Power distribution cabinet monitoring device and monitoring method based on infrared-visible light double-source cooperation

Abstract

The invention discloses a power distribution cabinet monitoring device and monitoring method based on infrared-visible light double-source cooperation, and relates to the technical field of power equipment monitoring. A double-source cooperation monitoring module is arranged to synchronously collect visible light image data and infrared temperature data of a power distribution cabinet and complete accurate calibration; the fault identification module integrates environment temperature and humidity data and typical fault samples to perform fusion analysis on the calibrated double-source data so as to realize fault identification and generate a disposal instruction, and then the environment regulation and control module responds to the instruction to execute a corresponding disposal action; a full-link monitoring operation and maintenance system of double-source data acquisition calibration-multi-dimensional fusion fault recognition-precise linkage disposal is formed, and the technical problem that the recognition accuracy and timeliness of potential risks of equipment are insufficient due to the defect of incomplete sensing caused by single sensing and data splitting in an existing power distribution cabinet monitoring operation and maintenance technology is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment monitoring, and in particular to a power distribution cabinet monitoring device and method based on infrared-visible light dual-source cooperation. BACKGROUND

[0002] With the continuous evolution of the power system towards intelligence and high reliability, the stable operation of the power distribution cabinet as the terminal core equipment of the power distribution network is directly related to the continuity and safety of industrial production and people's livelihood electricity, especially in a closed and high-load operating environment, potential risks such as equipment heating, component aging, and environmental factors occur frequently, and the industry has an increasingly urgent demand for intelligent monitoring, accurate early warning, and efficient disposal of power distribution cabinets.

[0003] At present, power distribution cabinet monitoring has gradually transformed from traditional manual inspection to automatic monitoring, and the application of various sensing technologies and control components has reduced the dependence on manual work to some extent, but the existing technology still has key bottlenecks: most solutions rely on a single sensing mode for monitoring, which is easily affected by cabinet structure shielding, forming a monitoring blind area, even though some solutions introduce multi-source sensing data, there is still a lack of effective and accurate cooperation mechanism, resulting in different dimensions of data being isolated from each other, making it difficult to achieve comprehensive perception of equipment status; at the same time, the existing technology relies on a single data dimension for fault identification, does not fully combine the actual operating environment of the equipment with typical fault characteristics, and the fault identification and subsequent disposal are independent of each other, lacking a direct linkage closed-loop design, often resulting in identification lag and single disposal, which cannot quickly capture the core characteristics of potential equipment risks, nor can it take timely and accurate measures to address the risks, these problems collectively restrict the comprehensiveness, accuracy, and efficiency of power distribution cabinet operation and maintenance, making it difficult to meet the high-reliability operation requirements in complex scenarios. SUMMARY

[0004] The present application provides a power distribution cabinet monitoring device and method based on infrared-visible light dual-source cooperation, which solves the problem of incomplete perception caused by single sensing and data fragmentation in existing power distribution cabinet monitoring and maintenance technology, and improves the accuracy and timeliness of identifying potential equipment risks.

[0005] The present application provides a power distribution cabinet monitoring device and method based on infrared-visible light dual-source cooperation, which solves the problem of incomplete perception caused by single sensing and data fragmentation in existing power distribution cabinet monitoring and maintenance technology, and improves the accuracy and timeliness of identifying potential equipment risks.

[0006] The dual-source cooperative monitoring module is configured to collect visible light image data and external infrared temperature measurement data of the power distribution cabinet to be detected and perform dual-source data calibration to obtain a calibrated dual-source graph.

[0007] The fault identification module is configured to obtain environmental temperature and humidity data and typical fault samples of the power distribution cabinet, perform fault identification in combination with the calibrated dual-source graph, and generate disposal instructions according to the fault identification results.

[0008] an environmental regulation module configured to perform an associated treatment action according to the treatment instruction.

[0009] Optionally, the dual-source cooperative monitoring module comprises:

[0010] a camera group configured to collect visible light image data of the power distribution cabinet to be detected;

[0011] an infrared temperature measurement array configured to collect external infrared temperature measurement data of the power distribution cabinet to be detected;

[0012] a dual-source data synchronous calibration unit configured to perform coordinate matching on the visible light image data and the external infrared temperature measurement data to obtain a calibrated dual-source atlas.

[0013] Optionally, the dual-source cooperative monitoring module further comprises a mounting bracket;

[0014] the camera group and the infrared temperature measurement array are both mounted on the mounting bracket;

[0015] the mounting bracket is configured to adjust the monitoring angles of the camera group and the infrared temperature measurement array until a preset standard collection condition is met;

[0016] the preset standard collection condition is specifically that the monitoring coverage of all key monitoring areas in the power distribution cabinet to be detected is greater than a preset standard coverage threshold, and the field of view overlap degree between the camera group and the infrared temperature measurement array meets a preset standard overlap threshold.

[0017] Optionally, the coordinate matching process is specifically as follows:

[0018] a calibration reference point in the power distribution cabinet to be detected is determined, and the calibration reference point is specifically a fixed mark pre-marked in the power distribution cabinet to be detected;

[0019] the pixel coordinates of the visible light image data and the temperature measurement point coordinates of the external infrared temperature measurement data are matched with reference to the calibration reference point to obtain a plurality of pixel-temperature atlases;

[0020] the plurality of pixel-temperature atlases are integrated to obtain a calibrated dual-source atlas.

[0021] Optionally, the fault identification module comprises:

[0022] an edge computing chip configured to receive environmental temperature and humidity data of the power distribution cabinet to be detected, and to perform noise reduction on the calibrated dual-source atlas to obtain a dual-source feature matrix, which is specifically a two-dimensional standardized matrix composed of pixel visual features and temperature features, and to perform normalization on the environmental temperature and humidity data to obtain environmental feature parameters;

[0023] a feature database configured to store typical fault samples of the power distribution cabinet;

[0024] a fault identification model configured to perform fault identification by using the dual-source feature matrix, the environmental feature parameter, and the typical fault samples of the power distribution cabinet, and generate an abnormal confidence level;

[0025] The edge computing chip is further configured to generate an associated treatment instruction according to a preset fault level interval to which the abnormal confidence level belongs.

[0026] Optionally, the process of the fault identification is specifically as follows:

[0027] performing visual and temperature dual-channel feature extraction on the dual-source feature matrix to obtain visual feature parameters and temperature feature parameters;

[0028] constructing a visual feature vector by using the visual feature parameters;

[0029] constructing a temperature-environment feature vector by using the temperature feature parameters and the environmental feature parameter;

[0030] performing weighted fusion on the visual feature vector and the temperature-environment feature vector to obtain a fusion feature vector;

[0031] calculating a cosine similarity between the fusion feature vector and the typical fault samples of the power distribution cabinet to determine an abnormal confidence level.

[0032] Optionally, the environmental regulation module comprises a treatment priority decision unit, a circuit breaker linkage interface, an audible and visual alarm, an environmental regulation component, and a sensor device.

[0033] The treatment priority decision unit is configured to match and execute a treatment action corresponding to an instruction type according to the instruction type of the treatment instruction.

[0034] The circuit breaker linkage interface is configured to perform fault loop isolation.

[0035] The audible and visual alarm is configured to perform audible and visual alarm.

[0036] The environmental regulation component comprises an intelligent fan and a dehumidifier.

[0037] The intelligent fan is configured to perform heat dissipation when the fault type is an overheating fault.

[0038] The dehumidifier is configured to perform dehumidification when the fault type is a condensation fault.

[0039] The sensor device comprises a temperature and humidity sensor and a condensation sensor.

[0040] The temperature and humidity sensor is used for collecting the environmental temperature and humidity data in the power distribution cabinet to be detected in real time and transmitting to the fault identification module.

[0041] The condensation sensor is used for collecting condensation data in the power distribution cabinet to be detected.

[0042] Optionally, the process of matching the treatment action corresponding to the instruction type and executing according to the instruction type of the treatment instruction is specifically as follows:

[0043] When the treatment instruction is an emergency fault treatment instruction, the fault loop isolation is executed through the circuit breaker linkage interface, and the audible and visual alarm is started;

[0044] When the treatment instruction is a serious fault treatment instruction, the corresponding environmental regulation component is started according to the fault type, and the audible and visual alarm is started;

[0045] When the treatment instruction is an environmental anomaly treatment instruction, the environmental regulation component is started;

[0046] When the treatment instruction is a slight fault treatment instruction, the audible and visual alarm is started, and the sensor device is controlled to monitor the environmental state of the power distribution cabinet to be detected.

[0047] Optionally, the low-power operation and maintenance module is further included.

[0048] The low-power operation and maintenance module includes a local storage module, a wireless communication module and a low-power management unit.

[0049] The local storage module is connected with the dual-source cooperative monitoring module and the fault identification module respectively, and is used for data classification storage.

[0050] The wireless communication module is connected with the local storage module, and is used for uploading the classified data stored in the local storage module to a remote operation and maintenance platform.

[0051] The low-power management unit is used for switching the monitoring device to a low-power operation and maintenance mode when the monitoring device is not in a monitoring stage.

[0052] The second aspect of the present application provides a monitoring method applied to the power distribution cabinet monitoring device based on infrared-visible light dual-source cooperation.

[0053] Collecting visible light image data and external red temperature measurement data of the power distribution cabinet to be detected and performing dual-source data calibration to obtain a calibrated dual-source graph;

[0054] Obtaining environmental temperature and humidity data and power distribution cabinet typical fault samples, combining the calibrated dual-source graph to perform fault identification, and generating a treatment instruction according to the fault identification result.

[0055] performing an associated treatment action according to the treatment instruction.

