Substation multi-target photoelectric monitoring device and method based on multispectral fusion

Abstract

The application discloses a kind of based on multispectral fusion's transformer substation multi-objective photoelectric monitoring device and method, transformer substation multi-objective photoelectric monitoring device includes multispectral image acquisition module and multispectral fusion temperature measurement processing module;Multispectral image acquisition module is used to obtain the mid-infrared thermal imaging image of the to-be-measured electric power equipment, near-infrared temperature data, visible light imaging picture data;Multispectral fusion temperature measurement processing module includes data fusion unit and bias calibration unit;Data fusion unit is used to process mid-infrared thermal imaging image, near-infrared temperature data, visible light imaging picture data, carries out to-be-measured electric power equipment identification;Bias calibration unit is used to carry out bias calibration to the pixel of mid-infrared thermal imaging image according to near-infrared temperature data, based on physical radiation transfer equation and neural network residual fitting, obtains pixel-level temperature measurement data.The device and method disclosed in the application can improve the accuracy of electric power equipment temperature measurement and the reliability of monitoring data.

Description

Technical Field

[0001] This invention relates to the field of power equipment safety monitoring technology, and in particular to a multi-target photoelectric monitoring device and method for substations based on multispectral fusion. Background Technology

[0002] During long-term operation, various electrical equipment in substations may experience problems such as poor contact, insulation aging, and overload operation. These problems often manifest as abnormal heating of the equipment, making electrical equipment temperature monitoring a crucial means of assessing equipment operating status in power grid maintenance. Conventional fixed infrared temperature measurement systems use a single infrared lens to collect data, and the measurement results are easily affected by factors such as detection distance, ambient temperature, and atmospheric transmittance, resulting in significant temperature drift and making it difficult to achieve high-precision temperature measurement of electrical equipment.

[0003] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a multi-target photoelectric monitoring device and method for substations based on multispectral fusion, so as to solve the problem that the existing temperature measurement system is not accurate enough for wide-area and long-distance power equipment.

[0005] The technical solution of the present invention is as follows: In a first aspect, the present invention provides a multi-target photoelectric monitoring device for substations based on multispectral fusion, comprising: a multispectral image acquisition module and a multispectral fusion temperature measurement and processing module; The multispectral image acquisition module is used to acquire mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test. The multispectral fusion temperature measurement processing module includes a data fusion unit and a deviation calibration unit; The data fusion unit is used to process the mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data to identify the power equipment under test and extract the area of ​​the power equipment under test. The deviation calibration unit is used to perform deviation calibration on the pixels of the mid- and far-infrared thermal imaging image based on the near-infrared temperature data, using the physical radiation transfer equation and neural network residual fitting, to obtain pixel-level temperature measurement data.

[0006] In a further embodiment of the present invention, the monitoring device further includes a spatial positioning module for acquiring the spatial location information of the power equipment under test. The spatial positioning module includes a binocular ranging unit and an angle calculation unit. The binocular ranging unit is used to acquire the spatial distance of the power equipment under test, and the angle calculation unit is used to acquire the azimuth angle data of the power equipment under test.

[0007] In a further embodiment of the present invention, the monitoring device also includes an inspection and target tracking module; the inspection and target tracking module includes a tracking calculation control unit and a CNC pan-tilt unit, and the multispectral image acquisition module is fixed on the CNC pan-tilt unit; the tracking calculation control unit locates the power equipment under test based on the spatial location information of the power equipment under test obtained by the spatial positioning module, and issues a movement tracking command to the CNC pan-tilt unit; the inspection and target tracking module embeds a deep learning algorithm to automatically detect personnel and foreign object intrusion targets, and performs identification, capture, storage and uploading operations on intrusion targets.

[0008] A further feature of the present invention is that the visible light imaging unit captures the visible light imaging image data. The monitoring device also includes a multi-mode temperature measurement execution module, which is used to set the temperature measurement points within a single field of view of the visible light imaging unit on the CNC pan-tilt unit. The shape of the temperature measurement area formed by the temperature measurement points is unlimited, and multiple temperature measurement areas are measured in parallel.

[0009] A further provision of the present invention includes the step of processing the mid- and far-infrared thermal imaging image, the near-infrared temperature data, and the visible light imaging image data, comprising: The visible light imaging data and the mid- and far-infrared thermal imaging image are registered pixel-level using a deep learning network to obtain registered image data. The visible light imaging data and the mid- and far-infrared thermal imaging image are then fed into a trained semantic segmentation network to complete the identification of the power equipment under test and extract the region of the power equipment under test.

[0010] In a further embodiment of the present invention, the monitoring device also includes a temperature compensation module; The temperature compensation module is used to compensate the pixel-level temperature measurement data of the power equipment under test obtained after deviation calibration, specifically as follows: Q ; Where Q is the compensated temperature measurement value. This represents the original temperature measurement weighting coefficient. Temperature data for the near-infrared auxiliary thermal detection unit. This is the distance compensation weighting coefficient. The temperature compensation component corresponds to the detection distance. The ambient air temperature compensation weighting coefficient. This is the temperature compensation component corresponding to the ambient air temperature. This is the emissivity compensation weighting coefficient. The temperature compensation component corresponding to the surface emissivity of the electrical equipment under test; weighting coefficient. Determined by blackbody calibration combined with least squares fitting.

[0011] In a further embodiment of the present invention, the weighting coefficient Determined through blackbody calibration combined with least squares fitting, specifically including: Using a standard blackbody source as the calibration reference, the correction formula Q is established. The quaternary linear regression model transforms the correction formula into a linear regression form and then into a matrix form. Perform least squares fitting to find the weight coefficients. To minimize the sum of squared residuals between the model predictions of all calibration samples and the true blackbody temperature of the standard blackbody source, the matrix inversion operation is used to obtain... .

[0012] A further feature of the present invention includes a multi-target temperature measurement calibration module, which has a built-in standard blackbody source communication interface to obtain the real-time temperature of the standard blackbody source. This temperature is used to correct the compensation coefficient A of the temperature compensation component corresponding to the detection distance, the compensation coefficient B of the temperature compensation component corresponding to the ambient air temperature, and the compensation coefficient C of the temperature compensation component corresponding to the surface emissivity of the power equipment under test.

