A substation switch cabinet abnormality identification method and system

Abstract

This invention relates to the field of substation anomaly identification technology, and discloses a method and system for anomaly identification in substation switchgear. The method includes: acquiring real-time temperature data from multiple monitoring points inside the switchgear and uploading it to a big data analysis platform via wired or wireless communication to achieve continuous monitoring of equipment operating status. Based on the real-time temperature data, the highest temperature of the monitoring area, the highest temperature difference between adjacent monitoring periods, and temperature abrupt changes are calculated. When these parameters exceed preset thresholds, corresponding alarm strategies are triggered in a timely manner. By generating infrared thermal images, extracting areas of temperature distribution difference, and marking high-risk monitoring points, the accuracy and timeliness of anomaly identification are further improved. Simultaneously, a temperature trend prediction model is used to predict future temperature trends, generate early warning signals in advance, and activate a preventative alarm mechanism, providing strong protection for the safe and stable operation of the power system.

Description

Technical Field

[0001] This invention relates to the field of substation equipment anomaly identification technology, and more specifically, to a method and system for anomaly identification of substation switchgear. Background Technology

[0002] For the identification of abnormal conditions and fault prevention of substation switchgear, the traditional methods commonly used in the industry mainly include periodic manual inspections and basic temperature monitoring technology.

[0003] However, these conventional methods, which have been used for many years, have revealed many obvious defects and shortcomings in practical applications.

[0004] First, manual inspection requires professional personnel to regularly go to the site to conduct visual inspections, infrared temperature measurements, and other operations. The whole process not only requires a lot of human resources and consumes a lot of working time, but also, due to the limitations of inspection frequency and time, it is impossible to continuously monitor the operating status of the equipment, and it is easy to miss some sudden or intermittent abnormal signs.

[0005] Secondly, the basic temperature monitoring methods currently widely used, such as the arrangement of local temperature measurement points or simple infrared temperature measurement, are often limited to specific locations. The temperature data obtained lacks comprehensiveness and systematicity, making it difficult to truly and accurately reflect the thermal state distribution and dynamic change trends of various key parts inside the switchgear.

[0006] More importantly, with the increasing complexity of the operating environment of modern substations, factors such as frequent equipment load fluctuations and drastic changes in ambient temperature and humidity have brought severe challenges to traditional monitoring methods. These methods, which are based on manual and experience-based judgment, are inadequate when faced with complex operating conditions, and their reliability and timeliness are difficult to meet the high standards of equipment operation safety and stability requirements of the current power system. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of the prior art, the present invention provides a method for identifying anomalies in substation switchgear, which solves the problems of low monitoring efficiency, data bias, and insufficient ability to cope with complex operating conditions in traditional monitoring methods. In addition, the present invention also provides a substation switchgear anomaly identification system.

[0008] Technical solution: According to a first aspect of the present invention, a method for identifying anomalies in substation switchgear is provided, the method comprising:

[0009] A miniature infrared thermal imaging device is installed on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device acquires a complete frame of infrared thermal images of the inside of the switchgear according to a preset scanning frequency. Spatial mapping and ambient temperature compensation processing are performed on each frame of infrared thermal images.

[0010] The acquired real-time temperature data is structured using a physical model-driven environmental reflection component stripping method, then stored, assigned a unique serial number, and uploaded to the big data analysis platform.

[0011] The big data analysis platform calculates the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature abrupt change value based on the uploaded real-time temperature data. When the highest temperature of the monitoring area exceeds a preset high temperature threshold, or the difference between the highest temperatures of adjacent monitoring periods exceeds a preset temperature rise threshold, or the temperature abrupt change value exceeds a preset abrupt change threshold, an equipment abnormality alarm signal is generated. The highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring period. The difference between the highest temperatures of adjacent monitoring periods refers to the numerical difference between the highest temperature of the current period and the highest temperature of the previous period. The temperature abrupt change value refers to the temperature difference corresponding to the rate of temperature change exceeding a preset rate of change threshold per unit time. The preset rate of change threshold is a critical rate of change used to determine whether a sudden temperature rise has occurred.

[0012] Collect alarm information records of the same type of switchgear at all levels within the most recent monitoring periods, and calculate and execute the judgment of first-level alarm, second-level alarm and third-level alarm respectively.

[0013] Furthermore, including:

[0014] The process of performing spatial mapping and ambient temperature compensation on each frame of infrared thermal image includes:

[0015] Based on the pre-defined spatial coordinate mapping relationship, the position of each pixel in the infrared thermal image is mapped to the specific physical location of the moving contact, stationary contact and their connection structure inside the switch cabinet;

[0016] Non-uniformity correction and ambient temperature drift compensation are performed on the original infrared radiation signal to eliminate measurement deviations caused by air convection, metal reflection, or background heat sources inside the cabinet. Ambient temperature drift compensation refers to dynamically correcting the infrared temperature measurement results based on the reference temperature of the environment inside the cabinet to eliminate the impact of environmental fluctuations on measurement accuracy.

[0017] Based on the temperature distribution of the current pixel and its neighborhood, the representative temperature value of the specific physical location corresponding to the pixel is extracted and marked as a valid monitoring point; the valid monitoring point refers to the monitoring location located at the moving contact, stationary contact or key connection part and whose temperature data has been verified by noise filtering and jump rejection.

[0018] The temperature values ​​of all valid monitoring points, along with their spatial coordinates and acquisition timestamps, are encapsulated to form a structured real-time temperature dataset for subsequent uploading and analysis. The structured real-time temperature dataset refers to a standardized set of data units containing triples of temperature value, spatial coordinates, and timestamp.

[0019] Furthermore, including:

[0020] The physical model-driven environmental reflection component stripping method includes:

[0021] The system receives real-time temperature data from a miniature infrared thermal imaging device and converts it into a raw temperature matrix. It then performs neighborhood consistency checks on the temperature value of each monitoring point. If the temperature difference between a monitoring point and its neighboring monitoring points exceeds a preset jump tolerance threshold, the point is marked as a suspicious anomaly.

[0022] Determine whether the number of suspicious anomalies in the current scanning cycle exceeds the preset upper limit of the total number of monitoring points. If it does not exceed the limit, perform environmental temperature drift compensation on all non-suspicious anomalies, correct the infrared temperature measurement deviation based on the reading of the reference temperature sensor in the cabinet, and generate a pre-cleaned temperature dataset. Otherwise, discard all temperature data in the current scanning cycle, record the invalid data event log, and trigger the self-test process of the miniature infrared thermal imaging device to check for lens contamination, power supply fluctuations, or communication interruption. After completing the self-test, wait for the next scanning cycle to re-acquire data.

[0023] To centralize the temperature data after initial cleaning, each temperature value is bound to a collection timestamp and a spatial location code. The timestamp comes from the high-precision real-time clock inside the device, and the spatial location code is generated based on the pre-calibrated mapping relationship between infrared image pixels and the internal structure of the switch cabinet.

[0024] Temperature data with spatiotemporal tags is encapsulated into structured temperature data packets and written to the tail of the first-in-first-out queue of the local non-volatile storage module. After assigning a unique sequence number, the local cache management program is started.

[0025] The system checks the current communication link status. If the link is stable, it prepares to upload data in batches. If the link is interrupted, it retains the data packets until the link is restored.

[0026] Furthermore, including:

[0027] The process of performing environmental temperature drift compensation on all non-suspicious anomalies includes:

[0028] At least one high-stability ambient reference temperature sensor is fixedly installed in the non-heating area inside the switch cabinet. The surface of the sensor is coated with a low emissivity material and shielded from direct thermal radiation, and is used to collect the background ambient temperature inside the cabinet in real time.

[0029] Simultaneously acquire the raw infrared radiation intensity matrix output by the miniature infrared thermal imaging device and the reading of the environmental reference temperature sensor, and align the timestamps of both to the same scanning cycle;

[0030] Based on the infrared thermal imaging physical model, the theoretical radiation intensity of an ideal blackbody at the current ambient temperature is calculated and compared with the actual infrared radiation intensity to derive the proportion of environmental reflection components.

[0031] For each pixel in the original infrared radiation intensity matrix, the spurious radiation value contributed by the environmental reflection component is subtracted, and the true component that comes only from the thermal radiation of the measured contact itself is retained.