[0056] From the above technical solutions, the present application has the following advantages:

[0057] The present application provides a kind of based on infrared-visible light double source cooperation's switchgear monitoring device and monitoring method, by setting up double source cooperative monitoring module synchronous acquisition switchgear visible light image data and infrared temperature data and complete accurate calibration, with fault identification module integration environmental temperature and humidity data and typical fault sample are fused to the double source data after calibration It is analyzed to realize fault identification and generate treatment instruction, again by environmental control module response instruction executes corresponding treatment action, forms double source data acquisition calibration-multidimensional fusion fault identification-accurate linkage treatment's whole link monitoring operation and maintenance system;The present application is accurately calibrated by double source data cooperative cooperation, let visible light component positioning advantage and infrared heat perception ability deep fusion, both completely fill the monitoring blind area caused by cabinet shielding, and realize the spatialization perception of equipment state by pixel-temperature binding, solve the problem of not comprehensive perception caused by single sensing and data fragmentation;With the scene fusion of environmental temperature and humidity data and typical fault sample, fault identification process can fully consider the actual operation environment of closed high load switchgear, effectively avoid the misjudgment caused by single data and scene separation, greatly improve the adaptability and accuracy of fault identification;And seamless closed loop linkage of monitoring acquisition, fault identification and hierarchical treatment, can trigger targeted treatment action as soon as fault is identified, avoid the lag of identification and treatment in traditional technology, solve the pain point of not timely response in prior art, finally realize the operation and maintenance effect of more comprehensive monitoring, more accurate identification and more efficient treatment of switchgear, provide active protection for stable operation of power distribution equipment. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0059] Figure 1 It is a module architecture schematic diagram of the switchgear monitoring device based on infrared-visible light double source cooperation of the embodiment of the present application.

[0060] Figure 2 It is a double source data acquisition calibration process schematic diagram of the embodiment of the present application.

[0061] Figure 3 It is a fault identification and hierarchical treatment process schematic diagram of the embodiment of the present application.

[0062] Figure 4 This is a schematic diagram of the low-power operation and maintenance cycle and data interaction process according to an embodiment of the present invention;

[0063] Figure 5 This is a flowchart illustrating the steps of a monitoring method for a distribution cabinet monitoring device based on infrared-visible light dual-source coordination, according to an embodiment of the present invention. Detailed Implementation

[0064] This invention provides a power distribution cabinet monitoring device and method based on infrared-visible light dual-source collaboration, which solves the technical problem that existing power distribution cabinet monitoring and maintenance technologies suffer from incomplete perception due to single sensors and data fragmentation, resulting in insufficient accuracy and timeliness in identifying potential risks to equipment.

[0065] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention solves the pain points of the prior art through an innovative combination of dual-source collaborative monitoring + multi-dimensional data fusion AI recognition + adaptive linkage processing + edge computing low-power operation and maintenance.

[0066] The core differences are: First, an innovative dual-source collaborative structure combining a wide-angle camera array, an infrared temperature measurement array, and a 360° rotating bracket completely eliminates blind spots caused by cabinet obstruction, achieving precise fusion of temperature and visual data; second, a multi-dimensional fusion recognition algorithm based on CNN and attention mechanisms is proposed, integrating visual, temperature, and environmental parameter features to significantly improve the accuracy of identifying hidden faults; third, a dual-layer handling mechanism of environmental control and fault isolation is constructed, which actively suppresses fault-causing factors such as temperature, humidity, and condensation, while quickly preventing the spread of faults; and fourth, a low-power solution using edge computing and dual-mode communication is adopted, processing core data locally and only uploading fault and critical status information, significantly reducing transmission pressure and maintenance costs.

[0067] Please see Figure 1 This invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination, comprising: a dual-source coordinated monitoring module, used to collect visible light image data and infrared temperature measurement data of the power distribution cabinet under test and perform dual-source data calibration to obtain a calibrated dual-source spectrum; a fault identification module, used to acquire environmental temperature and humidity data and typical fault samples of the power distribution cabinet, combine them with the calibrated dual-source spectrum to identify faults, and generate handling instructions based on the fault identification results; and an environmental control module, used to execute associated handling actions according to the handling instructions.

[0068] Dual-Source Collaborative Monitoring Module: This modular component integrates a camera array, an infrared temperature measurement array, and a dual-source data synchronization calibration unit. It features visible light image acquisition, infrared temperature acquisition, and precise correlation of dual-source data. Its core function is to overcome the limitations of single-sensor data, enabling dual visual and temperature perception and data collaboration for the distribution cabinet's status. Visible Light Image Data: This refers to visual image data of the internal components of the distribution cabinet under test, acquired through optical imaging. It includes visual features such as the component's shape, position, and appearance. Infrared Temperature Measurement Data: This refers to temperature data of various areas inside the distribution cabinet under test, acquired through infrared sensing technology. It includes temperature features such as component surface temperature and ambient temperature distribution. Dual-Source Data Calibration: This refers to the process of accurately matching and collaboratively correlating the two types of data by eliminating spatial positional deviations and data dimensional differences caused by equipment installation errors and sensor characteristic differences during the acquisition of visible light image data and infrared temperature measurement data. Dual-Source Calibration Atlas: This refers to a unified atlas formed by integrating the visual features of visible light images and the temperature features of infrared thermography data after processing by a dual-source data synchronization calibration unit. It simultaneously includes the visual and temperature features of the distribution cabinet under test, with a one-to-one spatial correspondence between the two types of features. Its core attribute is the correlation between spatial location, visual information, and temperature information, which can be directly used for fault feature extraction and identification. Fault Identification Module: This module possesses data reception, feature analysis, fault judgment, and command generation functions. Its core function is to combine multi-dimensional data to accurately identify distribution cabinet faults and output targeted handling commands. Ambient Temperature and Humidity Data: This refers to the temperature and relative humidity data of the environment inside or around the distribution cabinet under test. These are important environmental factors affecting the operating status of the distribution cabinet and the induction of faults. Typical Distribution Cabinet Fault Samples: This refers to a pre-stored set of feature data corresponding to various common distribution cabinet faults (such as component overheating, condensation, component loosening, etc.), including the visual features, temperature features, and environmental correlation features of the faults. Handling instructions: These are instructions generated by the fault identification module based on the fault identification results, used to instruct the environmental control module to perform specific actions. They include core information such as fault level, handling type, and execution parameters. Environmental control module: This refers to the execution module with fault response and environmental regulation functions. Its core function is to receive handling instructions and execute corresponding actions to achieve fault blocking, mitigation, or early warning.

[0069] It should be noted that this device is suitable for monitoring the operating status of power distribution terminal equipment such as high and low voltage distribution cabinets and distribution boxes. It can be deployed in various power distribution scenarios such as industrial plants, commercial buildings, and residential communities. Through the coordinated work of the three core modules, it can achieve comprehensive perception of the operating status of the distribution cabinet, accurate fault identification, and rapid response, ensuring the stable operation of power distribution equipment.

[0070] In this embodiment of the invention, the dual-source collaborative monitoring module is deployed at a preset installation location inside the distribution cabinet under test. After startup, it synchronously collects visible light image data and infrared temperature measurement data of the distribution cabinet under test. The built-in basic calibration algorithm eliminates the acquisition deviation of the dual-source data and outputs a calibrated dual-source spectrum. This spectrum integrates the visual information and temperature information of the equipment components, providing core data support for fault identification. The fault identification module is communicatively connected to the dual-source collaborative monitoring module, receives the calibrated dual-source spectrum in real time, and obtains the ambient temperature and humidity data around the distribution cabinet under test through a preset data interface. Combined with pre-stored typical fault samples of distribution cabinets, it performs fault identification on the calibrated dual-source spectrum, identifies whether the distribution cabinet under test has operational abnormalities and the type and severity of the abnormalities, and then generates corresponding handling instructions. The environmental control module is signal-connected to the fault identification module. After receiving the handling instructions, it triggers preset associated handling actions to complete the initial response to the fault or abnormal state, forming a complete working link of data acquisition-fault identification-handling response.

[0071] Please see Figure 1 The present invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination. The dual-source coordination monitoring module includes: a camera group for acquiring visible light image data of the power distribution cabinet to be tested; an infrared temperature measurement array for acquiring infrared temperature measurement data of the power distribution cabinet to be tested; and a dual-source data synchronization calibration unit for performing coordinate matching between the visible light image data and the infrared temperature measurement data to obtain a calibrated dual-source spectrum.

[0072] Camera Array: This refers to an integrated optical imaging component adapted to the internal environment of the power distribution cabinet (temperature resistance -10℃~60℃, electromagnetic interference resistant). It contains at least two high-definition cameras. Its core function is to collect visible light image data of the components inside the power distribution cabinet, capturing visual features such as the appearance, location distribution, and surface anomalies (e.g., oxidation, loosening, ablation marks). Visible Light Image Data: This refers to image data reflecting the internal scene of the power distribution cabinet, acquired by the camera array through optical imaging principles. Its core features include pixel coordinates, grayscale / color channel information, component outlines, and spatial relationships. It is the basic data for locating component positions and identifying appearance anomalies. Infrared Temperature Measurement Array: This refers to an array-type temperature measurement component composed of multiple infrared sensing units, such as a 32×32 pixel array. Its core function is to convert the infrared radiation signals received from objects into temperature values, enabling simultaneous multi-point temperature measurement of the entire internal area of ​​the power distribution cabinet and acquiring temperature distribution data. Infrared temperature measurement data: This refers to the quantitative data collected by the infrared temperature measurement array, reflecting the temperature of various areas inside the distribution cabinet. Its core characteristics include the coordinates of the measurement point, the temperature value, and the temperature distribution gradient. It is the core data for identifying component overheating and temperature anomalies. Dual-source data synchronous calibration unit: This refers to a processing unit (which can be integrated into an edge computing chip or dedicated processing module) with data preprocessing, coordinate matching, and data integration functions. Its core function is to eliminate the acquisition deviations (spatial deviation and temporal deviation) between visible light image data and infrared temperature measurement data, achieving accurate correlation between the two types of data. Coordinate matching: This refers to establishing a mapping relationship between the visible light image pixel coordinate system and the infrared temperature measurement point coordinate system, using a preset calibration benchmark as a reference. This ensures that visual pixels at the same physical location correspond one-to-one with the temperature measurement point, eliminating spatial misalignment problems caused by multi-source data.