[0013] Secondly, the present invention provides a monitoring method for the aforementioned multi-target photoelectric monitoring device for substations based on multispectral fusion, comprising: Acquire mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test; The mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data are subjected to feature-level fusion processing. The identification of the power equipment under test is completed and the area of ​​the power equipment under test is extracted. The pixel deviation of the mid- and far-infrared thermal imaging image is calibrated by combining the near-infrared temperature data. The correction model of the deviation calibration is based on the physical radiation transfer equation and neural network residual fitting to obtain pixel-level temperature measurement data of the power equipment under test.

[0014] A further feature of the present invention is that the pixel-level temperature measurement data of the power device under test is compared with a preset threshold in real time. If the data exceeds the preset threshold, it is marked as abnormal, and a temperature change curve of the power device under test is generated within a preset time period.

[0015] The present invention provides a multi-target photoelectric monitoring device and method for substations based on multispectral fusion, which has the following beneficial effects: The present invention integrates multi-source data from mid-infrared, far-infrared, and visible light sources through synchronous acquisition of multispectral images and feature-level fusion processing. This provides multi-dimensional equipment thermal status monitoring and testing data for substation control systems, improves the accuracy of temperature measurement of power equipment and the reliability of monitoring data. The near-infrared auxiliary channel can calibrate mid-infrared temperature drift, and visible light imaging can clearly present the details of equipment appearance, thus improving the overall availability of monitoring and testing data and the accuracy of fault location. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0017] Figure 1 This is a test flowchart of a multi-target photoelectric monitoring device for substations based on multispectral fusion, according to one embodiment of the present invention.

[0018] Figure 2 This is a flowchart of a multi-target photoelectric monitoring method for substations based on multispectral fusion, according to one embodiment of the present invention.

[0019] Figure 3 This is a flowchart illustrating the intelligent temperature compensation and temperature measurement process in one embodiment of the present invention.

[0020] Figure 4 This is a flowchart of intelligent inspection and target tracking management in one embodiment of the present invention.

[0021] Figure 5 This is a closed-loop diagram of anomaly monitoring and data interaction in one embodiment of the present invention. Detailed Implementation

[0022] This invention provides a multi-target photoelectric monitoring device and method for substations based on multispectral fusion. To make the objectives, technical solutions, and effects of this invention clearer and more explicit, the invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0023] In the implementation methods and claims, unless otherwise specified in the text, the terms "a," "an," "the," and "the" may also include plural forms. If the embodiments of the present invention involve descriptions of "first," "second," etc., such descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features.

[0024] It should be further understood that the term "comprising" as used in this specification means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when an element is referred to as "connected" or "coupled" to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements present. Furthermore, "connected" or "coupled" as used herein can include wireless connections or wireless coupling. The term "and / or" as used herein includes all or any unit and all combinations of one or more associated listed items.

[0025] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0026] Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0027] Traditional power equipment temperature measurement mainly relies on manual handheld temperature measuring devices for inspection. This has a limited measuring range and poor real-time data, making continuous 24 / 7 monitoring impossible. Manual inspections are also affected by outdoor environment and weather conditions, making it difficult to cover all equipment locations in a substation. Conventional fixed infrared temperature measurement systems use a single infrared lens to collect data; the measurement results are easily affected by factors such as detection distance, ambient temperature, and atmospheric transmittance, resulting in significant temperature drift. Regular on-site calibration is required, and the installation location must be close to the equipment being measured, making wide-area, long-distance deployment impossible.

[0028] Traditional substation temperature monitoring systems have significant limitations in functionality. Temperature measurement modes are limited to single-point and rectangular area measurement, failing to meet diverse needs such as global and linear temperature measurement. The number of target points that can be set within a single field of view is also limited, making it difficult to adapt to scenarios involving parallel monitoring of multiple devices in a substation. These systems lack target spatial positioning capabilities, cannot obtain the actual detection distance of the measured equipment, and temperature compensation relies solely on manually inputted empirical parameters, resulting in limited effectiveness. Abnormal temperature monitoring only provides numerical alarms, failing to accurately mark abnormal points or store abnormal images. Temperature data display is limited, and it does not support simultaneous comparison of multiple temperature curves, making it impossible for maintenance personnel to intuitively analyze the temperature change characteristics of a group of devices.

[0029] Existing temperature measurement devices have significant shortcomings in terms of intelligence and environmental adaptability. Most devices do not employ multispectral fusion technology or integrate near-infrared auxiliary temperature measurement channels, failing to improve measurement accuracy through multi-source data fusion. The devices themselves lack robust adaptive heat dissipation and high-level protection structures, resulting in insufficient long-term operational stability in the high-temperature, high-humidity, and dusty outdoor environments of substations. Furthermore, existing devices do not support automatic inspection and target tracking by CNC pan-tilt units, limiting the coverage of a single device and requiring large-scale deployment for comprehensive monitoring. They also lack integrated intelligent intrusion detection functions, making it impossible to simultaneously monitor for personnel and foreign object intrusion. Data storage capacity is limited, and remote communication protocol compatibility is poor, hindering efficient integration with centralized monitoring platforms and cloud systems, thus failing to meet the development needs of unmanned operation, real-time comprehensive monitoring, and remote intelligent operation and maintenance in smart substations.

[0030] This invention provides a preferred embodiment of a multi-target photoelectric monitoring device for substations based on multispectral fusion, comprising: a multispectral image acquisition module and a multispectral fusion temperature measurement processing module; the multispectral image acquisition module is used to acquire mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test; the multispectral fusion temperature measurement processing module includes a data fusion unit and a deviation calibration unit; the data fusion unit is used to process the mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data to identify the power equipment under test and extract the area of ​​the power equipment under test; the deviation calibration unit is used to perform deviation calibration on the pixels of the mid- and far-infrared thermal imaging images based on the near-infrared temperature data, using the physical radiative transfer equation and neural network residual fitting, to obtain pixel-level temperature measurement data.