[0032] The corrected infrared radiation intensity is converted into surface temperature values ​​to generate a temperature matrix after environmental drift compensation.

[0033] Furthermore, including:

[0034] The big data analysis platform calculates the highest temperature in the monitored area, the difference in highest temperature between adjacent monitoring periods, and temperature abrupt changes based on the uploaded real-time temperature data, including:

[0035] Obtain the real-time temperature values ​​of each monitoring point at the current moment, and determine the maximum value as the highest temperature;

[0036] Obtain the historical highest temperature value of each monitoring point in the previous monitoring period, and calculate the difference between the current highest temperature and the historical highest temperature as the highest temperature difference between adjacent monitoring periods;

[0037] Time series analysis is performed on the temperature data within the current monitoring period to calculate the rate of temperature change per unit time. When the rate of change is greater than a preset rate of change threshold, the temperature difference corresponding to that rate of change is determined as the temperature abrupt change value.

[0038] Furthermore, including:

[0039] The big data analysis platform calculates temperature fluctuation values ​​based on the uploaded real-time temperature data, including:

[0040] Structured temperature data of the same monitoring point in the most recent consecutive monitoring periods are extracted from the local cache to construct the temperature time series of that point;

[0041] Using a fixed time interval as the step size, the ratio of the temperature difference to the time difference between two adjacent points is calculated to obtain the instantaneous temperature change rate per unit time.

[0042] The calculated rate of change at each moment is compared with a preset rate of change threshold. If any rate of change exceeds the threshold, the actual temperature difference within the corresponding time period is recorded as a candidate mutation value. The largest candidate mutation value among all monitoring points is selected as the temperature mutation value for this monitoring cycle for subsequent alarm judgment.

[0043] Furthermore, including:

[0044] The statistical analysis of alarm information records at all levels for the same type of switchgear within the most recent monitoring periods, and the calculation and execution of judgments for Level 1, Level 2, and Level 3 alarms, includes:

[0045] Collect historical temperature data of the same type of switchgear under rated load, full load and slight overload conditions during continuous operation, and extract the highest temperature distribution range of the moving contact and stationary contact areas under thermal stability conditions.

[0046] Determine whether the distribution range covers at least thirty complete monitoring cycles. If it does, calculate the high percentile value of the highest temperature distribution range and round it up by a fixed temperature rise step as the preset high temperature threshold. Ensure that this threshold is higher than the upper limit of the normal thermal steady state of the equipment but lower than the thermal failure critical point of the insulation material. Otherwise, continue to collect data until the cycle requirement is met.

[0047] Based on the sequence of highest temperature differences between adjacent monitoring periods in historical temperature data, the maximum temperature rise difference under normal load fluctuation conditions is statistically analyzed.

[0048] Determine if there are any records of non-abnormal load change events caused by external scheduling. If so, remove the temperature rise difference for the corresponding time period, recalculate the maximum value, multiply the corrected maximum temperature rise difference by a safety factor greater than one and round it down to the nearest integer as the preset temperature rise threshold. This threshold is used to distinguish between abnormal temperature rise caused by reasonable load changes and contact resistance degradation. Otherwise, directly multiply the corrected maximum temperature rise difference by a safety factor greater than one and round it down to the nearest integer as the preset temperature rise threshold. This threshold is used to distinguish between abnormal temperature rise caused by reasonable load changes and contact resistance degradation.

[0049] Retrieve all confirmed temperature mutation events from the historical fault case database and extract the temperature change rate from normal operation to the last effective monitoring period before the fault occurred.

[0050] Determine whether each event has complete pre-monitoring data and no communication interruption. If it does, calculate the minimum temperature change rate at the onset of the mutation in all valid events, set the minimum value as the preset change rate threshold, and calculate the corresponding temperature difference value in combination with the current system monitoring cycle time interval as the preset mutation threshold. Otherwise, remove the event and continue to process the next case.

[0051] The minimum temperature change rate at the onset of mutation in all valid events is statistically analyzed. This minimum value is set as the preset change rate threshold, and the corresponding temperature difference is calculated by combining it with the current system monitoring cycle time interval, which is then used as the preset mutation threshold.

[0052] The preset high temperature threshold, preset temperature rise threshold, and preset sudden change threshold are written into the alarm rule engine of the big data analysis platform and bound to the switch cabinet model, rated current, and ambient temperature compensation coefficient.

[0053] Once the newly connected switchgear completes its initial identification, the system automatically matches its model and operating parameters, loads the corresponding high temperature threshold, preset temperature rise threshold and preset sudden change threshold parameter set, and performs real-time alarm judgment.

[0054] After the system is put into operation, a threshold verification process is initiated once every quarter. The historical statistical distribution is updated using newly added operational data to determine whether the false alarm rate in the current alarm record exceeds the preset false alarm limit or the missed alarm rate exceeds the preset missed alarm limit.

[0055] Furthermore, including:

[0056] The method also includes: closed-loop optimization of the baseline heatmap, alarm threshold, and prediction model based on operation and maintenance feedback and operational data, specifically including the following steps:

[0057] Step 1: Receive structured anomaly handling feedback information submitted by maintenance personnel on mobile terminals. The feedback information includes fault authenticity determination, retested temperature data, physical defect location, and execution status of handling measures.

[0058] Step 2: Determine whether this abnormal event is a real fault. If the feedback information confirms that there are physical defects such as contact surface burning, loose bolts or insulation deterioration, proceed to step 3; otherwise, proceed to step 6.

[0059] Step 3: Detect the load status and environmental stability of the switchgear during continuous operation after the anomaly is handled. If the load rate is within the rated range and the fluctuation of the ambient temperature inside the cabinet is lower than the stability threshold, proceed to step 4; otherwise, proceed to step 6.

[0060] Step 4: After collecting and processing infrared thermal image data for multiple consecutive monitoring cycles, perform a difference comparison between each frame of thermal image and the current baseline thermal image, and calculate the area ratio of the temperature distribution difference region.

[0061] Step 5: Determine whether the area ratio of temperature distribution difference regions in all acquisition frames does not exceed the preset ratio. If so, use the heat map data of this period to reconstruct the baseline heat map and replace the original baseline heat map. At the same time, store the complete temperature evolution sequence of this abnormal event, the spatial location of high-risk monitoring points and the final fault type into the fault mode library, and trigger the incremental training process of the temperature trend prediction model.

[0062] Step 6: Analyze the alarm records of the same type of switchgear at all levels in the most recent monitoring periods, and calculate the false alarm rate and false negative rate of Level 1 alarm, Level 2 alarm and Level 3 alarm respectively.

[0063] On the other hand, the present invention also provides a substation switchgear anomaly identification system, the system comprising:

[0064] The data acquisition and processing module is used to install a miniature infrared thermal imaging device on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device acquires a complete frame of infrared thermal images of the inside of the switchgear according to a preset scanning frequency; and performs spatial mapping and ambient temperature compensation processing on each frame of infrared thermal images.

[0065] The structured processing module is used to process the acquired real-time temperature data using a physical model-driven environmental reflection component stripping method, store the data, assign a unique serial number, and upload it to the big data analysis platform.

[0066] The equipment anomaly alarm module is used by the big data analysis platform to calculate the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature mutation value based on the uploaded real-time temperature data. When the highest temperature of the monitoring area exceeds a preset high temperature threshold, or the difference between the highest temperatures of adjacent monitoring periods exceeds a preset temperature rise threshold, or the temperature mutation value exceeds a preset mutation threshold, an equipment anomaly alarm signal is generated. The highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring period. The difference between the highest temperatures of adjacent monitoring periods refers to the numerical difference between the highest temperature of the current period and the highest temperature of the previous period. The temperature mutation value refers to the temperature difference corresponding to the rate of temperature change exceeding a preset rate of change threshold per unit time. The preset rate of change threshold is a critical rate of change used to determine whether a sudden temperature rise has occurred.

[0067] The alarm execution module is used to collect alarm information records of the same type of switchgear at all levels within the most recent monitoring periods, and to calculate and execute the judgment of level 1 alarm, level 2 alarm and level 3 alarm respectively.