[0073] In this embodiment of the invention, the dual-source collaborative monitoring module is deployed in a pre-set installation position at the top or side of the distribution cabinet to be monitored, avoiding high-voltage core components such as busbars and circuit breakers, and ensuring that it is not obstructed by itself. Through the synchronous acquisition of data by the camera group and the infrared temperature measurement array, and the collaborative processing of the dual-source data synchronous calibration unit, accurate binding of visual information and temperature information is achieved. The specific working process is as follows:

[0074] After the module is started, the camera group and the infrared temperature measurement array start data acquisition on the same time axis: the camera group continuously acquires visible light image data of the inside of the distribution cabinet under test according to the preset resolution (e.g., 1920×1080 pixels) and frame rate (e.g., 10 frames / second), completely capturing the appearance of the internal components of the cabinet, such as wiring terminals, circuit breaker contacts, busbars, installation positions and surface conditions; the infrared temperature measurement array simultaneously acquires infrared temperature measurement data of the entire area inside the cabinet according to the preset temperature measurement range (e.g., -20℃~150℃) and temperature measurement accuracy (e.g., ±0.5℃), realizing the perception of temperature distribution of each component and environmental area.

[0075] During the data acquisition process, the acquisition cycles of the camera group and the infrared temperature measurement array are strictly synchronized (time deviation ≤ 10ms) to ensure that visual data and temperature data under the same monitoring scenario are acquired at the same time. After the acquisition is completed, the two types of raw data are synchronously transmitted to the dual-source data synchronization calibration unit through the internal data transmission interface (such as SPI interface).

[0076] After receiving the data, the dual-source data synchronous calibration unit first corrects the distortion of the visible light image data to eliminate the influence of lens distortion, and filters the infrared temperature measurement data for noise to remove abnormal temperature points caused by electromagnetic interference. Then, using the pre-marked fixed markers inside the distribution cabinet (such as metal positioning holes and standard mounting screws) as the calibration benchmark, a unified spatial coordinate system is established. The pixel coordinates of the visible light image are mapped and matched one-to-one with the temperature measurement point coordinates of the infrared temperature measurement data to ensure that the location of each component in the visible light image corresponds to a unique temperature data. Finally, the matched visual features and temperature features are integrated to output a calibration dual-source map containing three-dimensional correlation information of spatial location, appearance status, and temperature value. This map can intuitively present the real-time status and temperature distribution of each component inside the distribution cabinet, providing a highly reliable core data source for the fault identification module.

[0077] Please see Figure 1 This invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination. The dual-source coordination monitoring module also includes a mounting bracket. Both the camera group and the infrared temperature measurement array are mounted on the mounting bracket. The mounting bracket is used to adjust the monitoring angles of the camera group and the infrared temperature measurement array until the preset standard acquisition conditions are met. The preset standard acquisition conditions are specifically that the monitoring coverage of all key monitoring areas in the power distribution cabinet to be tested is greater than the preset standard coverage threshold, and the field of view overlap between the camera group and the infrared temperature measurement array meets the preset standard overlap threshold.

[0078] Mounting bracket: This refers to a support component adapted to the internal environment of the distribution cabinet, used to fix and adjust the installation position and monitoring angle of the camera group and infrared temperature measurement array. Its core function is to ensure the installation stability, comprehensive monitoring coverage, and field-of-view coordination of the dual-source acquisition equipment. Key monitoring area: This refers to the core components and areas within the distribution cabinet that are prone to failure and crucial to power distribution safety, including but not limited to the surfaces and surrounding areas of components such as circuit breaker contacts, terminals, busbars, fuses, and capacitor components. Monitoring coverage: This refers to the coverage ratio of the camera group's (or infrared temperature measurement array's) field of view over all key monitoring areas, and is a core indicator for measuring the comprehensiveness of monitoring. Preset standard coverage threshold: This refers to the pre-set minimum percentage standard for determining whether the monitoring coverage meets the standard. In this embodiment, it is set to ≥99%, and can be flexibly adjusted according to the distribution cabinet model and internal structure differences. Field-of-view overlap: This refers to the overlap ratio between the visible light monitoring field of view of the camera group and the infrared monitoring field of view of the infrared temperature measurement array, and is a core indicator for measuring the synergy of dual-source data. Preset standard overlap threshold: refers to the minimum ratio standard set in advance to determine whether the dual-source vision coordination meets the standard. In this embodiment, it is set to ≥95%. Its value setting needs to match the coordinate matching accuracy requirements of dual-source data synchronization calibration.

[0079] In this embodiment of the invention, the mounting bracket is integrally formed from lightweight, high-strength aluminum alloy, with a temperature resistance range of -10℃ to 60℃, suitable for the enclosed environment inside the distribution cabinet and electromagnetic interference scenarios. The entire bracket is fixed to the top center position or a pre-set side mounting beam inside the distribution cabinet to be tested using expansion bolts. The installation position avoids high-voltage core components such as busbars and circuit breakers, as well as the cabinet door opening and closing path, to prevent obstruction or interference with the normal operation and maintenance of the distribution cabinet. The camera group and infrared temperature measurement array are symmetrically mounted on the front mounting panel of the mounting bracket via a detachable snap-fit ​​structure. The installation distance between the two is fixed at 5~8cm (the optimal spatial distance for dual-source data synchronous calibration), and the mounting panel has a pre-reserved fine-tuning groove, supporting horizontal ±15mm and vertical ±10mm positional adjustments.

[0080] The specific adjustment process for the mounting bracket is as follows:

[0081] After initial installation, rough adjustments are made using the horizontal rotation axis (rotation angle range 0°~180°) and the pitch adjustment axis (adjustment angle range -30°~60°) of the bracket to ensure that the initial field of view of the camera group and the infrared temperature measurement array roughly covers the entire interior of the power distribution cabinet.

[0082] The camera array and infrared temperature measurement array were activated for trial data acquisition to obtain initial visible light images and infrared temperature distribution maps. Key data calculations were then performed using the built-in field-of-view analysis tool.

[0083] Coverage calculation: The outlines of all key monitoring areas (such as circuit breaker contacts, terminals, etc.) within the distribution cabinet to be monitored are pre-marked in the system, and their total pixel area is calculated. In the acquired visible light images (or infrared thermograms), identify and calculate the pixel area of ​​the key monitoring region covered by the field of view. Specifically:

[0084] ;

[0085] Overlap Calculation: For the visible light field of view of the camera group: The acquired visible light images are preprocessed, and invalid areas such as cabinet edges, mounting brackets, and non-power distribution component areas are identified and removed using image segmentation algorithms. Only the area covering the internal components of the power distribution cabinet is retained, and the number of pixels in this area is counted, which is the effective pixel area of ​​the visible light field of view of the camera group. For the infrared field of view of the infrared temperature measurement array: the infrared temperature distribution map acquired in the trial is preprocessed to remove invalid areas such as those exceeding the physical boundary of the distribution cabinet and blank areas with no temperature measurement data, retaining only the temperature measurement area inside the distribution cabinet, and counting the number of pixels in this area, which is the effective pixel area of ​​the infrared field of view of the infrared temperature measurement array. Using pre-marked fixed markers (such as metal positioning holes or standard mounting screws) within the distribution cabinet as registration feature points, the pixel coordinates of these fixed markers are extracted from the visible light image, and the coordinates of the corresponding temperature measurement points are extracted from the infrared temperature distribution map. Through a coordinate transformation algorithm, the coordinate system of the infrared temperature distribution map is aligned to the pixel coordinate system of the visible light image, achieving spatial alignment of the dual-source fields of view. The registered dual-source field of view region is then binarized (the effective field of view region is marked as 1, and the invalid region as 0). The intersection region, i.e., the region simultaneously covered by both the visible light field of view (marked as 1) and the infrared field of view (marked as 1), is taken. The number of pixels in this intersection region is counted, which represents the pixel area of ​​the overlapping region of the dual-source fields of view. Specifically:

[0086]

[0087] If the calculated monitoring coverage is <99% (preset standard coverage threshold), it indicates that there are key monitoring areas that are not covered. The coverage range can be supplemented by fine-tuning the position of the slide groove on the mounting panel or adjusting the pitch angle. If the field of view overlap is <95% (preset standard overlap threshold), the field of view orientation of both can be fine-tuned by rotating the horizontal axis until both calculated values ​​meet the preset standard.

[0088] After adjustment, the positions of each adjustment shaft and slide groove are fixed by the locking knob on the bracket to prevent angle deviation caused by cabinet vibration during long-term operation and to ensure monitoring stability.

[0089] In this embodiment, the preset standard coverage threshold is set to ≥99%, and the preset standard overlap threshold is set to ≥95%. The above simplified calculation method can quickly determine whether the monitoring field of view meets the standard, providing a precise spatial basis for the coordinate matching of the subsequent dual-source data synchronization calibration unit.

[0090] Please see Figure 2 After the dual-source collaborative monitoring module is installed, the entire process of generating the calibration dual-source spectrum is as follows:

[0091] Module startup initialization: After the dual-source collaborative monitoring module is powered on, it automatically completes hardware self-test (detects the circuit connectivity and functional effectiveness of the camera group, infrared temperature measurement array, and dual-source data synchronization calibration unit), and loads preset acquisition parameters (such as camera resolution and infrared temperature measurement range) and calibration reference point configuration information (such as characteristic parameters of fixed markings in the power distribution cabinet). After confirming that the status of each component is normal, it enters the standby state.