[0031] Furthermore, a mid-to-far infrared detector can be selected to acquire thermal imaging images of the power equipment under test, a near-infrared auxiliary thermal detection unit can be used to measure the temperature data of the power equipment under test, and a visible light imaging unit can be used to capture image data of the power equipment under test; the frame synchronization signal of the mid-to-far infrared detector is used as a reference to trigger synchronous exposure of the infrared auxiliary thermal detection unit and the visible light imaging unit. In this embodiment of the invention, the operating wavelength range of the infrared auxiliary thermal detection unit is 0.78 to 2.5 μm, performing auxiliary detection and not undertaking the main temperature measurement function; the operating wavelength range of the mid-to-far infrared detector is 3 to 14 μm, serving as the main temperature measurement channel. Figure 1 As shown, by synchronously acquiring multispectral images and performing feature-level fusion processing, multi-source data from mid-infrared, near-infrared, and visible light sources are integrated. This provides substation control systems with multi-dimensional equipment thermal status monitoring and testing data, ultimately yielding pixel-level temperature measurement data. This improves the accuracy of power equipment temperature measurement and the reliability of monitoring data. The near-infrared auxiliary channel can calibrate mid-infrared temperature drift, and visible light imaging can clearly present equipment appearance details, comprehensively improving the usability of monitoring and testing data and the accuracy of fault location. The timing deviation of the three data streams (mid-infrared, near-infrared, and visible light) is less than 1 millisecond, providing a time alignment basis for pixel-level fusion calibration. In summary, the acquisition end integrates infrared, near-infrared, and visible light sensors to acquire thermal imaging, calibration data, and high-definition images of the equipment. The positioning end uses binocular ranging and angle calculation to obtain the spatial orientation of objects. The monitoring device performs time-series synchronization and feature fusion of these two types of heterogeneous data, thereby accurately locating target equipment in the substation against complex backgrounds and completing pixel-level temperature measurement point matching.

[0032] In this embodiment of the invention, the monitoring device further includes a spatial positioning module for acquiring the spatial location information of the power equipment under test. The spatial positioning module includes a binocular ranging unit and an angle calculation unit. The binocular ranging unit is used to acquire the spatial distance to the power equipment under test, and the angle calculation unit is used to acquire the azimuth angle data of the power equipment under test. Please refer to the appendix of the specification. Figure 1 .

[0033] In embodiments of the present invention, such as Figure 4As shown, the monitoring device also includes an inspection and target tracking module; the inspection and target tracking module includes a tracking calculation control unit and a CNC pan-tilt unit, with the multispectral image acquisition module fixed on the CNC pan-tilt unit; the tracking calculation control unit locates the power equipment under test based on the spatial location information obtained by the spatial positioning module, and issues a movement tracking command to the CNC pan-tilt unit; the inspection and target tracking module embeds a deep learning algorithm to automatically detect personnel and foreign object intrusion targets, and performs identification, capture, storage, and uploading operations on the intrusion targets. The substation multi-target photoelectric monitoring device loads and parses the pre-set monitoring point information in the background, controls the CNC pan-tilt unit to drive the sensor array to automatically cruise, and continuously tracks several power equipment under test. During the inspection process, the substation multi-target photoelectric monitoring device uses the built-in algorithm to lock and track key targets and judge the temperature difference in real time. If personnel or foreign object intrusion is detected, the system will automatically perform capture and uploading. This process realizes wide-area inspection of the entire substation area and uninterrupted monitoring of key targets. The pan-tilt unit has a horizontal rotation range of 0° to 360°, capable of continuous rotation, and a pitch range of -90° to +90°. The preset position accuracy is ≤0.05°. The pan-tilt unit's single field of view refers to the entire spatial range that the camera or detector can capture or detect at a fixed position and angle without rotating the pan-tilt unit. In other words, both the pitch (up / down) and horizontal (left / right) axes of the pan-tilt unit are fixed at a specific angle without rotation or scanning.

[0034] In this embodiment of the invention, visible light imaging data is captured by a visible light imaging unit. The monitoring device also includes a multi-mode temperature measurement execution module for setting the temperature measurement points within a single field of view of the visible light imaging unit on the CNC pan-tilt unit. The shape of the temperature measurement area formed by these points is unrestricted, and multiple temperature measurement areas can be measured in parallel. It also includes a multi-spectral fusion temperature measurement processing module that maps the coordinates of the visible light imaging unit's image data to the pixel coordinates of the thermal imaging image from the mid-to-far-infrared detector, enabling the definition of temperature measurement areas of arbitrary shapes. This meets the requirements for full-area and localized temperature measurement of large equipment. Figure 3 As shown, the monitoring system collects ambient temperature, radiation, atmospheric transmittance, target emissivity, and distance parameters calculated by the spatial positioning module in real time, and performs comprehensive compensation and correction on the raw data output by the mid- and far-infrared detectors. The corrected and accurate temperature data is sent to the execution module, which performs temperature measurement tasks in various modes such as single-point, linear, rectangular, or global as needed, and supports parallel processing of an unlimited number of points within a single field of view.

[0035] like Figure 5The diagram illustrates the backend management and early warning feedback process. The monitoring device compares the measured temperature data with preset thresholds in real time to identify high and low temperatures and temperature differences. Abnormal results are overlaid onto a high-definition screen via a visualization module, generating temperature change curves. Data is uploaded to the monitoring center and cloud platform via Ethernet or mobile network. Simultaneously, the alarm module sends alerts based on the alarm level, while the health management module monitors the device's own heat dissipation and status parameters to ensure the long-term stability of the entire operation and maintenance system.