[0068] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0069] This application proposes a method for anomaly identification in substation switchgear. By installing a miniature infrared thermal imaging device on the switchgear door, it achieves non-contact real-time temperature monitoring of the moving and stationary contacts inside the switchgear, effectively solving the problem of data partiality in traditional monitoring methods. This method can acquire real-time temperature data from multiple monitoring points inside the switchgear and upload it to a big data analysis platform via wired or wireless communication, enabling continuous monitoring of equipment operating status. In the big data analysis platform, based on the real-time temperature data, the highest temperature of the monitoring area, the highest temperature difference between adjacent monitoring periods, and temperature abrupt changes are calculated. When these parameters exceed preset thresholds, an anomaly alarm signal can be generated promptly, triggering corresponding alarm strategies, thereby improving monitoring efficiency and the ability to handle complex operating conditions. Furthermore, this method further enhances the accuracy and timeliness of anomaly identification by generating infrared thermal images, extracting areas of temperature distribution difference, and marking high-risk monitoring points. Simultaneously, by using a temperature trend prediction model to predict future temperature trends, early warning signals are generated in advance, and a preventative alarm mechanism is activated, providing strong support for the safe and stable operation of the power system. Attached Figure Description

[0070] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as limiting the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 A flowchart illustrating a method for identifying anomalies in a substation switchgear, as provided in one embodiment of this application;

[0072] Figure 2 An internal structural diagram of an electronic device for a substation switchgear anomaly identification method provided in one embodiment of this application;

[0073] Figure 3 This is an example diagram of the first interface of a substation switchgear anomaly identification system provided in one embodiment of this application;

[0074] Figure 4 This is an example diagram of the second interface of a substation switchgear anomaly identification system provided in one embodiment of this application;

[0075] Figure 5 This is an example diagram of the third interface of a substation switchgear anomaly identification system provided in one embodiment of this application;

[0076] Figure 6 This is an example diagram of the fourth interface of a substation switchgear anomaly identification system provided in one embodiment of this application. Detailed Implementation

[0077] 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, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0078] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0079] Example 1: Refer to Figure 1 This is the first embodiment of the present application, which provides a method for identifying anomalies in substation switchgear. Figure 1 The flowchart of the substation switchgear anomaly identification method of this embodiment is shown, including:

[0080] S101, a miniature infrared thermal imaging device is installed on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device is used for non-contact real-time temperature monitoring of the moving and stationary contacts inside the switchgear.

[0081] It should be noted that a miniature infrared thermal imaging device is installed on the cabinet door of the substation switchgear. This device performs non-contact real-time temperature monitoring of the moving and stationary contacts inside the switchgear.

[0082] Furthermore, after installation, the miniature infrared thermal imaging device continuously performs periodic thermal imaging scans of key areas inside the switchgear at a preset scanning frequency. During each scan, the device acquires complete temperature distribution data covering the moving and stationary contact areas, generating an infrared thermal image containing temperature values ​​from multiple monitoring points. By performing pixel-level temperature analysis on the infrared thermal image, the system can accurately extract the real-time temperature value corresponding to each monitoring point. Based on the spatial coordinate mapping relationship, the temperature data is mapped one-to-one with the specific physical location inside the switchgear, thereby achieving continuous, interference-free, and high-precision monitoring of the thermal state of the moving and stationary contacts.

[0083] Among them, the miniature infrared thermal imaging device refers to a miniaturized sensing device installed on the inside of the switch cabinet door, which has the functions of infrared radiation detection and thermal imaging. The moving contact refers to a conductive component that moves during the switch operation and comes into contact with or separates from the stationary contact. The stationary contact refers to a conductive component that is fixed inside the switch cabinet and works with the moving contact to complete the circuit switching. Non-contact real-time temperature monitoring refers to the process of continuously acquiring the surface temperature data of the component being measured through infrared radiation sensing without physical contact with the component being measured.

[0084] In this embodiment of the application, the miniature infrared thermal imaging device periodically scans the interior of the switchgear according to a preset scanning frequency, and the method further includes:

[0085] After each scan is completed, an infrared thermogram containing the temperature distribution of multiple monitoring points is generated;

[0086] The infrared thermal image is compared with a baseline thermal image under historical normal conditions to extract areas of temperature distribution difference.

[0087] If the area of ​​temperature distribution difference exceeds a preset proportion, the monitoring point corresponding to that area will be listed as a high-risk monitoring point and marked as a priority monitoring level in the big data analysis platform.

[0088] Specifically, the miniature infrared thermal imaging device can be controlled to periodically scan the inside of the switch cabinet according to the preset scanning frequency. After each scan, an infrared thermal image containing the temperature distribution of multiple monitoring points is generated. The infrared thermal image records the real-time temperature values ​​of each position in the moving and stationary contact areas in a two-dimensional temperature matrix.

[0089] Furthermore, the currently generated infrared thermal image is compared pixel by pixel with the baseline thermal image stored under historical normal conditions, the temperature deviation at each corresponding position is calculated, and the continuous area where the temperature deviation exceeds the preset temperature difference tolerance threshold is extracted as the temperature distribution difference area.

[0090] Furthermore, if the area of ​​the temperature distribution difference region accounts for more than a preset proportion of the total area of ​​the entire monitoring area, the monitoring point corresponding to the region in the spatial coordinate system is marked as a high-risk monitoring point, and a priority monitoring level is assigned to it in the big data analysis platform, so that it can obtain higher weight in subsequent temperature change value calculation and alarm judgment.

[0091] Among them, the preset scanning frequency refers to the time interval at which the miniature infrared thermal imaging device performs a complete scan every 30 seconds to 5 minutes; the infrared thermal image refers to the two-dimensional image data output by the miniature infrared thermal imaging device that reflects the surface temperature distribution of each monitoring point inside the switchgear; the reference thermal image refers to the infrared thermal image collected by the switchgear under historical normal operating conditions and verified to be without abnormalities; the temperature distribution difference area refers to the area in the current infrared thermal image whose temperature deviation exceeds the preset temperature difference tolerance threshold and has spatial continuity compared with the reference thermal image; the preset ratio refers to the area proportion threshold used to determine whether it constitutes a significant abnormality; the high-risk monitoring point refers to the specific monitoring location that is marked by the system as having potential fault risk due to abnormal temperature distribution; and the priority monitoring level refers to the data processing and alarm response priority set for high-risk monitoring points in the big data analysis platform, which is higher than that for ordinary monitoring points.

[0092] Based on the above embodiments, for the technical solution of installing a miniature infrared thermal imaging device on the cabinet door of a substation switchgear to achieve non-contact real-time temperature monitoring of the internal moving and stationary contacts, the following steps are further taken to achieve precise deployment and functional coordination of the device:

[0093] A1. Fix the miniature infrared thermal imaging device inside the switch cabinet door, directly facing the area of ​​the moving and stationary contacts. If the field of view cannot completely cover all contact surfaces and surrounding connection structures, adjust the installation height or lateral offset until the entire target area is included in the imaging range, and then proceed to step A2.

[0094] A2. Use a non-metallic insulating bracket to rigidly fix the miniature infrared thermal imaging device. Determine whether the bracket material has anti-corona and low dielectric constant characteristics. If the requirements are met, proceed to step A3. Otherwise, replace the bracket material and repeat step A2.

[0095] A3. Calibrate the optical axis of the miniature infrared thermal imaging device so that it is perpendicular to the plane where the moving and stationary contacts are located. After calibration, lock the device posture. If vibration or temperature difference deformation risk is detected in the cabinet door, install a mechanical limiting structure to suppress displacement and proceed to step A4.

[0096] A4. Install a high-temperature resistant infrared transmission window in front of the lens of the miniature infrared thermal imaging device. Determine whether the window has dustproof, anti-condensation and anti-electromagnetic shielding functions. If all functions meet the standards, proceed to step A5. Otherwise, replace the window component and return to step A4.

[0097] A5. Configure the scanning parameters of the miniature infrared thermal imaging device, set the spatial resolution so that the actual physical size corresponding to each pixel does not exceed the preset upper limit, the temperature resolution is not lower than the preset lower limit, and the scanning frequency is set to complete one full frame acquisition at a fixed period. If the parameter configuration causes the power consumption to exceed the device's power supply capacity, reduce the scanning frequency and re-verify the temperature rise capture capability. After meeting the requirements, proceed to step A6.