[0092] The bracket angle adaptive adjustment: The module calls the built-in field of view analysis tool and combines the initial trial data collected by the camera group and infrared temperature measurement array to automatically control the horizontal rotation axis (0°~180°) and the pitch adjustment axis (-30°~60°) of the mounting bracket to make fine-tuning of the angle. It calculates the coverage of key monitoring areas and the overlap of dual-source field of view in real time until the two indicators reach the preset standards of ≥99% and ≥95% respectively. Then, the bracket adjustment axis is locked by locking the knob to avoid vibration and displacement.

[0093] Data Synchronous Acquisition: After the bracket is adjusted to the standard, the module controls the camera group (according to the preset resolution and frame rate) and the infrared temperature measurement array (according to the preset temperature measurement range and accuracy) to start data acquisition on the same time axis, ensuring that the acquisition time deviation of the two types of data is ≤10ms, and synchronously acquiring visible light image data (including component visual features) and infrared temperature measurement data (including area temperature features) inside the power distribution cabinet to be tested.

[0094] Dual-source coordinate precision: After calibration and acquisition, the dual-source data synchronous calibration unit first performs distortion correction on the visible light image (eliminating lens distortion) and noise filtering on the infrared temperature measurement data (removing abnormal points of electromagnetic interference); then, using the fixed markings in the power distribution cabinet (such as metal positioning holes and standard mounting screws) as calibration reference points, it extracts the coordinates of the reference points in the two types of data, and constructs a transformation matrix of "visible light pixel coordinate system → infrared temperature measurement coordinate system" through affine transformation to achieve a one-to-one mapping and matching between the coordinates of each pixel and the coordinates of the temperature measurement point.

[0095] Panoramic stitching and data fusion: For the collected data from multiple areas inside the power distribution cabinet, the module performs panoramic stitching according to the actual physical layout of the cabinet, integrates the "pixel position-visual features-temperature features" data matched by each area, eliminates data deviations at overlapping areas, and forms a fused data set covering the entire area of ​​the power distribution cabinet.

[0096] Calibration effect verification + data output: The module verifies the calibration effect of the fused data: verify that the coordinate matching error of the key area is ≤1 pixel and the correlation accuracy of temperature-visual features is ≥98%; after the verification is passed, the module outputs a calibration dual-source map containing "spatial location-visual features-temperature features" and transmits it to the fault identification module as the core data source for subsequent fault identification; if the verification fails, the module returns to the "adaptive adjustment of support angle" step to re-execute the process.

[0097] Please see Figure 1 The present invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination. The coordinate matching process is as follows: determining the calibration reference point in the power distribution cabinet to be tested, which is a fixed mark pre-marked in the power distribution cabinet to be tested; using the calibration reference point as a reference, matching the pixel coordinates of the visible light image data with the temperature measurement point coordinates of the infrared temperature measurement data to obtain multiple pixel-temperature maps; integrating the multiple pixel-temperature maps to obtain the calibration dual-source map.

[0098] Calibration reference point: Refers to a pre-marked, permanently fixed physical marker within the distribution cabinet under test. These are typically selected from components unaffected by maintenance operations, such as cabinet structural parts or standard fasteners. Their core function is to provide a unified reference for matching visible light pixel coordinates with infrared temperature measurement point coordinates, ensuring accurate coordinate mapping. Fixed marker: Refers to a stable physical feature (such as metal positioning holes or standard mounting screws) inherent to the distribution cabinet or additionally marked during installation. It serves as the concrete carrier of the calibration reference point. Pixel coordinates: In visible light image data, these are two-dimensional coordinates used to identify the spatial position of a single pixel. They reflect the relative position of a component in the visual image and are the basis for locating the physical position of the component. Temperature measurement point coordinates: In infrared temperature measurement data, these are two-dimensional coordinates used to identify the spatial position of a single temperature measurement point. They reflect the physical region corresponding to the temperature data and are the basis for associating the temperature status of the component. Pixel-temperature map: Refers to a set of local region data obtained through coordinate matching. Its core content is the correlation between visible light pixel coordinates and the corresponding infrared temperature measurement point temperature values. It is a local unit constituting the calibration dual-source map.

[0099] In this embodiment of the invention, the coordinate matching process is carried out based on visible light image data and infrared temperature measurement data collected by the dual-source collaborative monitoring module. The specific operation steps are as follows:

[0100] Determine the calibration reference point:

[0101] Select 2-3 pre-marked fixed markers inside the distribution cabinet to be tested as calibration reference points (the markers must meet the conditions of "long-term fixed location, unaffected by cabinet maintenance operations, and clearly detectable by both source and source devices"). In this embodiment, the selected calibration reference points are:

[0102] Reference point 1: The circular metal positioning hole on the inner wall of the top of the distribution cabinet (5mm in diameter, marked with high-temperature reflective paint during installation, with no obstruction on the surface).

[0103] Reference point 2: The M6 ​​standard mounting screw next to the circuit breaker (the screw head is marked with wear-resistant paint for high differentiation from surrounding components). These reference points are marked during the factory or installation phase of the distribution cabinet and remain fixed during subsequent operation, providing a stable reference for coordinate matching.

[0104] Coordinate matching with reference to the calibration benchmark:

[0105] First, the dual-source data is preprocessed: distortion correction is performed on the visible light image data to eliminate coordinate deviations caused by camera lens distortion; noise filtering is performed on the infrared temperature measurement data to remove abnormal temperature measurement points caused by electromagnetic interference.

[0106] Extracting the coordinate information of the reference points: Using a feature recognition algorithm, the pixel coordinates of two calibration reference points are extracted from the preprocessed visible light image data; at the same time, the coordinates of the temperature measurement points corresponding to the same reference point are extracted from the preprocessed infrared temperature measurement data through temperature distribution characteristics (the temperature of the metal reference point has a stable difference from that of the surrounding non-metallic parts).

[0107] Establish coordinate mapping relationship: Using the coordinates of two calibration reference points as a reference, a transformation matrix of "visible light pixel coordinate system → infrared temperature measurement coordinate system" is constructed through an affine transformation algorithm. Based on this matrix, the coordinates of each pixel in the visible light image data are mapped to the corresponding infrared temperature measurement point coordinates, realizing the association of "one visible light pixel coordinate corresponding to one infrared temperature measurement point coordinate". This operation is performed on key monitoring areas such as the busbar area, terminal area, and circuit breaker contact area in the distribution cabinet to obtain the "pixel position-temperature value" associated data set corresponding to each area, i.e., multiple pixel-temperature maps.

[0108] The resulting calibrated dual-source spectrum was obtained by integration:

[0109] Based on the actual physical layout inside the distribution cabinet, the pixel-temperature maps of each key monitoring area are stitched together. At the same time, the associated data of the overlapping areas are deduplicated (keeping the temperature values ​​with higher measurement accuracy). Finally, an integrated map covering all key monitoring areas of the distribution cabinet under test is formed, namely the calibrated dual-source map. Each pixel in this map contains both the corresponding component visual features (such as shape and appearance) and temperature features (such as real-time temperature value).

[0110] Please see Figure 1 This invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source collaboration. The fault identification module includes: an edge computing chip for receiving environmental temperature and humidity data of the power distribution cabinet under test, and denoising the calibrated dual-source spectrum to obtain a dual-source feature matrix. Specifically, the dual-source feature matrix is ​​a two-dimensional standardized matrix composed of pixel visual features and temperature features. The environmental temperature and humidity data is normalized to obtain environmental feature parameters; a feature database for storing typical fault samples of the power distribution cabinet; a fault identification model for using the dual-source feature matrix, environmental feature parameters, and typical fault samples of the power distribution cabinet to perform fault identification and generate anomaly confidence levels; and the edge computing chip is also used to generate associated handling instructions based on the preset fault level range to which the anomaly confidence level belongs.

[0111] ARM Cortex-A53 architecture: A low-power, high-performance embedded processor architecture whose core function is to provide local data processing hardware computing power for the fault identification module, supporting the stable operation of AI algorithms in the enclosed environment of the power distribution cabinet; Multi-dimensional fusion fault identification model: An intelligent identification model built on CNN + attention mechanism, whose core features are dual input channels + dynamic weight allocation, which can fuse visual, temperature, and environmental features to achieve accurate fault classification; Attention mechanism: An algorithm optimization mechanism whose core function is to dynamically increase the weight ratio of core features (such as temperature features) during the fault identification process, avoid interference from irrelevant features, and improve the identification accuracy; Dual input channels: The feature input structure of the fault identification model, divided into visual feature channels and temperature-environment feature channels, whose core function is to realize the parallel input and fusion analysis of multi-dimensional data; 12 typical fault features: A set of common fault types of power distribution cabinets pre-trained in the model, including overheating, condensation, loosening, abnormal noise, insulation aging, etc., which serve as the matching benchmark for fault identification; Online feature database update: A functional extension of the feature database, whose core function is to synchronize new fault samples from the remote platform and continuously optimize the model's identification capabilities in complex scenarios.