[0036] In some embodiments, the infrared-assisted thermal detection unit employs a near-infrared camera, the visible light imaging unit employs a visible light camera, the binocular ranging unit has a baseline length of 80mm–120mm, an angle resolution better than 0.01°, and a positioning information refresh rate of not less than 10Hz. The visible light imaging unit includes a complementary metal-oxide-semiconductor image sensor (CMOS sensor), with both color and monochrome working illuminance indices, and is equipped with backlight compensation, strong light suppression, and 3D digital noise reduction processing functions. Simultaneously, the mid- and far-infrared detectors are configured with dual-gain working modes, with high-gain and low-gain modes corresponding to different temperature measurement ranges. The mid- and far-infrared detectors can employ uncooled silicon oxide infrared focal plane detectors, and the near-infrared auxiliary detection unit has a fixed working wavelength of 830nm, with independent channels acquiring auxiliary temperature measurement data to calibrate infrared temperature measurement drift.

[0037] In this embodiment of the invention, the step of processing the mid- and far-infrared thermal imaging image, the near-infrared temperature data, and the visible light imaging image data includes: A deep learning network is used to perform pixel-level registration between the visible light imaging data and the mid-to-far infrared thermal imaging image to obtain registered image data. The visible light imaging data and the mid-to-far infrared thermal imaging image are then fed into a trained semantic segmentation network to identify the power equipment under test, extract the area of ​​the power equipment under test, shield background interference, and perform deviation calibration on the pixels of the thermal imaging image of the mid-to-far infrared detector using temperature data from the near-infrared auxiliary thermal detection unit. The deviation calibration correction model is based on the physical radiative transfer equation and neural network residual fitting. Specifically, the 830nm near-infrared auxiliary thermal detection unit and the mid-to-far infrared detector simultaneously observe the same target. These two bands have different characteristics affected by atmospheric attenuation and detector temperature drift. By comparing these two data streams, the drift amount of temperature measurement by the mid-to-far infrared detector can be identified. First, the correction basis is established using the physical radiative transfer equation, and then the nonlinear deviation that the physical radiative transfer equation cannot cover is corrected using neural network residual fitting. Finally, the pixel-level temperature measurement data of the power equipment under test after calibration is output.

[0038] It also includes a temperature compensation module, which integrates distance compensation, ambient temperature compensation, and emissivity compensation mechanisms to correct the temperature measurement data. The compensation component is smoothed using Kalman filtering. It automatically switches between high-gain and low-gain modes based on the temperature range of the power equipment under test. The high-gain mode corresponds to temperature measurement under normal operating conditions, while the low-gain mode corresponds to temperature measurement under abnormal high-temperature conditions. Simultaneously, it combines load current data to perform temperature-load correlation analysis, filtering out non-fault-related temperature rise alarms caused by short-term overload.

[0039] The temperature compensation module is used to compensate the pixel-level temperature measurement data of the power equipment under test obtained after deviation calibration, specifically as follows: Q .

[0040] Where Q is the compensated temperature measurement value. This represents the original temperature measurement weighting coefficient. Temperature data for the near-infrared auxiliary thermal detection unit. This is the distance compensation weighting coefficient. The temperature compensation component corresponds to the detection distance. The ambient air temperature compensation weighting coefficient. This is the temperature compensation component corresponding to the ambient air temperature. This is the emissivity compensation weighting coefficient. The temperature compensation component corresponding to the surface emissivity of the electrical equipment under test; weighting coefficient. Determined through blackbody calibration combined with least squares fitting, and supports online dynamic updates via a multi-objective temperature measurement calibration module, specifically including: Using a standard blackbody source as the calibration reference, the correction formula Q is established. The four-variable linear regression model is used to transform the correction formula into a linear regression form and then into a matrix form; least squares fitting is performed to find the weight coefficients. To minimize the sum of squared residuals between the model predictions of all calibration samples and the true blackbody temperature of the standard blackbody source, the matrix inversion operation is used to obtain... It also includes a multi-target temperature measurement calibration module, which has a built-in standard blackbody source communication interface to acquire the real-time temperature of the standard blackbody source. This temperature is used to correct the compensation coefficients A (corresponding to the detection distance), B (corresponding to the ambient air temperature), and C (corresponding to the surface emissivity of the electrical equipment under test). If the deviation exceeds ±1℃, the correction algorithm is automatically invoked to update the compensation coefficients, eliminating the need for manual intervention. Near-infrared auxiliary calibration provides real-time pixel-level drift suppression at the front end, while the blackbody calibration module performs periodic system coefficient recalibration at the back end. Together, they ensure long-term stable temperature measurement accuracy.

[0041] Temperature compensation component corresponding to detection distance ,in To determine the straight-line distance to the target power equipment, a real-time measurement based on the near-infrared laser time-of-flight principle is employed: The near-infrared auxiliary thermal detection unit emits an 830nm near-infrared laser and receives the echo, processing the time difference Δt between emission and echo. Real-time acquisition of ambient temperature, humidity, and air pressure parameters is used, and atmospheric refractive index correction is applied. This is combined with the coaxiality error compensation coefficient between the gimbal axis and the laser emission axis calibrated during installation, as well as the real-time calculated atmospheric turbulence disturbance correction coefficient. Through multi-echo effective signal filtering logic, invalid echoes from nearby obstructions such as tree branches, dust covers, and spider webs are automatically filtered out. Finally, the straight-line distance between the target power equipment and the temperature-measuring radar is calculated and output. .

[0042] Where c is the current ambient light speed, η is the multi-echo effective signal filtering coefficient, k is the atmospheric refractive index correction coefficient, γ is the coaxiality error compensation coefficient, and ε is the atmospheric turbulence disturbance correction coefficient.

[0043] The distance-temperature compensation coefficient A is determined through calibration using a standard blackbody source. Infrared thermometry is performed on a standard blackbody with a known true temperature at different distances (e.g., 10m, 20m, 30m, 50m, 65m, etc.); the deviation between the original temperature measurement value and the true temperature of the blackbody is recorded for each distance; and the temperature attenuation corresponding to each meter increase in distance is obtained through least-squares fitting, i.e., coefficient A.

[0044] Temperature compensation component corresponding to ambient air temperature The current ambient temperature E is collected in real time by the device's built-in environmental sensors (temperature, humidity, and air pressure sensors), providing the current ambient temperature reference data for temperature calibration. The calibration reference ambient temperature E0 is the reference temperature set during the system's factory calibration, typically 25°C.