[0098] A6. Establish the spatial mapping relationship between the pixel coordinates of the infrared thermal image and the internal structure of the switch cabinet. Calibrate the moving contact, stationary contact and their connection points one by one. If a certain monitoring point does not have a clear corresponding pixel in the thermal image, supplement the auxiliary view or adjust the device angle and recalibrate. After all calibrations are completed, store the mapping relationship in the local storage module and proceed to step A7.

[0099] A7, during the system initialization phase, initiates the baseline thermal image acquisition process, controls the miniature infrared thermal imaging device to run continuously for no less than one complete day and night cycle, and monitors the switch cabinet load rate and ambient temperature stability in real time.

[0100] A8. Determine whether the current load rate is in the high range of the rated value and the ambient temperature fluctuation is less than the preset stable threshold. If the conditions are met, use the infrared thermal image collected during this period to construct the baseline thermal image and proceed to step A9. Otherwise, continue to wait for a collection window that meets the conditions.

[0101] A9 performs a consistency check on the collected multi-frame normal state heatmaps, removes data frames with abnormal jumps or communication interference, and generates a unique baseline heatmap by averaging and fusing the remaining frames, which is then uploaded to the big data analysis platform to end the initialization process.

[0102] Therefore, through the above closed-loop installation, calibration and standardization process, the miniature infrared thermal imaging device can achieve high-precision, blind-spot-free, and location-based non-contact temperature monitoring in complex electromagnetic and thermal environments, providing a reliable data foundation for subsequent anomaly identification.

[0103] Building upon the aforementioned embodiments, there are also situations where the substation switchgear operates in environments with strong electromagnetic interference, mechanical shocks caused by frequent operations, or drastic diurnal temperature variations. In such cases, the cabinet door structure may experience minor deformations or periodic vibrations due to long-term uneven stress or thermal expansion and contraction, leading to a shift in the installation posture of the miniature infrared thermal imaging device and affecting the spatial mapping accuracy of the infrared thermal image and the accuracy of temperature monitoring. For example, in switchgear deployed in coastal areas with high humidity and high salinity, the metal material of the cabinet is susceptible to corrosion stress. During peak load periods, the thermal effect of current exacerbates local expansion of the cabinet door, causing the device's optical axis to deviate from its original calibration position. Similarly, in hub substations with frequent switching operations, mechanical vibrations caused by the actions of adjacent equipment are transmitted through the cabinet to the cabinet door, causing high-frequency micro-amplitude jitter in the infrared thermal imaging device, resulting in misalignment of consecutive inter-frame images.

[0104] Meanwhile, to ensure that the miniature infrared thermal imaging device maintains a stable imaging benchmark under complex working conditions and to avoid measurement errors caused by cabinet door displacement, this solution implements targeted deployment and risk response for the mechanical limiting structure through the following steps:

[0105] a1. Real-time acquisition of triaxial accelerometer data and temperature gradient sensor data near the installation point on the inside of the cabinet door, constructing a dynamic state feature vector of the cabinet door. When the acceleration amplitude continuously exceeds the preset vibration threshold or the temperature difference change rate of adjacent temperature measurement points exceeds the preset deformation rate threshold, it is determined that there is a risk of vibration or temperature difference deformation. Relevant technicians can preset the relevant threshold size according to actual needs, but this invention does not limit it.

[0106] a2. Based on the risk assessment results, the mechanical limit structure deployment process is initiated. The mechanical limit structure consists of a rigid support arm, an elastic buffer pad, and a magnetic base. One end of the rigid support arm is fixed to the main frame of the switch cabinet, and the other end is connected to the back of the shell of the miniature infrared thermal imaging device through a universal joint.

[0107] a3. Adjust the length of the rigid support arm and the angle of the universal joint so that the optical axis of the miniature infrared thermal imaging device remains perpendicular to the plane where the moving contact and the stationary contact are located in the limited state, and the field of view completely covers the target monitoring area. After the attitude is locked, activate the electromagnetic locking mechanism of the magnetic base.

[0108] a4. After the limit structure is installed, perform an infrared thermal image spatial mapping recalibration. Use the calibration target point with known physical coordinates to verify whether the pixel and position correspondence is offset. If the offset exceeds the allowable tolerance, update the spatial mapping parameter table in the local storage module.

[0109] a5 records this limit event in the device health file, including the triggering cause, limit structure configuration parameters and calibration correction amount, and pushes it to the big data analysis platform for environmental disturbance factor compensation in the subsequent temperature trend prediction model.

[0110] In this embodiment, a Long Short-Term Memory (LSTM) network is used as the base network for the temperature trend prediction model. LSTM has significant advantages in time series temperature prediction. It can solve the gradient vanishing problem in traditional RNNs and better capture dependencies in long sequences. Environmental disturbances such as temperature, humidity, airflow, and air pressure are considered as influencing factors affecting temperature prediction by the LSTM network.

[0111] This embodiment senses the dynamic state of the cabinet door and actively deploys a mechanical limiting structure to effectively suppress device displacement caused by vibration or thermal deformation without interrupting monitoring, ensuring the geometric stability and temperature measurement consistency of infrared thermal imaging, and improving the robustness of anomaly identification in harsh operating environments.

[0112] Among them, the field of view refers to the spatial angle range in which the miniature infrared thermal imaging device can effectively image; the non-metallic insulating bracket refers to a fixed support structure made of non-conductive materials such as engineering plastics or ceramics, used to avoid introducing parasitic capacitance or eddy current losses; the high-temperature resistant infrared transmission window refers to an optical window that allows infrared radiation to penetrate and can work stably in a high-temperature environment for a long time; the spatial mapping relationship refers to the one-to-one coordinate relationship between the position of infrared image pixels and the internal physical structure of the switchgear; the reference thermal map refers to the standard temperature distribution map generated under normal equipment conditions for subsequent difference comparison; the high load rate range refers to the operating range in which the actual current of the switchgear is a relatively high proportion of the rated current but not overloaded; the ambient temperature stability threshold refers to the maximum allowable fluctuation range for determining whether the background temperature inside the cabinet is suitable as a reference acquisition condition.

[0113] Building upon the above embodiments, there are also situations where the internal structure of the substation switchgear is complex, the moving and stationary contact areas are obstructed, or there is a multi-layered layout. In such cases, if the miniature infrared thermal imaging device is only installed on the inside of the cabinet door facing forward, it may not be able to completely cover the field of view of all critical contact connection points. For example, in double busbar structures or handcart-type switchgear, some stationary contacts are embedded deep within the cabinet, and there are insulating partitions or metal brackets obstructing their view in front, causing the single-view infrared thermal imaging device to be unable to directly acquire the infrared radiation signal from the surface of the obstructed contact.

[0114] Meanwhile, to ensure blind-spot-free, non-contact, real-time temperature monitoring of all moving and stationary contact areas, this solution employs the following steps.

[0115] B1. Based on the internal three-dimensional structural drawings of the switch cabinet and the physical layout information of the moving and stationary contacts, identify the location and spatial coordinates of all key contacts that need to be monitored, and mark the areas where the line of sight is obstructed.

[0116] B2, For contact areas where the view is obstructed, at least one auxiliary miniature infrared thermal imaging device is added at different heights or lateral offsets on the inside of the cabinet door, so that the field of view of the main device and the auxiliary device can form complementary coverage in space.

[0117] B3. Perform independent optical axis calibration on each miniature infrared thermal imaging device to ensure that the imaging plane of each device is perpendicular to the plane of its corresponding moving and stationary contacts, and map the pixel coordinate system of each device to the same global space coordinate system.

[0118] B4. Establish a multi-view infrared thermal image fusion mechanism. After each scanning cycle, the infrared thermal images generated by the main device and the auxiliary device are stitched together according to the spatial coordinate mapping relationship to generate a complete temperature distribution map covering the entire moving and stationary contact area.

[0119] B5 performs a unified data cleaning, spatiotemporal labeling, and high-risk monitoring point identification process on the stitched complete temperature distribution map to ensure that subsequent temperature rise difference calculation, abrupt value analysis, and alarm judgment are based on data from the entire region.

[0120] B6, during the system initialization phase, controls all miniature infrared thermal imaging devices to synchronously acquire infrared thermal images under normal operating conditions, and generates a unique reference thermal image based on the stitching results. This reference thermal image contains the temperature distribution characteristics of all contact areas.