[0112] In this embodiment of the invention, a local high-computing power + lightweight intelligent algorithm is used as the core architecture. The entire system is installed in the low-voltage area inside the power distribution cabinet via a guide rail. It communicates with the dual-source collaborative monitoring module and the environmental control module via an RS485 bus. Specifically, it consists of three parts: an edge computing chip, a fault identification model, and a feature database. Its working process is as follows:

[0113] This module uses an edge computing chip with an ARM Cortex-A53 architecture, which has low power consumption (power consumption ≤ 4W), local data processing capabilities, and does not rely on cloud computing power. First, it quickly receives the calibrated dual-source spectrum transmitted by the dual-source collaborative monitoring module and the ambient temperature and humidity data collected by the temperature and humidity sensors. Then, it performs Gaussian noise reduction on the calibrated dual-source spectrum to eliminate visual noise points and temperature fluctuation points caused by cabinet reflection and electromagnetic interference. Subsequently, it extracts the pixel visual features (such as component edge contours and surface gray-scale distributions) and temperature features (such as real-time temperature values) of each pixel in the denoised spectrum, combines the two types of features into a two-dimensional standardized matrix, namely the dual-source feature matrix, and at the same time uses the min-max normalization algorithm to map the ambient temperature and humidity data to the 0-1 interval (the temperature value range is preset to -10°C to 60°C, and the humidity is 0% to 100%) to obtain ambient feature parameters; the feature database配套 with the edge computing chip is a structured data set stored in an industrial-grade flash memory with a local capacity of ≥ 16GB. Its core content is 12 types of typical fault samples of distribution cabinets such as overheating, condensation, loosening, abnormal noise, and insulation aging. Each sample contains an image sample corresponding to the fault, a temperature threshold range, and ambient correlation parameters (matching intervals of temperature and humidity, condensation status), and at the same time, this database supports an online update function, and can synchronize new fault samples from the remote operation and maintenance platform through the wireless communication module to continuously optimize the adaptability of subsequent fault identification; the multi-dimensional fusion fault identification model is constructed based on CNN (Convolutional Neural Network) + attention mechanism. Different from the single-image CNN algorithm, it sets up a dual-input channel of a visual feature channel (visual features input into the dual-source feature matrix) and a temperature-ambient feature channel (temperature features + ambient feature parameters input into the dual-source feature matrix). At the same time, through the attention mechanism, the weight of the temperature feature is dynamically increased to 60% (visual feature weight 30%, ambient feature weight 10%) in the fault scenario to strengthen the recognition priority of the core fault features. Subsequently, the feature data of the dual-input channel is convolutionally matched with 12 types of fault samples in the feature database, calculates the cosine similarity between the current data and each fault sample, and finally outputs an abnormal confidence level ranging from 0 to 1 (the closer the value is to 1, the higher the fault matching degree), and the fault identification accuracy of this model is ≥ 99%; finally, the edge computing chip is built-in with a preset fault level interval (abnormal confidence level ≥ 0.85 is the emergency fault interval, 0.70 ≤ abnormal confidence level < 0.85 is the serious fault interval, 0.50 ≤ abnormal confidence level < 0.70 is the ambient anomaly interval, abnormal confidence level < 0.85 is the no-fault interval), and generates corresponding disposal instructions (including instruction type, target execution component, execution parameters) according to the interval where the current abnormal confidence level belongs. For example, when the abnormal confidence level is 0.92 (emergency fault interval), a disposal instruction of "breaker linkage interface executes fault loop isolation + audible and visual alarm starts" is generated.

[0114] Please see Figure 1 The present invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source collaboration. The fault identification process is as follows: visual and temperature dual-channel feature extraction is performed on the dual-source feature matrix to obtain visual feature parameters and temperature feature parameters; a visual feature vector is constructed using the visual feature parameters; a temperature-environment feature vector is constructed using the temperature feature parameters and environmental feature parameters; the visual feature vector and the temperature-environment feature vector are weighted and fused to obtain a fused feature vector; the cosine similarity between the fused feature vector and typical fault samples of the power distribution cabinet is calculated to determine the anomaly confidence level.

[0115] Dual-channel feature extraction (visual and temperature): This refers to the operation of extracting effective features separately from the visual and temperature feature components of the dual-source feature matrix, relying on a CNN network and fully connected layers. Its core function is to separate and enhance fault-related features of different dimensions. Visual feature parameters: These are high-dimensional quantified data extracted from the visual portion of the dual-source feature matrix, reflecting the surface state (e.g., oxidation, ablation) and contour features of the component. They are the foundation for constructing visual feature vectors. Temperature feature parameters: These are statistical quantified data extracted from the temperature portion of the dual-source feature matrix, reflecting the regional temperature distribution (e.g., peak value, gradient). They are the core elements for constructing temperature-environment feature vectors. Visual feature vector: This is a high-dimensional vector formed by concatenating visual feature parameters in a preset order. The unified expression form of visual features; Temperature-environment feature vector: refers to a high-dimensional vector formed by concatenating temperature feature parameters and environmental feature parameters, which is a unified expression form of temperature-environment correlation features; Weighted fusion: refers to the operation of linearly combining feature vectors of different dimensions according to the importance of features, and its core function is to integrate multi-dimensional features to improve the accuracy of fault identification; Fusion feature vector: refers to the unified feature vector obtained after weighted fusion, which integrates multi-dimensional fault correlation features of vision, temperature, and environment; Cosine similarity: refers to a quantitative index (range 0-1) that measures the angle between two high-dimensional vector spaces, and is used in this device to determine the degree of matching between the current fused feature vector and the fault sample vector, and the result is the anomaly confidence.

[0116] In this embodiment of the invention, a pre-loaded CNN network is first invoked to perform visual and temperature dual-channel feature extraction on the dual-source feature matrix. For the visual feature part corresponding to each pixel in the dual-source feature matrix, edge and texture features are extracted using the convolutional kernels of the CNN, and then the dimensions are compressed by pooling layers to finally obtain visual feature parameters containing the surface state and contour features of the component. For the temperature feature part in the dual-source feature matrix, statistical features such as temperature peak, regional mean, and gradient change are extracted through fully connected layers to obtain temperature feature parameters. Subsequently, these visual feature parameters are concatenated in a preset order to construct a visual feature vector. At the same time, the temperature features are... The parameters and environmental feature parameters are concatenated dimensionally to construct a temperature-environment feature vector. Then, based on the model's built-in attention mechanism, the weights of the two types of feature vectors are dynamically allocated according to the current monitoring scenario (the weight ratio of strengthening temperature-environment features in fault scenarios). The visual feature vector and the temperature-environment feature vector are then linearly fused with weights to obtain a fused feature vector that integrates multi-dimensional features. Finally, the fault feature vector corresponding to a typical fault sample of a distribution cabinet matching the current monitoring area is retrieved from the feature database. The cosine similarity between the fused feature vector and the fault sample vector is calculated, and the result is the anomaly confidence level corresponding to the current state.

[0117] Please see Figure 3 After receiving the data, the fault identification module completes the entire chain of fault determination and handling decision-making. The specific steps are as follows:

[0118] The edge computing chip of the multi-source data receiving + preprocessing fault identification module synchronously receives the calibration dual-source map transmitted by the dual-source collaborative monitoring module and the environmental data transmitted by the sensor device (temperature and humidity / condensation sensor). First, Gaussian noise reduction processing is performed on the calibration dual-source map to extract pixel visual features and temperature features and construct a dual-source feature matrix. At the same time, min-max normalization processing is performed on the environmental data to obtain environmental feature parameters.

[0119] The dual-channel feature parallel extraction invokes a multi-dimensional fusion fault identification model, initiating parallel extraction of the "visual feature channel" and the "temperature-environment feature channel": the visual feature channel extracts visual feature parameters such as component contours and surface states from the dual-source feature matrix through CNN convolutional kernels; the temperature-environment feature channel extracts feature parameters such as temperature distribution and environmental parameters through fully connected layers.

[0120] The CNN+attention mechanism fusion operation constructs visual feature vectors and temperature-environment feature vectors from the feature parameters extracted from the two channels. The weights are dynamically allocated through the built-in attention mechanism of the model (the weight of temperature-environment features is increased to 60% in the fault scenario). The two types of vectors are weighted and fused to obtain a fused feature vector that integrates multi-dimensional features. Then, the cosine similarity between this vector and typical fault samples in the feature database is calculated to obtain the anomaly confidence.

[0121] To determine whether an anomaly exists, a confidence level of ≥50% is used as the judgment threshold.

[0122] If the confidence level is less than 50% (no anomalies): mark the current data as "normal data" and store it in the local storage module;

[0123] If the confidence level is ≥50% (an anomaly exists): proceed to the subsequent anomaly level and handling decision-making process.

[0124] Anomaly level quantification combines anomaly confidence and fault type to match a preset fault level range:

[0125] Confidence level ≥ 85% → Emergency failure;

[0126] 70% ≤ confidence level < 85% → serious fault;

[0127] 50% ≤ confidence level < 70% → Abnormal environment;

[0128] It should be noted that, Figure 3 Minor faults correspond to mild scenarios within environmental anomalies in this embodiment of the invention.

[0129] Dynamic decision-making for handling priorities matches corresponding handling actions based on the level of abnormality:

[0130] Emergency Fault → Emergency Isolation + Alarm Linkage;

[0131] Critical fault → Warning + Cooling / Dehumidification linkage;

[0132] Abnormal environment → adaptive environmental regulation;

[0133] Minor faults → Early warning + trend tracking.

[0134] The closed-loop record of the handling results integrates information such as handling instructions, execution status, and environmental feedback data, stores them in the local storage module, and simultaneously uploads them to the remote operation and maintenance platform to complete the closed-loop record of this process.

[0135] Please see Figure 1This invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination. The environmental control module includes a priority decision unit, a circuit breaker linkage interface, an audible and visual alarm, environmental control components, and sensor devices. The priority decision unit is used to match and execute the corresponding action based on the instruction type of the instruction. The circuit breaker linkage interface is used to perform fault circuit isolation. The audible and visual alarm is used to provide audible and visual alarms. The environmental control components include an intelligent fan and a dehumidifier. The intelligent fan is used to dissipate heat when the fault type is overheating. The dehumidifier is used to dehumidify when the fault type is condensation. The sensor devices include a temperature and humidity sensor and a condensation sensor. The temperature and humidity sensor is used to collect the environmental temperature and humidity data inside the power distribution cabinet under test in real time and transmit it to the fault identification module. The condensation sensor is used to collect condensation data inside the power distribution cabinet under test.