[0045] The ambient temperature compensation coefficient B is determined through blackbody calibration: Under different ambient temperature conditions (such as low temperature, normal temperature, and high temperature environmental chambers), infrared temperature measurement is performed on a standard blackbody with a known true temperature; the deviation of the original temperature measurement value under each ambient temperature is recorded; through least squares fitting, the temperature deviation corresponding to each 1℃ deviation of the ambient temperature from the reference temperature is obtained, i.e., coefficient B.

[0046] Temperature compensation component of surface emissivity of the electrical equipment under test The calibration reference emissivity F is the emissivity of a standard blackbody, typically taken as 0.95, which is the reference emissivity standard used in calibration experiments. The actual emissivity of the target being measured... The multispectral fusion recognition unit automatically determines: the outline, texture, and identification features of power equipment are extracted from visible light images; the type, material, rated parameters, and operation and maintenance history of the target equipment are identified by matching the pre-stored power equipment ledger database; the actual emissivity parameters of the corresponding target are output by matching the pre-stored equipment material emissivity database; and the emissivity of all targets within the field of view is automatically obtained without the need for manual input of emissivity.

[0047] Emissivity compensation coefficient The method employed blackbody calibration: temperature measurement experiments were conducted using standard targets with different known emissivity (or by attaching materials with different emissivity in front of the blackbody); the deviation of the original temperature measurement values ​​under each set of emissivity conditions was recorded; and the temperature compensation amount corresponding to the emissivity deviation, i.e., the coefficient, was obtained through least-squares fitting. .

[0048] In this embodiment of the invention, an adaptive heat dissipation and protection module is also included. This module comprises a plate-fin heat sink, a sealed protective cavity, an internal temperature monitoring unit, and a speed-regulating fan array. It automatically adjusts the heat dissipation efficiency based on the internal operating temperature of the device. When the internal cavity temperature exceeds 40°C, the fan automatically starts; when it exceeds 55°C, it operates at full speed and reports an alarm. The sealed protective cavity is filled with desiccant to suppress condensation. The monitoring device of this invention is housed within the sealed protective cavity. A temperature data visualization module is also included, which overlays temperature data onto a visible light screen in real time, generating multiple temperature change curves. It supports simultaneous display of real-time trends and historical comparisons of no fewer than six temperature curves on the same interface, and allows for playback at different time scales. It also includes a data storage and playback module for storing historical temperature data, abnormal image records, and automatic inspection logs. It supports querying temperature curves by time range, with a storage medium capacity of at least 1TB, supporting cyclic overwrite storage. Temperature data can be exported as CSV and standard infrared thermal image formats. Inspection logs include complete fields for CNC pan-tilt-zoom angle, temperature measurement points, and environmental parameters, and are synchronously uploaded to the centralized monitoring center and private cloud platform via wired and wireless networks. The uploaded data packets use JSON format, including timestamps, device tags, and checksums, and support breakpoint resumption and data deduplication. A digital twin model of the device is constructed to analyze the deviation between actual and expected parameters; if the deviation exceeds the limit, a preventative maintenance reminder is triggered. Meanwhile, the monitoring device of this invention also includes a communication transmission module, which completes data uploading and remote command reception via adaptive Ethernet and mobile communication networks, and interfaces with the centralized monitoring center and private cloud platform. The communication protocols simultaneously support Real-Time Streaming Protocol, Open Network Video Interface Forum, Technical requirements for information transmission, switch and control in video surveillance networking system for public security, and Communication networks and systems in substations. A surge protector is configured for lightning protection. In addition, it also includes an alarm output module, which generates graded alarm signals according to the temperature anomaly level, and completes the synchronous push of alarm information from local alarm prompts and remote monitoring center. The alarm levels are divided into three levels: early warning, serious alarm, and critical alarm. Each level of alarm signal is output in parallel through dry contacts, audible and visual alarms, and network messages. The alarm information includes the device name, abnormal temperature value, alarm level, and abnormal screen screenshot.The system synchronizes local alarm notifications with alarm information from the remote monitoring center. Alarm information is pushed to the mobile maintenance app via the MQTT protocol, allowing maintenance personnel to remotely confirm, suppress, or escalate alarms. It also includes preset automatic inspection steps, loading pre-defined substation monitoring point information and controlling a CNC pan-tilt unit to complete automatic cyclic inspections of multiple areas. The CNC pan-tilt unit's movement uses trapezoidal speed curve control, and the dwell time after reaching its destination is settable, during which data acquisition and fusion processing of all temperature measurement points are completed. This monitoring device also includes an equipment health management module, which collects the device's own operating parameters and performs intelligent assessment of the device's operating status. These parameters include the internal temperature of the sealed protective cavity, the focal temperature of the silicon oxide uncooled infrared focal plane detector, the CNC pan-tilt unit motor current, the communication link bit error rate, and the remaining lifespan of the storage medium. The module identifies device faults and outputs early warning information. Figure 2 As shown, the system first acquires the raw data and spatial location of the power equipment under test simultaneously through a multispectral image acquisition module and a spatial positioning module. Then, it enters the core processing stage, performing data fusion identification and multi-parameter temperature compensation correction. Next, the monitoring device performs intelligent inspections according to a preset path and executes temperature measurements in various modes. Finally, through anomaly monitoring, data visualization, and cloud communication, a closed-loop management system is achieved, enabling tiered alarm push notifications and device health self-evaluation.

[0049] This invention employs a multi-dimensional intelligent temperature compensation mechanism, combining parameters such as detection distance, ambient temperature, and equipment emissivity to correct temperature measurement data. It eliminates the need for regular manual on-site calibration, maintaining stable accuracy of long-term monitoring and testing data. This ensures the continuous reliability of equipment temperature status data input to the control system, providing a high-confidence data foundation for control system status assessment and anomaly diagnosis decisions. It offers multiple temperature measurement execution modes, including global, single-point, linear, and rectangular area measurement. It supports unlimited temperature measurement point settings within a single field of view and can simultaneously perform parallel temperature measurements in multiple areas, fully adapting to the differentiated monitoring and testing needs of substation control systems for different types and layouts of power equipment. Equipped with a CNC pan-tilt unit and intelligent inspection module, this invention enables automatic cyclic inspection of preset points and continuous multi-target tracking. A single device can complete continuous monitoring and testing covering the entire substation, reducing the number of devices deployed and system integration complexity. Embedded deep learning algorithms can simultaneously detect personnel and foreign object intrusion, expanding the comprehensive capabilities of substation control systems for full-area safety monitoring and equipment status diagnosis.