[0121] Therefore, this solution solves the monitoring blind spot problem caused by structural obstruction by using a multi-view deployment and thermal image fusion strategy, ensuring that both moving and stationary contacts can be effectively included in the real-time temperature monitoring range, regardless of whether they are located in the shallow or deep layers of the cabinet, thus significantly improving the completeness and reliability of anomaly identification.

[0122] Based on the above embodiments, there are also cases where the internal structure of the substation switchgear adopts a multi-layer layout or a handcart-type push-pull mechanism. In this case, some stationary contacts are embedded in the depth area of ​​the cabinet, and are blocked by insulating partitions, metal brackets, or adjacent feeder units in front. This causes a single miniature infrared thermal imaging device installed on the inside of the cabinet door facing forward to be unable to obtain the complete infrared radiation signal of the blocked contact area. For example, in a handcart-type cabinet with dual busbars operating in sections, the lower row of stationary contacts is located at the rear of the bottom of the cabinet, and is completely blocked in front by the grounding switch operating mechanism. The field of view of the main infrared thermal imaging device can only cover the upper row of contacts, resulting in the lower row of contacts being in a monitoring blind zone.

[0123] Meanwhile, to ensure that all moving and stationary contacts, regardless of their location within the shallow or deep layers of the cabinet, are included in the non-contact real-time temperature monitoring range, this solution determines the number of auxiliary miniature infrared thermal imaging devices and their spatial deployment logic based on obstruction geometry and field-of-view coverage requirements. This is achieved through the following steps:

[0124] b1. Extract the spatial coordinates of all contacts to be monitored based on the three-dimensional structural drawings of the switch cabinet, and identify whether each contact is within the direct field of view of the main micro infrared thermal imaging device. If there is an opaque structure between a contact and the main device, mark it as a contact with obstructed view.

[0125] b2, perform cluster analysis on all line-of-sight obstruction contacts, group contacts that are spatially adjacent and have the same obstruction direction into the same obstruction group, and each obstruction group corresponds to an independent supplementary observation view;

[0126] b3. For each shielding group, an auxiliary miniature infrared thermal imaging device is deployed on the inside of the cabinet door, offset from the main device installation position in the vertical or horizontal direction, so that the optical axis of the auxiliary device points to the center area of ​​the shielding group and the field of view completely covers all contacts in the group.

[0127] b4. If a single auxiliary device still cannot cover all the contacts in the same shielding group, a second auxiliary miniature infrared thermal imaging device shall be added to the cabinet door area corresponding to the shielding group. The installation height or lateral offset of the second auxiliary device shall be arranged in a non-collinear manner with the first auxiliary device to ensure that the fields of view of the two devices do not overlap or lack in space.

[0128] b5 performs unified spatial coordinate system calibration on the main device and all auxiliary devices, and transforms the pixel coordinates of the infrared thermal images output by each device into a global three-dimensional coordinate system based on the overall structure of the switch cabinet.

[0129] b6. Establish multi-view thermal image stitching rules. Based on the projection relationship of each device in the global coordinate system, perform geometric correction and pixel-level fusion on multiple frames of infrared thermal images collected in the same scanning cycle to generate a complete temperature distribution map covering all moving and stationary contact areas.

[0130] Among them, the line-of-sight obstruction area refers to the contact area that cannot be directly observed by a single infrared thermal imaging device due to obstruction by metal supports, insulating partitions, or adjacent equipment inside the cabinet; complementary field-of-sight coverage refers to the spatial connection or overlap of the observation ranges of multiple miniature infrared thermal imaging devices, jointly covering the entire target area; the global spatial coordinate system refers to a unified three-dimensional coordinate system established based on the overall structure of the switch cabinet, used for spatial alignment of data collected by each device; the multi-view infrared thermal image fusion mechanism refers to the image processing process of geometrically correcting and pixel-level stitching infrared thermal images from different perspectives according to spatial mapping relationships; and the complete temperature distribution map refers to a two-dimensional temperature matrix that covers all moving and stationary contact areas after fusion, with each pixel corresponding to the actual physical location and temperature value.

[0131] It should be noted that this step provides a high spatial resolution, non-invasive source of raw data for the entire anomaly identification process. Since the device directly faces the high-risk areas of moving and stationary contacts, and does not require opening cabinets or contact with live components, it ensures both monitoring safety and the continuity and accuracy of data acquisition, laying a physical foundation for obtaining accurate real-time temperature data subsequently.

[0132] S102 acquires real-time temperature data from multiple monitoring points inside the switch cabinet using a miniature infrared thermal imaging device;

[0133] Acquiring high-precision, comprehensive temperature data is a prerequisite for anomaly identification. Only by accurately capturing the true thermal state of the moving and stationary contact areas during operation can subsequent temperature rise trend analysis, abrupt change detection, and high-risk point identification be supported. Relying solely on coarse single-frame imaging or failing to compensate for environmental interference will lead to distorted temperature data, resulting in false alarms or missed alarms.

[0134] Specifically, a combination of periodic scanning and pixel-level temperature analysis can be used. A miniature infrared thermal imaging device can collect complete frames of the inside of the switch cabinet at a preset scanning frequency, and perform spatial mapping and ambient temperature compensation processing on each frame of infrared thermal image.

[0135] Specifically, based on the pre-defined spatial coordinate mapping relationship, each pixel position in the infrared thermal image is mapped to the specific physical location of the moving contact, stationary contact and their connection structure inside the switch cabinet;

[0136] Then, the original infrared radiation signal is corrected for non-uniformity and compensated for ambient temperature drift to eliminate measurement deviations caused by air convection inside the cabinet, metal reflection, or background heat sources. For example, in a single scan, the device outputs a two-dimensional temperature matrix containing hundreds of pixels. The system determines the central area of ​​the contact surface of the stationary contact by using coordinate mapping, with the pixel in the fifteenth row and eighth column corresponding to the central area of ​​the contact surface of the stationary contact.

[0137] Next, based on the temperature distribution of the pixel and its neighborhood, the monitoring point is extracted. The monitoring point is the temperature value corresponding to the physical location of the current pixel, and is marked as a valid monitoring point.

[0138] Finally, the temperature values ​​of all valid monitoring points, along with their spatial coordinates and acquisition timestamps, are encapsulated to form a structured real-time temperature dataset for subsequent uploading and analysis.

[0139] Among them, pixel-level temperature analysis refers to the process of converting the infrared radiation intensity corresponding to each pixel in the infrared thermal image into the actual surface temperature value; spatial coordinate mapping relationship refers to the one-to-one correspondence between the pixel position of the infrared thermal image and the internal physical structure of the switch cabinet, which is established through calibration during the system initialization phase; environmental temperature drift compensation refers to the dynamic correction of the infrared temperature measurement results based on the reference temperature of the environment inside the cabinet to eliminate the impact of environmental fluctuations on measurement accuracy; effective monitoring point refers to the monitoring location located at the moving contact, stationary contact, or key connection parts, and whose temperature data has been verified through noise filtering and jump rejection; structured real-time temperature dataset refers to a standardized data unit set containing a triplet of temperature value, spatial coordinates, and timestamp.

[0140] It should be noted that this step converts infrared radiation information into structured multi-point temperature values, realizing the transformation from image perception to quantitative data. The acquired multi-point data covers the contact and its surrounding area, forming a complete thermal state snapshot, which enables accurate calculation of the maximum temperature, temperature rise difference, and abrupt change value in the subsequent big data analysis platform, avoiding missed or false judgments caused by single-point temperature measurement.

[0141] S103 uploads real-time temperature data to the big data analysis platform via wired or wireless communication.

[0142] It should be noted that real-time temperature data is uploaded to the big data analysis platform via wired or wireless communication. After the miniature infrared thermal imaging device completes the temperature acquisition of multiple monitoring points inside the switch cabinet and generates the corresponding infrared thermal images, the system immediately analyzes the infrared thermal images, extracts the temperature values ​​of each monitoring point, and encapsulates them into structured data packets. These data packets are encoded in a unified data format to ensure full compatibility with the interface protocol of the big data analysis platform.

[0143] Furthermore, the system then transmits data packets to the big data analysis platform in real time via industrial Ethernet or RS485 bus through wired communication mode, or via LoRa, NB-IoT or 5G communication network through wireless communication mode, according to the pre-configured communication method.

[0144] Furthermore, during data transmission, the system employs a data verification and retransmission mechanism to ensure data integrity and prevent temperature data loss or errors due to communication interference.

