sensor temperature less than or equal to 45℃, avoid thermal noise drift.

5. The power plant equipment intelligent monitoring system based on AI video fusion according to claim 1, characterized in that, The monitoring result output module displays: Dynamic vibration displacement matrix Pseudo-color cloud plot with color scale mapping vibration amplitude; SRII index curve changes with time, mark threshold θ and historical extreme value; Dynamic interference mask of the boundary contour.

6. The power plant equipment intelligent monitoring system based on AI video fusion according to claim 1, characterized in that, The core measurement module: In response to the device start-stop event sent by the access control system and the passing event sent by the vehicle system, a corresponding reference matrix is dynamically loaded : Load static thermal deformation compensation matrix when the equipment is stopped; Load typical vibration interference template matrix when the vehicle passes.

7. An AI video fusion-based power plant equipment intelligent monitoring method, implemented based on an AI video fusion-based power plant equipment intelligent monitoring system according to any one of claims 1-6. Comprise: S1: collect equipment surface video stream through short-wave infrared sensor; S2: Perform blackbody radiation compensation and phase motion amplification, generate dynamic vibration displacement matrix ; S3: generate dynamic mask based on UWB / RFID signal to suppress non-equipment motion vector; S4: calculate SRII index, if SRII<θ and lasts for 3 periods, trigger laser calibration scanning; S5: output vibration cloud picture, SRII curve and mask boundary.

8. The power plant equipment intelligent monitoring method based on AI video fusion according to claim 7, characterized in that, Phase motion amplification in step S2: S21. performing a complex steerable pyramid decomposition on the video frame, extracting scales , directions and phases ; S22. The method of S21, further comprising: time bandpass filtering to separate the vibration component ; S23. multiplying the motion magnification factor by the scaling factor , reconstructing the motion magnified video.​​

Images