[0050] Meanwhile, the present invention also has complete abnormal temperature monitoring and visualization functions, which can accurately mark abnormal points and store abnormal images, support the synchronous comparison and display of multiple temperature curves, and overlay temperature data onto the visible light screen in real time, providing timely and accurate judgment input for the control system to trigger protection linkage and other automatic response strategies, and shortening the time delay from abnormality detection to control response.

[0051] In some embodiments of this invention, an adaptive heat dissipation protection module and an equipment health management module are configured to adapt to the complex outdoor operating environment of substations, ensuring the long-term operational stability of the monitoring and testing device itself, preventing the control system from losing its monitoring data source due to device failure, and simultaneously monitoring the device's own operating status in real time, promptly identifying faults and outputting early warnings. A large-capacity data storage and multi-protocol compatible communication transmission module is adopted to store historical temperature measurement data and inspection records for a long time, efficiently connecting to the substation integrated automation system, centralized monitoring center, and cloud platform, establishing a data link from the field monitoring device to the control system, and supporting the control system's trend analysis and status prediction based on historical data. A multi-target temperature measurement calibration module is set up to dynamically calibrate temperature measurement accuracy based on operating data, continuously eliminating temperature drift caused by long-term device operation, maintaining the stability and consistency of temperature measurement results, ensuring the comparability and continuity of equipment status data acquired by the control system at different operating stages, and further improving the overall efficiency of multi-target monitoring and testing in substations.

[0052] This invention also provides a multi-target photoelectric monitoring method for substations based on multispectral fusion, comprising: acquiring mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test; performing feature-level fusion processing on the mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data; identifying the power equipment under test and extracting the region of the power equipment under test; and performing pixel deviation calibration on the mid- and far-infrared thermal imaging images in conjunction with the near-infrared temperature data, wherein the deviation calibration correction model is based on the physical radiative transfer equation and neural network residual fitting to obtain pixel-level temperature measurement data of the power equipment under test.

[0053] Furthermore, the temperature data of the measured power equipment can be compared with a preset temperature threshold. The temperature threshold includes three levels of criteria: absolute temperature threshold, temperature rise rate threshold, and phase-to-phase temperature difference threshold. If any criterion is triggered, the measured power equipment is determined to be in abnormal temperature condition, and the abnormal state of the power equipment is marked, and the abnormal point is marked and the abnormal image is stored.

[0054] In this embodiment of the invention, the method further includes: comparing the pixel-level temperature measurement data of the power device under test with a preset threshold in real time; if the data exceeds the preset threshold, it is marked as abnormal; and generating a temperature change curve of the power device under test within a preset time period.

[0055] This invention provides an embodiment of a multi-target photoelectric monitoring device for substations based on multispectral fusion, applied to multi-target photoelectric temperature measurement in 110kV substations. This embodiment is applied to the outdoor equipment area of ​​a 110kV substation, and the monitoring objects include transformers, circuit breakers, disconnect switches, instrument transformers, and busbar equipment. The device is fixedly installed on the outer support of the equipment area, without needing to be close to the equipment being measured, and achieves full-area multi-target temperature measurement within a range of 10 meters to 65 meters. The device fully operates all functional modules and the entire temperature measurement method process.

[0056] In this embodiment, the multispectral image acquisition module operates continuously. The uncooled silicon oxide infrared focal plane detector acquires mid- and far-infrared thermal imaging data at a resolution of 640×512. It automatically switches gain modes based on device temperature: high gain mode is used for normally operating equipment, covering a temperature range of 20℃ to 150℃; low gain mode is used for equipment with potential high-temperature anomalies, covering a temperature range of 0℃ to 550℃, with thermal sensitivity stabilized below 60mK. The pixel size is set to 17μm. The near-infrared auxiliary detection unit acquires auxiliary temperature measurement data at a fixed wavelength of 830nm, calibrating infrared temperature drift in real time. The visible light imaging unit uses a 2-megapixel CMOS sensor, outputting high-definition images at a resolution of 1920×1080. Illuminance is maintained at 0.005Lux in color mode and 0.001Lux in black and white mode. Through backlight compensation, strong light suppression, and 3D digital noise reduction, starlight-level night vision imaging is achieved, clearly presenting the device's appearance and connection point details. The spatial positioning module, through a binocular ranging unit and an angle calculation unit, calculates the detection distance and horizontal and vertical azimuth angles of each measured power device in real time, generating three-dimensional spatial positioning information. This positioning data is then synchronously transmitted to the multispectral fusion temperature processing module, providing spatial parameter support for subsequent temperature measurement and compensation. The multispectral fusion temperature processing module receives data from mid-infrared, far-infrared, and visible light sources, performs feature-level fusion processing, and, based on a deep learning model, intelligently identifies various types of power equipment such as transformers and circuit breakers. It accurately locates pixel-level temperature measurement data across the entire area and at specific points, eliminating invalid temperature measurement data from equipment accessories and environmental interference, and extracting valid equipment temperature measurement information.

[0057] The intelligent temperature compensation module collects ambient air temperature, ambient radiation temperature, atmospheric transmittance, and equipment surface emissivity parameters in real time. Combined with the detection distance output by the spatial positioning module, it activates a triple mechanism of distance compensation, ambient temperature compensation, and emissivity compensation to comprehensively correct the original temperature measurement data and generate accurate temperature values. The correction process requires no manual intervention and continuously maintains stable temperature measurement accuracy.