[0145] Furthermore, upon receiving the data packet, the big data analytics platform immediately decodes, aligns the timestamps, and maps the spatial location, accurately linking the real-time temperature data to the corresponding switch cabinets and monitoring points, providing continuous and reliable data input for subsequent temperature trend analysis, anomaly identification, and risk warning.

[0146] In this embodiment of the invention, when performing specific data transmission, it is considered that in the substation environment, switchgear is usually widely distributed and some areas have strong electromagnetic interference or difficult wiring. If real-time temperature data cannot be transmitted stably and completely to the big data analysis platform, it will lead to delays or even failures in anomaly identification. Therefore, it is necessary to establish a data upload mechanism with anti-interference capabilities, support for breakpoint resumption, and the ability to adaptively switch transmission modes according to network conditions.

[0147] Specifically, a structured data packet segmentation and caching strategy combined with dual-mode communication can be adopted to temporarily store the cleaned temperature data in the local storage module and dynamically select wired or wireless communication mode for uploading based on the current communication link quality.

[0148] Specifically, the structured temperature data packets with spatiotemporal tags are first compressed and encoded, and then divided into multiple fixed-length data segments. For example, a data packet containing temperature values, coordinates, and timestamps of 120 monitoring points is divided into three consecutive data segments. Next, the system checks whether the current wired communication link is connected and whether the bit error rate is below a preset threshold. If the conditions are met, the data is uploaded segment by segment via wired communication mode through industrial Ethernet or RS485 bus. If a wired link interruption or signal quality degradation is detected, the system automatically switches to wireless communication mode and sends data segments via LoRa, NB-IoT, or 5G networks. For example, if the RS485 line is temporarily disconnected due to construction near the cabinet during an upload, the system immediately activates the built-in NB-IoT module to continue transmitting the unfinished data segments. Finally, after each successful reception confirmation, the local storage module clears the uploaded data segments. If a transmission failure occurs, the data segment is retained and retransmitted with priority when communication is restored, ensuring that all temperature data ultimately arrives completely at the big data analysis platform.

[0149] Among them, structured temperature data packets refer to standardized data units composed of temperature values, spatial coordinates, and timestamps; dual-mode communication cooperation strategy means that the system supports both wired and wireless communication methods simultaneously and can automatically switch according to the link status to ensure transmission continuity; fixed-length data segments refer to transmission units after the original data packets are evenly divided into preset byte numbers; bit error rate refers to the ratio of the number of erroneous bits received per unit time to the total number of transmitted bits, used to measure the reliability of the communication link; local storage module refers to an embedded non-volatile storage unit deployed on the switch cabinet side for temporarily caching data to be transmitted.

[0150] In this embodiment of the application, before uploading the real-time temperature data to the big data analysis platform, the method further includes:

[0151] Step 1: Receive the real-time temperature data output by the miniature infrared thermal imaging device and convert it into the raw temperature matrix. Perform neighborhood consistency verification on the temperature value of each monitoring point. If the temperature difference between a monitoring point and its surrounding neighboring monitoring points exceeds the preset jump tolerance threshold, mark the point as a suspicious anomaly.

[0152] Step 2: Determine whether the number of suspicious anomalies in the current scanning cycle exceeds the preset upper limit of the total number of monitoring points. If it does not exceed the limit, proceed to Step 3; otherwise, proceed to Step 6.

[0153] Step 3: Perform ambient temperature drift compensation on all non-suspicious anomalies, correct infrared temperature measurement deviation based on the readings of the reference temperature sensor inside the cabinet, and generate a temperature dataset after preliminary cleaning.

[0154] Step 4: Bind the acquisition timestamp and spatial location code to each temperature value in the temperature data after the initial cleaning. The timestamp comes from the high-precision real-time clock inside the device, and the spatial location code is generated based on the pre-calibrated mapping relationship between infrared image pixels and the internal structure of the switch cabinet.

[0155] Step 5: Encapsulate the temperature data with spatiotemporal tags into a structured temperature data packet, write it to the tail of the first-in-first-out queue of the local non-volatile storage module, assign a unique sequence number, and then proceed to step 7.

[0156] Step 6: Discard all temperature data in the current scanning cycle, record the invalid data event log, and trigger the self-test process of the miniature infrared thermal imaging device to check for lens contamination, power supply fluctuations, or communication interruption. After completing the self-test, wait for the next scanning cycle to re-acquire data.

[0157] Step 7: Start the local cache management program, check the current communication link status. If the link is stable, prepare to upload in batches. If the link is interrupted, retain the data packets until the link is restored, and end the current processing flow.

[0158] The aforementioned ambient temperature drift compensation can be established and operated according to the following specific logic:

[0159] Step 11: Fix at least one high-stability ambient reference temperature sensor in the non-heating area inside the switch cabinet. The sensor surface is coated with a low emissivity material and shielded from direct thermal radiation, and is used to collect the background ambient temperature inside the cabinet in real time.

[0160] Step 12: Simultaneously acquire the raw infrared radiation intensity matrix output by the miniature infrared thermal imaging device and the reading of the environmental reference temperature sensor, and align the timestamps of both to the same scanning cycle.

[0161] Step 13: Based on the infrared thermal imaging physical model, calculate the theoretical radiation intensity of an ideal blackbody at the current ambient temperature, and compare it with the actual collected infrared radiation intensity to derive the proportion of the environmental reflection component.

[0162] Step 14: For each pixel in the original infrared radiation intensity matrix, subtract the false radiation value contributed by the environmental reflection component, and retain the true component that comes only from the thermal radiation of the tested contact itself.

[0163] Step 15: Convert the corrected infrared radiation intensity into surface temperature values ​​to generate a temperature matrix after environmental drift compensation.

[0164] Step 16: Bind the temperature values ​​of each monitoring point in the compensated temperature matrix with the corresponding spatial location code and collection timestamp to form a structured temperature data unit for subsequent data cleaning and uploading.

[0165] Therefore, this embodiment effectively suppresses infrared temperature measurement deviation caused by changes in the background temperature inside the cabinet by introducing a physical model-driven environmental reflection component stripping method, so that the temperature data truly reflects the thermal state of the contact parts, and significantly improves the stability and accuracy of anomaly identification under different seasons and load conditions.

[0166] It should be noted that this step establishes a data pathway between front-end sensing and back-end intelligent analysis. Through a reliable communication mechanism, the cleaned structured temperature data is transmitted in real time to a centralized platform, enabling subsequent complex calculations (such as time series analysis, threshold comparison, and trend prediction) to be executed efficiently on a high-performance server. Simultaneously, it supports multi-cabinet data fusion analysis, providing unified and timely data support for alarm generation.

[0167] S104, in the big data analysis platform, calculates the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring cycles, and the temperature change value based on real-time temperature data;

[0168] In this embodiment of the application, based on real-time temperature data, the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature abrupt change value are calculated, including:

[0169] Obtain the real-time temperature values ​​of each monitoring point at the current moment, and determine the maximum value as the highest temperature;

[0170] Obtain the historical highest temperature value of each monitoring point in the previous monitoring period, and calculate the difference between the current highest temperature and the historical highest temperature as the highest temperature difference between adjacent monitoring periods;

[0171] Time series analysis is performed on the temperature data within the current monitoring period to calculate the rate of temperature change per unit time. When the rate of change is greater than a preset rate of change threshold, the temperature difference corresponding to that rate of change is determined as the temperature abrupt change value.

[0172] Specifically, the system can obtain the real-time temperature values ​​of all monitoring points within the current monitoring period, extract the temperature with the highest value as the highest temperature of the monitoring area. For example, if the system obtains temperature data from 128 monitoring points in a single scan, it can traverse all the data and determine that the highest value is 87.3 degrees Celsius, and record it as the highest temperature of the current period.

[0173] Furthermore, the highest historical temperature value stored in the previous monitoring cycle is obtained, and the difference between it and the highest temperature in the current cycle is calculated to obtain the highest temperature difference between adjacent monitoring cycles. For example, if the highest temperature in the previous cycle was 79.1 degrees Celsius, then the current temperature rise difference is 8.2 degrees Celsius.