[0136] Environmental Control Module: This modular component integrates decision-making, execution, and sensing functions. It responds to handling commands, performs fault isolation, provides alarms, and optimizes the environment. Its core function is to rapidly handle distribution cabinet faults and dynamically adjust the operating environment, ensuring safe equipment operation. Priority Decision Unit: The core control component of the environmental control module. Its core function is to receive and parse handling commands, match corresponding handling action combinations according to preset priority rules, and coordinate the orderly operation of various execution components. Circuit Breaker Linkage Interface: A signal interface using standardized electrical design. Its core function is to establish a communication connection between the environmental control module and the circuit breaker, executing fault circuit isolation operations through output control signals. Audible and Visual Alarm: A warning component with dual audible and visual alarm functions. Its core function is to provide feedback on the distribution cabinet fault status to maintenance personnel and surrounding personnel through perceptible audible and visual signals, reminding them to handle the issue promptly. Environmental Control Components: A collection of execution components used to regulate the internal temperature and humidity environment of the distribution cabinet. Its core function is to optimize the operating environment and alleviate or eliminate faults such as overheating and condensation through heat dissipation and dehumidification. Intelligent Fan: A heat dissipation component with dynamic speed adjustment. Its core function is to address overheating faults by achieving efficient heat dissipation through precise speed control, thereby reducing the temperature inside the cabinet. Dehumidifier: A humidity control component employing condensation dehumidification technology. Its core function is to eliminate condensation inside the cabinet, reduce ambient humidity, and prevent problems such as insulation degradation and component corrosion caused by condensation. Sensor Equipment: A collection of sensing components used to collect environmental parameters inside the distribution cabinet. Its core function is to provide real-time environmental data for the fault identification module and to provide feedback for the start and stop of the environmental control components. Temperature and Humidity Sensor: A sensing device used to collect real-time temperature and relative humidity data inside the distribution cabinet. Its core function is to provide accurate environmental temperature and humidity data support. Condensation Sensor: A sensing device used to detect whether condensation occurs inside the distribution cabinet and the level of condensation. Its core function is to provide data for condensation fault diagnosis and dehumidifier control.

[0137] In this embodiment of the invention, the environmental control module is installed on a guide rail inside the low-voltage area of ​​the distribution cabinet to be tested. It establishes real-time communication with the fault identification module and the dual-source collaborative monitoring module via an RS485 bus. Each component works collaboratively in a closed-loop logic of decision-execution-feedback. The specific implementation process is as follows:

[0138] The core of the module, the priority decision-making unit, pre-stores a mapping table of "instruction type - handling action" and handling priority rules (emergency fault handling priority > serious fault handling priority > environmental anomaly handling priority > minor fault handling priority). After receiving handling instructions from the fault identification module in real time, it first parses the core information contained in the instruction, such as the fault type, severity level, and target area. Then, it matches the corresponding handling action combination according to the mapping table and triggers execution in priority order. The circuit breaker linkage interface adopts a standardized electrical interface design, directly connecting to the control circuit of the circuit breaker in the distribution cabinet. When an emergency fault handling instruction is received, it immediately outputs a switch signal to control the circuit breaker to trip, quickly isolating the fault circuit and cutting off the connection between the fault area and the power supply to prevent the fault from escalating. Audible and visual alarms are installed inside the distribution cabinet door and on the top outside of the cabinet (dual deployment ensures that maintenance personnel and surrounding personnel can perceive them). When various fault handling instructions are received, an audible alarm (alarm intensity ≥85dB, frequency 500Hz~1kHz) and a visual alarm (red LED flashing, frequency 2 times / second) are simultaneously activated until the fault is cleared or manually reset. The intelligent fan and air conditioner in the environmental control component... All humidifiers adopt a low-power DC drive design and are installed at the ventilation openings inside the distribution cabinet. The intelligent fan automatically adjusts its speed based on temperature feedback signals. When the handling command determines that the fault type is overheating, it immediately starts and dynamically adjusts the fan speed according to the difference between the temperature of the fault area and the preset threshold to achieve precise heat dissipation. The dehumidifier starts when the handling command determines that the fault type is condensation, or when the condensation data collected by the condensation sensor reaches the preset alarm threshold. It reduces the humidity inside the cabinet through condensation dehumidification until the humidity returns to a safe range. The temperature and humidity sensors in the sensor equipment are installed in the central area inside the distribution cabinet. (Avoiding heat sources and vents to ensure accurate data collection) The temperature and humidity data inside the cabinet are collected in real time at a cycle of 5 seconds, and transmitted to the fault identification module in real time through the communication interface to provide environmental data support for fault identification; The condensation sensor is installed at the bottom of the power distribution cabinet (the area where condensation is prone to occur), and collects condensation data inside the cabinet at a cycle of 10 seconds (including whether condensation exists and the level of condensation). This data is transmitted to the fault identification module to assist in fault judgment and also fed back to the priority decision-making unit to provide direct basis for the start and stop of the dehumidifier, forming a closed-loop control of fault response-environmental adjustment-data feedback.

[0139] Please see Figure 1The present invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source coordination. The process of matching and executing the corresponding handling action according to the type of the handling instruction is as follows: when the handling instruction is an emergency fault handling instruction, the fault circuit is isolated through the circuit breaker linkage interface, and an audible and visual alarm is activated; when the handling instruction is a serious fault handling instruction, the corresponding environmental control component is activated according to the fault type, and an audible and visual alarm is activated; when the handling instruction is an environmental anomaly handling instruction, the environmental control component is activated; when the handling instruction is a minor fault handling instruction, the audible and visual alarm is activated, and the sensor equipment is controlled to monitor the environmental status of the power distribution cabinet under test.

[0140] Emergency Fault Handling Command: This command, generated by the fault identification module for scenarios requiring immediate risk isolation, such as short-circuit precursors or component overheating and fire risks, is the highest priority command type. Its core function is to trigger rapid risk isolation actions. Severe Fault Handling Command: This command, generated for scenarios such as continuous overheating or severe condensation that may damage equipment, triggers environmental control and alarm linkage actions to mitigate fault development. Abnormal Environment Handling Command: This command, generated for scenarios where cabinet temperature and humidity exceed safe thresholds but do not cause equipment failure, triggers environmental control components to optimize the operating environment. Minor Fault Handling Command: This command, generated for scenarios such as slightly loose components or slightly elevated local temperatures that do not reach the severe fault threshold, triggers alarm prompts and enhanced monitoring to track fault development trends. Fault Circuit Isolation: This refers to a safety protection action that uses the circuit breaker linkage interface to control the circuit breaker to trip, cutting off the electrical connection between the faulty area and the power supply, preventing the fault from spreading to other areas. Handling Actions: These refer to the specific operations performed by the various execution components of the environmental control module (circuit breaker linkage interface, audible and visual alarms, environmental control components, etc.) according to handling instructions. They are the direct means of responding to faults or anomalies. Enhanced Monitoring: This refers to increasing the data acquisition frequency of sensor devices in minor fault scenarios to focus on monitoring the fault-related areas and the overall environment. Its core function is to accurately capture the development trend of the fault.

[0141] In this embodiment of the invention, after receiving the handling instructions transmitted by the fault identification module in real time, the handling priority decision unit first analyzes the instruction type (emergency fault handling instruction, serious fault handling instruction, environmental anomaly handling instruction, minor fault handling instruction), fault type (such as overheating, condensation, short circuit precursor, etc.), fault location information, and other core parameters. Then, according to the preset handling priority rules (emergency fault handling has the highest priority, followed by serious faults, environmental anomalies, and minor faults), the corresponding handling process is initiated. When the analysis determines that the handling instruction is an emergency fault handling instruction (corresponding to a fault type such as short circuit precursor, component overheating and fire risk, etc., which require immediate blocking), the handling priority decision unit first triggers the circuit breaker linkage interface, outputs standardized control signals to link with the circuit breaker control circuit in the distribution cabinet, quickly executes fault circuit isolation, cuts off the electrical connection between the fault area and the main power supply, and simultaneously activates the audible and visual alarm, continuously outputting audible and visual alarm signals of preset intensity until the fault is cleared or the maintenance personnel manually reset it. During this process, it will provide real-time feedback to the fault identification module on "circuit isolation completed" and "alarm activated". The status signal; when the handling instruction is determined to be a serious fault handling instruction (corresponding to fault types such as continuous overheating, severe condensation, etc., which may cause equipment damage), the handling priority decision unit first matches the corresponding environmental control component according to the fault type in the instruction. If it is an overheating fault, the intelligent fan is activated; if it is a condensation fault, the dehumidifier is activated. At the same time, the audible and visual alarm is activated to issue a continuous alarm signal. During the operation of the environmental control component, it receives environmental data fed back by sensor devices in real time and dynamically adjusts operating parameters (such as fan speed and dehumidifier power) until the fault is relieved or the environment returns to a safe range; when the handling instruction is determined to be an environmental abnormality handling instruction (corresponding to the scenario where the temperature and humidity inside the cabinet exceed the safe threshold but do not cause equipment failure), the handling priority decision unit directly activates the corresponding environmental control component (activating the intelligent fan when the temperature is too high, activating the dehumidifier when the humidity is too high or there is condensation). In the event of a dehumidifier malfunction, the system does not require the activation of the audible and visual alarms. Instead, it uses sensors to collect environmental data in real time and feeds it back to the decision-making unit, forming a closed-loop regulation until the environmental parameters inside the cabinet return to the preset safe range. When the handling instruction is determined to be a minor fault handling instruction (corresponding to scenarios such as slightly loose components or slightly high local temperatures that have not reached the threshold for serious faults), the handling priority decision-making unit immediately activates the audible and visual alarms to issue intermittent audible and visual alarm signals (distinguishing itself from the continuous alarms for emergency and serious faults to avoid unnecessary interference). At the same time, it sends encrypted monitoring instructions to the sensor devices, controlling the temperature and humidity sensors and condensation sensors to increase the data acquisition frequency (shortening the normal acquisition cycle to 1 / 3 of the normal cycle). This allows for focused monitoring of the fault-related area and the overall environmental status inside the cabinet, and real-time feedback of monitoring data to the fault identification module, providing support for judging the fault development trend.