[0058] An adaptive heat dissipation and protection module operates synchronously. The plate-fin radiator automatically adjusts its heat dissipation efficiency based on the internal operating temperature of the device. A sealed protective cavity provides dust and water protection, and an internal temperature monitoring unit monitors the cavity temperature in real time, ensuring long-term stable operation of the device in complex outdoor environments. A multi-target temperature measurement calibration module establishes temperature measurement benchmark parameters at different detection distances and ambient temperatures. It combines multispectral fusion data to perform dynamic calibration of temperature measurement accuracy, eliminating temperature drift caused by long-term operation and maintaining consistency of temperature measurement results.

[0059] In some embodiments, the intelligent inspection and target tracking module loads preset monitoring point information and achieves automatic cruise switching through the CNC pan-tilt control unit. An unlimited number of temperature measurement points can be set within the field of view of a single preset point, allowing a single device to complete full-area coverage inspection of the substation. The module embeds a deep learning algorithm to automatically detect personnel and foreign object intrusion targets, completing the identification, capture, storage, and uploading of multiple intrusion targets to achieve comprehensive safety monitoring. The multi-mode temperature measurement execution module provides multiple modes, including global temperature measurement, single-point temperature measurement, linear temperature measurement, and rectangular area temperature measurement. It switches temperature measurement modes according to maintenance needs, simultaneously completing parallel temperature measurement of multiple areas to meet the temperature monitoring requirements of different devices. The abnormal temperature monitoring and marking module compares preset temperature thresholds with area temperature difference thresholds to identify high temperature, low temperature, and temperature difference abnormalities, accurately marking abnormal points and storing abnormal images in real time. The temperature data visualization module overlays temperature values ​​onto a visible light screen in real time, generating multiple temperature change curves and simultaneously comparing the temperature characteristics of multiple devices, intuitively displaying the temperature differences between the device group. The equipment health management module collects the device's operating parameters in real time, performs intelligent assessment of its operating status, identifies device faults, and outputs early warning information.

[0060] This embodiment implements a multi-target photoelectric monitoring method for substations based on multispectral fusion, sequentially completing the following steps: synchronous multispectral data acquisition, spatial three-dimensional positioning, multispectral feature fusion and identification, multi-parameter temperature compensation, multi-mode parallel temperature measurement, automatic inspection of preset points, temperature anomaly monitoring and marking, temperature data visualization, data storage and remote transmission, hierarchical alarm information push, and device operation health monitoring. Multispectral feature fusion and identification extracts the device outline using visible light and combines it with infrared temperature and near-infrared calibration data to complete multi-target matching, unaffected by outdoor environments. Multi-parameter temperature compensation continuously updates parameters and dynamically adjusts them, automatically switching temperature measurement gain modes to achieve accurate temperature measurement across the entire area.

[0061] Table 1: Comparison of core indicators of the device of the present invention with traditional infrared temperature measurement systems.

[0062]

[0063] Table 1 shows the data from the continuous operation test results of the 110kV substation in this embodiment. Traditional infrared temperature measurement systems have short effective distances, limited temperature measurement modes, a limited number of temperature measurement points per field of view, rely on manual calibration, and can only output numerical alarms for anomaly monitoring. The device of this invention, relying on multispectral fusion and spatial positioning, significantly improves the effective temperature measurement distance, supports multiple temperature measurement modes and unlimited temperature measurement points, automatically completes temperature measurement calibration and anomaly marker storage, and comprehensively improves the coverage, flexibility, and intelligence level of multi-target temperature measurement in substations, adapting to the monitoring needs of all equipment in a 110kV substation.

[0064] This invention also provides an embodiment of a multi-target photoelectric monitoring device for substations based on multispectral fusion applied to wide-area photoelectric temperature measurement in a 500kV converter station. This embodiment is applied to the outdoor wide-area equipment area of ​​a 500kV converter station, and the monitoring objects include converter transformers, smoothing reactors, converter valve groups, high-voltage switchgear, and grounding devices. The equipment is widely distributed and spaced apart. The device is installed in the central monitoring tower of the converter station, realizing accurate temperature measurement of multiple targets across the entire area, and fully operating all modules of the device and the entire temperature measurement method. In this embodiment, the multispectral image acquisition module maintains high-load operation, the infrared detector stably outputs dual-gain temperature measurement data, the near-infrared auxiliary detection unit continuously calibrates temperature drift, and the visible light imaging unit outputs high-definition images after noise reduction and enhancement, clearly presenting the detailed structure of large equipment in the converter station, meeting the needs of detailed equipment identification under wide-area monitoring. The spatial positioning module accurately calculates the detection distance and azimuth angle of distant equipment for the widely distributed equipment in the converter station, generating three-dimensional positioning information, providing accurate spatial parameters for wide-area temperature measurement, and solving the problem of inaccurate positioning for long-distance temperature measurement. The multispectral fusion temperature measurement processing module completes the feature fusion of multi-source data, identifies large-scale power equipment unique to the converter station based on a deep learning model, accurately extracts pixel-level temperature measurement data, eliminates the influence of electromagnetic interference and environmental noise on the temperature measurement, and ensures the validity of the temperature measurement data.

[0065] The temperature compensation module, designed for the wide-area, long-distance temperature measurement characteristics of converter stations, strengthens distance and environmental compensation. Combining real-time collected environmental parameters with the detection distance, it accurately corrects the original temperature measurement data, ensuring stable accuracy in long-distance temperature measurement. This embodiment employs a photoelectric monitoring method, ensuring stable operation throughout the entire process. Multispectral feature fusion identification resists interference from the complex environment of the converter station. Multi-parameter temperature compensation adapts to wide-area, long-distance temperature measurement, automatically switching the temperature measurement gain mode to ensure accurate temperature measurement of all equipment in the converter station. This invention's device maintains high temperature matching degree, anomaly detection rate, and operational stability for various large-scale power equipment in converter stations. Multispectral fusion technology and multi-dimensional compensation mechanisms adapt to the wide-area, long-distance, and multi-device temperature measurement characteristics of converter stations. Spatial positioning and automatic inspection functions cover the entire monitoring area. The device's protective structure and intelligent calibration mechanism ensure long-term stable operation, meeting the high-precision, wide-area, and intelligent temperature monitoring needs of 500kV converter stations and improving the intelligent level of converter station equipment operation and maintenance.