[0174] Furthermore, during switchgear operation, sudden contact degradation or partial discharge can cause a rapid temperature rise within a short period. Such anomalies are characterized by rapid development and short duration. Relying solely on the highest temperature or the temperature difference between adjacent cycles can easily miss early warning windows. Therefore, it is essential to accurately capture the rate of temperature change per unit time through time series analysis and identify abrupt changes exceeding normal fluctuation ranges, thereby enabling rapid response to sudden faults.

[0175] Specifically, a time series differential algorithm based on a sliding window can be used to calculate the continuous rate of change of the historical temperature values ​​of each monitoring point in the current monitoring period and the previous few periods.

[0176] Specifically, the system first extracts the structured temperature data of the same monitoring point in the most recent consecutive monitoring periods from the local cache, and constructs the temperature time series of that point; for example, for a monitoring point with location code R18C7, the system retrieves its temperature values ​​in the current period and the two previous periods to form a ternary time series.

[0177] Then, using a fixed time interval as the step size, the ratio of the temperature difference to the time difference between two adjacent points is calculated to obtain the instantaneous temperature change rate per unit time. For example, if the temperature of the monitoring point is 63 degrees Celsius in the first cycle, 71 degrees Celsius in the second cycle, and 89 degrees Celsius in the third cycle, the system calculates the change rate from the second to the third cycle as 18 degrees Celsius per unit time.

[0178] Next, the calculated rate of change at each moment is compared with a preset rate of change threshold. If any rate of change exceeds the threshold, the actual temperature difference within the corresponding time period is recorded as a candidate mutation value.

[0179] Finally, the largest candidate mutation value among all monitoring points is selected as the temperature mutation value for this monitoring cycle for subsequent alarm judgment; for example, if the system finds that the temperature difference at point R18C7 is 18 degrees Celsius and the rate of change exceeds the limit, while other points do not exceed the limit, then 18 degrees Celsius is determined as the temperature mutation value for the current cycle.

[0180] Among them, the highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring cycle; the difference between the highest temperatures of adjacent monitoring cycles refers to the numerical difference between the highest temperature of the current cycle and the highest temperature of the previous cycle; the temperature change value refers to the temperature difference corresponding to the rate of temperature change exceeding the preset rate of change threshold per unit time; the preset rate of change threshold is the critical rate of change used to determine whether a sudden temperature rise has occurred, and the unit is degrees Celsius per second.

[0181] In this embodiment of the application, when calculating the temperature mutation value, temperature data from high-risk monitoring points are preferentially used for calculation;

[0182] When generating an equipment malfunction alarm signal, if the temperature parameter of a high-risk monitoring point meets any alarm condition, the alarm will be triggered in advance and the alarm level will be upgraded by one level.

[0183] Specifically, when calculating temperature mutation values, temperature data from high-risk monitoring points can be prioritized for analysis. For example, when the system identifies an area as a high-risk monitoring point because the difference between the infrared thermal image and the baseline thermal image exceeds 10%, the calculation of temperature mutation values ​​in each subsequent monitoring cycle will be based solely on the temperature change rate of that high-risk monitoring point and its neighboring points.

[0184] Furthermore, when generating equipment abnormality alarm signals, if the temperature parameter of a high-risk monitoring point meets any alarm condition, an alarm will be triggered immediately without waiting for the end of the normal judgment cycle. For example, if the current temperature of a high-risk monitoring point is 82 degrees Celsius, although it has not reached the 90-degree Celsius high temperature threshold for the first-level alarm, the temperature rise difference between adjacent cycles is 12 degrees Celsius, which exceeds the 10-degree Celsius temperature rise threshold for the second-level alarm, and the system will immediately generate an alarm signal.

[0185] At the same time, the alarm signal level is upgraded by one level based on the original judgment. For example, a situation that should have triggered a level two alarm is directly upgraded to a level one alarm because it occurs at a high-risk monitoring point, and the corresponding level one alarm strategy is executed in conjunction with it.

[0186] Among them, high-risk monitoring points refer to monitoring locations marked by the system because the area of ​​temperature distribution difference between the infrared thermal image and the historical benchmark thermal image exceeds a preset proportion. Temperature mutation value refers to the temperature difference corresponding to the temperature change rate exceeding the preset change rate threshold per unit time. Alarm conditions refer to any one of the following situations: the highest temperature exceeds the preset high temperature threshold, the highest temperature difference between adjacent monitoring periods exceeds the preset temperature rise threshold, or the temperature mutation value exceeds the preset mutation threshold. Alarm level upgrade means adjusting the alarm level up by one level based on the original alarm level. Level 1 alarm is the highest level, and Level 3 alarm is the lowest level.

[0187] It should be noted that this step, through the extraction of three types of features, comprehensively characterizes three typical modes of thermal anomalies in switchgear: continuous overheating, gradual degradation, and sudden failure. This multi-dimensional quantitative analysis significantly improves the sensitivity and specificity of anomaly identification, providing clear and comparable numerical basis for subsequent alarm judgment, and avoiding response lag or false triggering caused by relying on a single indicator.

[0188] S105: When the highest temperature exceeds the preset high temperature threshold, or the difference between the highest temperatures in adjacent monitoring cycles exceeds the preset temperature rise threshold, or the temperature change value exceeds the preset change threshold, an equipment abnormality alarm signal is generated.

[0189] Specifically, the highest temperature calculated can be compared with the preset high temperature threshold in real time. When the highest temperature reaches or exceeds 90 degrees Celsius, the alarm condition is determined to be met.

[0190] Simultaneously, the highest temperature difference between adjacent monitoring cycles is continuously monitored. When the difference reaches or exceeds 10 degrees Celsius, another alarm condition is determined to be met. Temperature change values ​​are also analyzed simultaneously. When they reach or exceed 8 degrees Celsius, a third alarm condition is determined to be met.

[0191] If any one of the above three conditions is met, the system will immediately generate an equipment abnormality alarm signal. For example, if the highest temperature in a certain monitoring cycle is 92 degrees Celsius, an alarm signal will still be triggered even though the temperature rise difference and sudden change value are not exceeded. Similarly, if the highest temperature is 85 degrees Celsius and is not exceeded, but the temperature rise difference between adjacent cycles is 11 degrees Celsius, an alarm signal will also be generated. Furthermore, if the highest temperature and temperature rise difference are normal, but the temperature at a high-risk monitoring point suddenly rises by 9 degrees Celsius within a unit of time, and the sudden change value exceeds the threshold, the system will also generate an alarm signal.

[0192] Among them, the highest temperature refers to the maximum value of the temperature among all monitoring points in the current monitoring cycle; the preset high temperature threshold is the upper limit of temperature used to determine whether the equipment is in a dangerous high temperature state, and its value is 90 degrees Celsius; the highest temperature difference between adjacent monitoring cycles refers to the difference between the highest temperature of the current cycle and the highest temperature of the previous cycle; the preset temperature rise threshold is the maximum allowable temperature rise between adjacent cycles, and its value is 10 degrees Celsius; the temperature mutation value refers to the actual temperature difference determined by the temperature change rate exceeding 0.5 degrees Celsius per second per unit time; the preset mutation threshold is the critical temperature difference used to identify sudden temperature rises, and its value is 8 degrees Celsius; the equipment abnormality alarm signal refers to the fault warning command automatically generated by the system when any abnormal temperature indicator exceeds the corresponding threshold.

[0193] In this embodiment of the application, after generating a device malfunction alarm signal, the method further includes:

[0194] Based on the type of abnormal alarm signal from the equipment, the corresponding alarm level is determined. Specifically, when the highest temperature exceeds the preset high temperature threshold, it is determined as a Level 1 alarm; when the difference between the highest temperatures of adjacent monitoring cycles exceeds the preset temperature rise threshold, it is determined as a Level 2 alarm; and when the temperature change value exceeds the preset change threshold, it is determined as a Level 3 alarm.

[0195] Based on the alarm level, the control alarm device executes the corresponding alarm strategy, where a level 1 alarm triggers a level 1 alarm strategy, a level 2 alarm triggers a level 2 alarm strategy, and a level 3 alarm triggers a level 3 alarm strategy.