[0142] Please see Figure 1The present invention provides a power distribution cabinet monitoring device based on infrared-visible light dual-source collaboration, which further includes a low-power operation and maintenance module; the low-power operation and maintenance module includes a local storage module, a wireless communication module, and a low-power management unit; the local storage module is connected to the dual-source collaborative monitoring module and the fault identification module respectively, and is used for data classification and storage; the wireless communication module is connected to the local storage module and is used to upload the classified data stored in the local storage module to the remote operation and maintenance platform; the low-power management unit is used to switch the monitoring device to a low-power operation and maintenance mode when the monitoring device is not in the monitoring stage.

[0143] Low-power operation and maintenance module: This modular component integrates data storage, remote communication, and power consumption management functions. It enables data traceability, remote operation and maintenance, and low-power operation of the monitoring device. Its core function is to reduce device energy consumption while ensuring operational and maintenance needs, thereby improving deployment practicality and lifespan. Local storage module: This storage component uses industrial-grade flash memory. Its core function is to categorize and store the collected data, identification results, and handling instructions generated by each module of the monitoring device, providing data support for remote uploading and operation and maintenance traceability. Wireless communication module: This communication component is based on low-power wide-area network (LPWAN) technology. Its core function is to enable bidirectional data transmission between locally stored data and the remote operation and maintenance platform, supporting batch uploading of regular data and emergency uploading of fault data. Low-power management unit: This is the control core of the low-power operation and maintenance module. Its core function is to monitor the device's operating status, determine whether to activate the low-power operation and maintenance mode, and optimize energy consumption by adjusting the operating parameters of each module. Low-power operation and maintenance mode: This refers to the energy-saving operation mode when the device is not in the monitoring phase. Its core characteristic is the shutdown or frequency reduction of unnecessary components, retaining only core functions for low-power operation to reduce energy consumption. Remote Operation and Maintenance Platform: Deployed as a centralized operation and maintenance management platform in the backend, its core function is to receive categorized data uploaded by monitoring devices, issue configuration commands, and realize remote control and operation and maintenance analysis of multiple monitoring devices. Categorized Storage: This refers to the method of dividing and storing data according to data type (collected data, fault data, command data, etc.) and timestamp. Its core function is to improve data retrieval and upload efficiency and ensure that critical data is not lost. Wake-up Conditions: These refer to the triggering factors that cause the low-power operation and maintenance mode to switch back to normal monitoring mode, including preset wake-up cycles, environmental anomaly detection, and remote wake-up commands. Their core function is to ensure that the device can quickly restore its monitoring capabilities when needed.

[0144] In this embodiment of the invention, the low-power operation and maintenance module adopts a modular integrated design. The entire module is installed on a rail in the low-voltage area inside the distribution cabinet to be tested. It establishes communication connections with the dual-source collaborative monitoring module, fault identification module, and environmental control module via an SPI bus. Each component works collaboratively according to the logic of data storage-remote upload-power control. Specifically, the core local storage module uses industrial-grade high-reliability flash memory (with shock and electromagnetic interference resistance, and a storage capacity ≥32GB). It establishes connections with the dual-source collaborative monitoring module and the fault identification module through bidirectional data interfaces, and stores data according to the rule of "data type + timestamp," including... The dual-source collaborative monitoring module transmits raw visible light image data, infrared temperature measurement data, and calibrated dual-source spectra, while the fault identification module transmits environmental characteristic parameters, anomaly confidence levels, handling instructions, and fault identification results. It also supports priority storage of critical fault data (fault data is marked as non-repeatable, while regular data is overwritten cyclically after reaching the storage limit), ensuring that core data for maintenance traceability is not lost. The wireless communication module bound to the local storage module uses low-power wide-area network technology (such as LoRa and NB-IoT) to support long-distance, low-speed data transmission. Its data upload logic combines "on-demand upload + triggered upload," and under normal conditions... The system uploads categorized and stored routine data in batches at preset intervals (configurable via a remote maintenance platform). When the monitoring device detects a fault, it immediately triggers an emergency upload mechanism, prioritizing the upload of fault-related data (fault type, handling instructions, calibration dual-source spectra at the time of the fault, etc.). It also supports receiving configuration instructions from the remote maintenance platform (such as adjusting the data upload cycle or updating the fault sample database), enabling bidirectional data interaction. The low-power management unit incorporates a condition detection circuit and power regulation algorithm to monitor the working status of each module of the monitoring device in real time, determining whether it is in the monitoring phase (defined as the process of dual-source data acquisition, fault identification, and handling response). When a fault is detected... When not in the monitoring phase (e.g., no data collection needs, no fault handling actions, and the preset idle time has been reached), the low-power operation and maintenance mode is automatically activated: the operating frequency of the edge computing chip is reduced by the control signal, the continuous acquisition function of the camera group and infrared temperature measurement array is turned off, the acquisition cycle of the sensor device is extended to 5 to 10 times the normal cycle, and only the low-power standby state of the local storage module and the wireless communication module and the working condition monitoring function of the low-power management unit are retained. When the wake-up conditions are met (e.g., the preset wake-up cycle is reached, the sensor detects an environmental anomaly, or a remote wake-up command is received), the mode is quickly switched back to normal monitoring mode to ensure a balance between energy consumption control and monitoring response.

[0145] Please see Figure 4After the device is powered on, the low-power operation and maintenance module realizes the full-link operation of "low-power cyclic operation + data uploading on demand". The specific steps are as follows: After the monitoring device is powered on, the low-power operation and maintenance module and other modules synchronously complete hardware self-test and initialization, load preset low-power parameters (such as sleep cycle and wake-up conditions), and then enter the cyclic workflow. The low-power management unit executes the "sleep-wake" cycle according to the preset cycle (which can be remotely configured, such as 10 minutes / time by default). In the sleep phase, the device switches to low-power operation and maintenance mode, shuts down unnecessary components (such as continuous camera acquisition and high-frequency infrared array temperature measurement), and only retains the working condition monitoring function of the low-power management unit. In the wake-up phase, when the preset wake-up cycle is reached or the wake-up condition is triggered (such as environmental abnormality or remote command), the device quickly resumes normal monitoring mode and starts the subsequent data acquisition and processing process. After waking up, the device starts key data acquisition. The dual-source collaborative monitoring module acquires visible light image data and infrared temperature measurement data, and the sensor device acquires temperature, humidity, and condensation data. All data is synchronously transmitted to the local storage module for temporary caching. Then, the edge computing chip preprocesses the collected key data, calibrating the dual-source data to obtain a calibrated dual-source map, normalizing the environmental data, and simultaneously executing a fault identification process to determine if fault data exists. This determination is based on the "abnormal confidence level ≥ 50%" output by the fault identification module. If fault data exists (yes), the "fault data upload" stage begins, where the wireless communication module initiates an emergency upload mechanism to prioritize uploading fault data (fault type, calibrated dual-source map, handling instructions, etc.) to the remote operation and maintenance platform. If no fault data exists (no), the "storage only, no upload" stage begins, where regular data is only stored in the local storage module and synchronized to the remote platform when a subsequent batch upload cycle is triggered. After receiving the data, the remote operation and maintenance platform can issue configuration instructions (such as adjusting the sleep cycle or updating fault samples). The low-power operation and maintenance module receives the instructions and updates the parameters, then returns to the "loop execution of the sleep-wake process" stage, continuously running in a loop.

[0146] Please see Figure 5 The present invention provides a monitoring method for a distribution cabinet monitoring device based on infrared-visible light dual-source coordination, comprising:

[0147] Step 101: Collect visible light image data and infrared temperature measurement data of the power distribution cabinet to be tested, and perform dual-source data calibration to obtain the calibrated dual-source spectrum.

[0148] Step 102: Obtain ambient temperature and humidity data and typical fault samples of the power distribution cabinet, combine them with the calibrated dual-source spectrum to identify faults, and generate handling instructions based on the fault identification results.

[0149] Step 103: Execute the associated handling actions according to the handling instructions.

[0150] Dual-source data calibration: refers to the process of eliminating the acquisition deviation between visible light image data and infrared temperature measurement data through preprocessing, coordinate matching and other operations, so as to achieve accurate spatial correlation between the two types of data. It is the core link in generating a calibrated dual-source spectrum.

[0151] Fault identification: refers to the process of integrating dual-source features, environmental parameters and typical fault samples, and using algorithm calculations to determine whether the equipment has a fault, the type of fault and the severity of the fault. The core output is the anomaly confidence level and the handling instructions.

[0152] Handling actions: These refer to the specific operations performed by each execution component of the environmental control module according to the handling instructions, including fault circuit isolation, audible and visual alarms, environmental adjustment, and encrypted monitoring. They are the direct means of responding to faults or anomalies.

[0153] Preset standard acquisition conditions: These refer to the quantitative standards set to ensure the comprehensiveness and coordination of data acquisition, including the coverage threshold of key monitoring areas and the overlap threshold of dual-source field of view. These are necessary conditions to be met before data acquisition.

[0154] Fusion feature vector: refers to a unified high-dimensional vector obtained by integrating visual feature vectors, temperature feature parameters and environmental feature parameters. It integrates multi-dimensional fault correlation features and is the core input data for fault identification.

[0155] Preset fault severity range: refers to the range of fault severity based on the degree of anomaly confidence. Its core function is to realize fault graded response and ensure that different levels of faults are handled appropriately.

[0156] Encrypted monitoring: refers to increasing the sensor acquisition frequency in minor fault scenarios to focus on monitoring the fault-related areas and the environment inside the cabinet. Its core function is to accurately capture the development trend of the fault.

[0157] Data classification storage: refers to the storage method that divides data according to data type (collected data, calibration data, identification results, instruction data, etc.) and timestamp. Its core function is to ensure data retrieval efficiency and prevent the loss of key data.