[0066] In summary, current substation control systems generally lack multi-dimensional online monitoring and testing methods for the thermal status of power equipment, and cannot incorporate equipment temperature status data into the control system's status assessment and decision-making model in real time. The multi-spectral fusion-based substation multi-target photoelectric monitoring device and method provided by this invention enables online continuous monitoring and anomaly testing and diagnosis of the operating temperature status of power equipment, improves monitoring coverage, enriches monitoring and testing methods, realizes automatic calibration and full-domain intelligent inspection, strengthens anomaly identification and data visualization capabilities, enhances environmental adaptability and remote operation and maintenance levels, provides reliable equipment operating status monitoring data support for substation control systems, and adapts to the intelligent monitoring and testing needs of substations.

[0067] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A multi-target photoelectric monitoring device for substations based on multispectral fusion, characterized in that, include: Multispectral image acquisition module and multispectral fusion temperature measurement and processing module; The multispectral image acquisition module is used to acquire mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test. The multispectral fusion temperature measurement processing module includes a data fusion unit and a deviation calibration unit; The data fusion unit is used to process the mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data to identify the power equipment under test and extract the area of ​​the power equipment under test. The deviation calibration unit is used to perform deviation calibration on the pixels of the mid- and far-infrared thermal imaging image based on the near-infrared temperature data, using the physical radiation transfer equation and neural network residual fitting, to obtain pixel-level temperature measurement data.

2. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 1, characterized in that, The monitoring device also includes a spatial positioning module for acquiring the spatial location information of the power equipment under test. The spatial positioning module includes a binocular ranging unit and an angle calculation unit. The binocular ranging unit is used to acquire the spatial distance of the power equipment under test, and the angle calculation unit is used to acquire the azimuth angle data of the power equipment under test.

3. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 2, characterized in that, The monitoring device also includes an inspection and target tracking module; the inspection and target tracking module includes a tracking calculation control unit and a CNC pan-tilt unit, and the multispectral image acquisition module is fixed on the CNC pan-tilt unit; the tracking calculation control unit locates the power equipment under test based on the spatial location information of the power equipment under test obtained by the spatial positioning module, and issues a movement tracking command to the CNC pan-tilt unit; the inspection and target tracking module embeds a deep learning algorithm to automatically detect personnel and foreign object intrusion targets, and performs identification, capture, storage and uploading operations on intrusion targets.

4. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 3, characterized in that, The monitoring device also includes a multi-mode temperature measurement execution module, which is used to set the temperature measurement points within a single field of view of the visible light imaging unit on the CNC pan-tilt unit. The shape of the temperature measurement area formed by the temperature measurement points is unlimited, and multiple temperature measurement areas can be measured in parallel.

5. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 1, characterized in that, The step of processing the mid- and far-infrared thermal imaging image, the near-infrared temperature data, and the visible light imaging image data includes: The visible light imaging data and the mid- and far-infrared thermal imaging image are registered pixel-level using a deep learning network to obtain registered image data. The visible light imaging data and the mid- and far-infrared thermal imaging image are then fed into a trained semantic segmentation network to complete the identification of the power equipment under test and extract the region of the power equipment under test.

6. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 1, characterized in that, The monitoring device also includes a temperature compensation module; The temperature compensation module is used to compensate the pixel-level temperature measurement data of the power equipment under test obtained after deviation calibration, specifically as follows: Q ； Where Q is the compensated temperature measurement value. This represents the original temperature measurement weighting coefficient. Temperature data for the near-infrared auxiliary thermal detection unit. This is the distance compensation weighting coefficient. The temperature compensation component corresponds to the detection distance. The ambient air temperature compensation weighting coefficient. This is the temperature compensation component corresponding to the ambient air temperature. This is the emissivity compensation weighting coefficient. The temperature compensation component corresponding to the surface emissivity of the electrical equipment under test; weighting coefficient. Determined by blackbody calibration combined with least squares fitting.

7. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 6, characterized in that, The weighting coefficient Determined through blackbody calibration combined with least squares fitting, specifically including: Using a standard blackbody source as the calibration reference, the correction formula Q is established. The quaternary linear regression model transforms the correction formula into a linear regression form and then into a matrix form. Perform least squares fitting to find the weight coefficients. To minimize the sum of squared residuals between the model predictions of all calibration samples and the true blackbody temperature of the standard blackbody source, the matrix inversion operation is used to obtain... .

8. The substation multi-target photoelectric monitoring device based on multispectral fusion according to claim 1, characterized in that, It also includes a multi-target temperature measurement calibration module, which has a built-in standard blackbody source communication interface to obtain the real-time temperature of the standard blackbody source, and is used to correct the compensation coefficient A of the temperature compensation component corresponding to the detection distance, the compensation coefficient B of the temperature compensation component corresponding to the ambient air temperature, and the compensation coefficient C of the temperature compensation component corresponding to the surface emissivity of the power equipment under test.

9. A monitoring method applied to the multi-target photoelectric monitoring device for substations based on multispectral fusion as described in any one of claims 1-8, characterized in that, include: Acquire mid- and far-infrared thermal imaging images, near-infrared temperature data, and visible light imaging data of the power equipment under test; The mid- and far-infrared thermal imaging images, the near-infrared temperature data, and the visible light imaging data are subjected to feature-level fusion processing. The identification of the power equipment under test is completed and the area of ​​the power equipment under test is extracted. The pixel deviation of the mid- and far-infrared thermal imaging image is calibrated by combining the near-infrared temperature data. The correction model of the deviation calibration is based on the physical radiation transfer equation and neural network residual fitting to obtain pixel-level temperature measurement data of the power equipment under test.

10. The monitoring method according to claim 9, characterized in that, The pixel-level temperature measurement data of the power device under test is compared with a preset threshold in real time. If the value exceeds the preset threshold, it is marked as abnormal, and a temperature change curve of the power device under test is generated within a preset time period.

Images