[0158] Low-power operation and maintenance mode: refers to the mode in which non-essential components are turned off or their frequency reduced when the device is idle, and only the core functions are retained for low-power operation. Its core function is to reduce energy consumption and extend the service life of the device deployment.

[0159] In this embodiment of the invention, in step 101, the acquisition and calibration stage, the monitoring angles of the camera group and the infrared temperature measurement array are first adjusted using the mounting bracket of the dual-source collaborative monitoring module until the preset standard acquisition conditions of "key monitoring area coverage ≥ 99% and dual-source field-of-view overlap ≥ 95%" are met. Then, the camera group acquires visible light image data of the inside of the power distribution cabinet under test according to the preset resolution and frame rate, and the infrared temperature measurement array simultaneously acquires full-area infrared temperature measurement data according to the preset temperature measurement range and accuracy. After the two types of data are transmitted to the dual-source data synchronous calibration unit through the internal interface, distortion correction (visible light image) and noise filtering (infrared temperature measurement data) are performed respectively. Then, a unified spatial coordinate system is established using the pre-marked fixed marks (such as metal positioning holes and standard mounting screws) in the power distribution cabinet as calibration reference points. Through coordinate matching, a one-to-one correspondence between the pixel coordinates of the visible light image and the coordinates of the infrared temperature measurement points is achieved. Finally, a calibration dual-source map that simultaneously contains the visual features and temperature features of the component is formed. Step 102: Fault identification and instruction generation. In this phase, the edge computing chip of the fault identification module synchronously receives the calibration dual-source map and the environmental temperature and humidity data transmitted by the sensor device (temperature and humidity sensor). First, the calibration dual-source map is denoised to obtain a dual-source feature matrix composed of pixel visual features and temperature features. The environmental temperature and humidity data is normalized to obtain environmental feature parameters. Then, the fault identification model performs visual and temperature dual-channel feature extraction to obtain visual feature parameters and temperature feature parameters respectively. Subsequently, a visual feature vector is constructed using the visual feature parameters, and a temperature-environment feature vector is constructed using the temperature feature parameters and the environmental feature parameters. The two types of vectors are weighted and fused based on the attention mechanism to obtain a fused feature vector. The cosine similarity between this fused feature vector and the typical fault samples of the distribution cabinet in the feature database is calculated to determine the anomaly confidence. Finally, the anomaly confidence is determined according to the preset fault level range (≥0.85 for emergency fault, 0.70~0.85 for severe fault, 0.50~0.70 for environmental anomaly, <0.(50 indicates a minor fault) Generates corresponding handling instructions; Step 103: During the handling action execution phase, after receiving the handling instructions, the handling priority decision unit of the environmental control module first parses the instruction type and fault type, and then matches and triggers the corresponding handling action according to the preset priority: If it is an emergency fault handling instruction, immediately execute the fault circuit isolation through the circuit breaker linkage interface and start the audible and visual alarm; if it is a serious fault handling instruction, start the corresponding environmental control component according to the fault type (start the intelligent fan for overheating fault, start the dehumidifier for condensation fault) and simultaneously start the audible and visual alarm; if it is an environmental abnormality handling instruction, directly start the corresponding environmental control component to optimize the cabinet environment; if it is a minor fault handling instruction, start the audible and visual alarm and control the sensor device to increase the acquisition frequency for encrypted monitoring. At the same time, the local storage module of the low-power operation and maintenance module stores the full-process data (acquisition data, calibration spectrum, identification results, handling instructions) according to the "data type + timestamp" rule. The wireless communication module uploads the data to the remote operation and maintenance platform through a combination of "regular batch upload + emergency fault upload". The low-power management unit switches to the low-power operation and maintenance mode when the device is not in the monitoring stage to balance energy consumption and monitoring needs. .

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0161] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0162] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power distribution cabinet monitoring device based on infrared-visible light dual-source coordination, characterized in that, include: The dual-source collaborative monitoring module is used to collect visible light image data and infrared temperature measurement data of the power distribution cabinet under test and perform dual-source data calibration to obtain a calibrated dual-source spectrum. The fault identification module is used to acquire ambient temperature and humidity data and typical fault samples of the power distribution cabinet, combine them with the calibrated dual-source spectrum to identify faults, and generate handling instructions based on the fault identification results. The environmental control module is used to execute associated treatment actions according to the treatment instructions.

2. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 1, characterized in that, The dual-source collaborative monitoring module includes: A camera array is used to collect visible light image data of the power distribution cabinet under test; An infrared temperature measurement array is used to collect the infrared temperature data of the power distribution cabinet under test; The dual-source data synchronization calibration unit is used to perform coordinate matching between the visible light image data and the external infrared temperature measurement data to obtain a calibrated dual-source spectrum.

3. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 2, characterized in that, The dual-source collaborative monitoring module also includes a mounting bracket; Both the camera group and the infrared temperature measurement array are mounted on the mounting bracket; The mounting bracket is used to adjust the monitoring angle of the camera group and the infrared temperature measurement array until the preset standard acquisition conditions are met. The preset standard acquisition conditions specifically require that the monitoring coverage of all key monitoring areas within the power distribution cabinet to be tested is greater than the preset standard coverage threshold, and that the field of view overlap between the camera group and the infrared temperature measurement array meets the preset standard overlap threshold.

4. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 2, characterized in that, The coordinate matching process is specifically as follows: Determine the calibration reference point inside the power distribution cabinet to be tested, specifically the fixed mark pre-marked inside the power distribution cabinet to be tested; Using the calibration reference point as a reference, the pixel coordinates of the visible light image data are matched with the temperature measurement point coordinates of the external infrared temperature measurement data to obtain multiple pixel-temperature maps; By integrating multiple pixel-temperature maps, a calibrated dual-source map is obtained.

5. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 1, characterized in that, The fault identification module includes: An edge computing chip is used to receive the ambient temperature and humidity data of the power distribution cabinet to be tested, and to reduce the noise of the calibration dual-source spectrum to obtain a dual-source feature matrix. The dual-source feature matrix is ​​specifically a two-dimensional normalized matrix composed of pixel visual features and temperature features. The ambient temperature and humidity data is normalized to obtain environmental feature parameters. A feature database is used to store typical fault samples of power distribution cabinets; A fault identification model is used to identify faults using the dual-source feature matrix, the environmental feature parameters, and typical fault samples of the power distribution cabinet, and to generate anomaly confidence. The edge computing chip is also used to generate associated handling instructions based on the preset fault level range to which the anomaly confidence level belongs.

6. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 5, characterized in that, The fault identification process is specifically as follows: Visual and temperature features are extracted from the dual-source feature matrix to obtain visual feature parameters and temperature feature parameters. Visual feature vectors are constructed using the aforementioned visual feature parameters; A temperature-environment feature vector is constructed using the temperature feature parameters and the environmental feature parameters. The visual feature vector and the temperature-environment feature vector are weighted and fused to obtain a fused feature vector; Calculate the cosine similarity between the fused feature vector and the typical fault samples of the distribution cabinet to determine the anomaly confidence level.

7. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 1, characterized in that, The environmental control module includes a priority decision-making unit, a circuit breaker linkage interface, an audible and visual alarm, environmental control components, and sensor devices. The disposal priority decision unit is used to match the disposal action corresponding to the disposal instruction type according to the instruction type of the disposal instruction and execute it. The circuit breaker linkage interface is used to perform fault circuit isolation; The audible and visual alarm is used to provide audible and visual alarms. The environmental control components include a smart fan and a dehumidifier; The intelligent fan is used to dissipate heat when the fault type is overheating fault. The dehumidifier is used to dehumidify when the fault type is condensation fault. The sensor device includes a temperature and humidity sensor and a condensation sensor; The temperature and humidity sensor is used to collect the ambient temperature and humidity data in the power distribution cabinet under test in real time and transmit it to the fault identification module. The condensation sensor is used to collect condensation data inside the power distribution cabinet to be tested.

8. The distribution cabinet monitoring device based on infrared-visible light dual-source coordination according to claim 7, characterized in that, The process of matching and executing the corresponding action based on the instruction type of the disposal instruction is as follows: When the handling instruction is an emergency fault handling instruction, the fault circuit is isolated through the circuit breaker linkage interface, and the audible and visual alarm is activated. When the handling instruction is a serious fault handling instruction, the corresponding environmental control component is activated according to the fault type, and the audible and visual alarm is activated. When the handling instruction is an environmental anomaly handling instruction, the environmental control component is activated; When the handling instruction is a minor fault handling instruction, the audible and visual alarm is activated, and the sensor equipment is controlled to monitor the environmental status of the power distribution cabinet under test.

9. The distribution cabinet monitoring device based on infrared-visible dual-source coordination according to any one of claims 1-8, characterized in that, It also includes a low-power operation and maintenance module; The low-power operation and maintenance module includes a local storage module, a wireless communication module, and a low-power management unit; The local storage module is connected to the dual-source collaborative monitoring module and the fault identification module respectively, and is used for data classification and storage; The wireless communication module is connected to the local storage module and is used to upload the classified data stored in the local storage module to the remote operation and maintenance platform; The low-power management unit is used to switch the monitoring device to a low-power operation and maintenance mode when the monitoring device is not in the monitoring phase.

10. A monitoring method applied to the distribution cabinet monitoring device based on infrared-visible dual-source coordination as described in any one of claims 1-9, characterized in that, include: Visible light image data and infrared temperature measurement data of the power distribution cabinet under test are collected and dual-source data calibration is performed to obtain the calibrated dual-source spectrum; Acquire ambient temperature and humidity data and typical fault samples of power distribution cabinets, combine them with the calibrated dual-source spectrum to identify faults, and generate handling instructions based on the fault identification results; Execute the associated handling action according to the handling instruction.

Images