[0196] In some embodiments, after generating a device malfunction alarm signal, "controlling the alarm device to execute the corresponding alarm strategy based on the alarm level, wherein a level 1 alarm triggers a level 1 alarm strategy, a level 2 alarm triggers a level 2 alarm strategy, and a level 3 alarm triggers a level 3 alarm strategy" includes the following steps:

[0197] Step 1.1 When the system determines that it is a Level 1 alarm, the Level 1 alarm strategy is activated, including immediately sending an emergency interlock command to the substation integrated automation system, forcibly disconnecting the circuit breaker control circuit of the corresponding switchgear, setting the "thermal fault interlock" status flag in the background monitoring system, and activating the fault isolation module of the local edge computing unit to automatically cut off the load distribution logic of the feeder where the switchgear is located, so as to prevent the overheating fault from spreading to adjacent equipment.

[0198] Step 1.2: When the system determines that it is a level 2 alarm, the level 2 alarm strategy is activated, including sending a temperature rise warning command to the substation energy management system, triggering the load dynamic scheduling mechanism, automatically transferring part of the load carried by the switchgear to the backup circuit, and starting a high-frequency temperature sampling mode in the big data analysis platform to shorten the scanning cycle of the miniature infrared thermal imaging device from 5 minutes to 30 seconds, so as to improve the monitoring density of the abnormal evolution process.

[0199] Step 1.3: When the system determines that it is a Level 3 alarm, the Level 3 alarm strategy is activated, including establishing a short-term temperature trend cache for the monitoring point in the local storage module, continuously recording complete infrared thermal image data for the next three monitoring cycles, and marking the data packet as "mutation analysis priority". In the next upload window, it is pushed to the big data analysis platform first for online fine-tuning of the temperature trend prediction model and abnormal pattern identification.

[0200] Step 1.4: While executing alarm strategies at all levels, the system dynamically adjusts the subsequent monitoring logic according to the alarm level: After a Level 1 alarm is triggered, the system automatically upgrades all monitoring points in the switch cabinet to high-risk monitoring points and forcibly includes them in the calculation of all subsequent temperature change values; after a Level 2 alarm is triggered, only the monitoring points corresponding to the abnormal temperature rise area are listed as high-risk; after a Level 3 alarm is triggered, only the change point itself is prioritized for monitoring within 72 hours, forming a hierarchical closed-loop response mechanism.

[0201] Based on the above embodiments, this application implements a scientific and engineered configuration of alarm thresholds through the following steps to ensure that the alarm level classification strictly matches the actual operating characteristics of the switchgear, taking into account the basis and determination method for setting different alarm thresholds:

[0202] C1. Collect historical temperature data of the same type of switchgear under rated load, full load and slight overload conditions during continuous operation, and extract the highest temperature distribution range of the moving contact and stationary contact areas under thermal stability conditions.

[0203] C2, determine whether the distribution range covers at least thirty complete monitoring cycles. If it does, proceed to step C3; otherwise, return to step C1 to continue collecting data until the cycle requirement is met.

[0204] C3 calculates the high percentile value of the highest temperature distribution range and rounds it up by a fixed temperature rise step as the preset high temperature threshold, ensuring that the threshold is higher than the upper limit of the normal thermal steady state of the equipment but lower than the thermal failure critical point of the insulation material.

[0205] C4, based on the sequence of highest temperature differences between adjacent monitoring periods in historical temperature data, statistically analyzes the maximum temperature rise difference under normal load fluctuation conditions;

[0206] C5. Determine if there are any records of non-abnormal load change events caused by external scheduling. If so, remove the temperature rise difference for the corresponding time period and recalculate the maximum value, then proceed to step C6. Otherwise, proceed directly to step C6.

[0207] C6, multiply the corrected maximum temperature rise difference by a safety factor greater than one and round it down to get the preset temperature rise threshold, which is used to distinguish between reasonable load changes and abnormal temperature rises caused by contact resistance deterioration.

[0208] C7 retrieves all confirmed temperature mutation events from the historical fault case library and extracts the temperature change rate from normal operation to the last effective monitoring period before the fault occurred.

[0209] C8. Determine whether each event has complete pre-monitoring data and no communication interruption. If it does, proceed to step C9; otherwise, remove the event and continue processing the next case.

[0210] C9, count the minimum temperature change rate at the start of the mutation in all valid events, set the minimum value as the preset change rate threshold, and calculate the corresponding temperature difference value in combination with the current system monitoring cycle time interval, as the preset mutation threshold;

[0211] C10 writes the preset high temperature threshold, preset temperature rise threshold, and preset sudden change threshold into the alarm rule engine of the big data analysis platform, and establishes a binding relationship with the switch cabinet model, rated current, and ambient temperature compensation coefficient.

[0212] C11: After the newly connected switchgear completes its initial identification, the system automatically matches its model and operating parameters, loads the corresponding three types of threshold parameter sets, and performs real-time alarm judgment.

[0213] C12 initiates a threshold verification process every quarter after the system is put into operation, using newly added operational data to update the historical statistical distribution;

[0214] C13, determine whether the false alarm rate in the current alarm record exceeds the preset false alarm limit or the missed alarm rate exceeds the preset missed alarm limit. If either condition is met, proceed to step C14; otherwise, end the verification process.

[0215] C14 triggers the threshold recalibration mechanism, re-executes steps C1 to C9, generates updated thresholds for the three categories and writes them into the alarm rule engine, replacing the original parameters.

[0216] Therefore, by using the above closed-loop threshold setting and dynamic calibration process, we can ensure that alarm judgment is always based on the actual operating characteristics and aging state of the equipment, effectively suppressing false alarms and missed alarms, and improving the engineering practicality and long-term adaptability of anomaly identification.

[0217] Among them, thermal stability state refers to the operating stage of the switchgear where the temperature fluctuation amplitude under continuous load is less than the preset stability tolerance; high percentile value refers to the temperature value at a higher cumulative probability position in the temperature distribution, used to characterize the upper limit boundary of normal operation; fixed temperature rise step size refers to the temperature increment unit uniformly increased to ensure safety margin; safety factor refers to the proportional factor that amplifies the temperature rise difference to eliminate normal fluctuation interference; preset false alarm limit refers to the maximum proportion of false alarm events allowed to account for the total number of alarms; preset missed alarm limit refers to the maximum proportion of missed alarm events allowed to account for the total number of actual faults; alarm rule engine refers to the logic execution module that automatically calls the corresponding threshold and performs multi-condition comparison based on equipment attributes; ambient temperature compensation coefficient refers to the correction parameter used to correct the influence of the background temperature inside the cabinet on infrared temperature measurement.

[0218] It should be noted that this step performs logical judgment based on the calculation results of the previous step, realizing a leap from "data" to "decision". Since the threshold setting is derived from the thermal characteristics of the equipment and historical fault statistics, the alarm signal has clear engineering significance and can accurately reflect the risk level of the equipment, thus providing reliable triggering conditions for subsequent alarm execution and operation and maintenance linkage.

[0219] S106, in response to an equipment malfunction alarm signal, triggers a high-temperature color alarm and / or audible alarm.

[0220] In this application embodiment, while triggering the high-temperature color alarm and / or sound alarm, the method further includes:

[0221] Push equipment malfunction alarm signals to the mobile terminal devices of maintenance personnel;

[0222] Display alarm location, alarm type, real-time temperature curve, and suggested handling measures on mobile terminal devices;

[0223] If no confirmation feedback is received from maintenance personnel within the preset response time, an upgrade alarm request will be automatically sent to the regional monitoring center.

[0224] Specifically, while triggering the high temperature color alarm and sound alarm, the device abnormality alarm signal, which includes the alarm level, monitoring point spatial coordinates, trigger conditions and timestamp, can be pushed to the bound operation and maintenance personnel's mobile terminal device through an encrypted communication channel.

[0225] Furthermore, the alarm details interface automatically pops up on the mobile terminal device, displaying the alarm location as the switch cabinet number and the coordinates of the internal contact area, the alarm type as a level 1 high temperature alarm, a level 2 temperature rise alarm, or a level 3 sudden change alarm, the real-time temperature curve as a temperature change trend graph of the monitoring point over the past three monitoring cycles, and simultaneously displaying the suggested handling measures generated by the system based on the fault mode library matching. For example, in the case of a level 1 alarm, it prompts "Immediately disconnect the power and arrange infrared retesting to check whether the moving contact contact surface is burned."

[0226] Furthermore, if no confirmation operation or processing feedback is received from the maintenance personnel on the mobile terminal within the preset response time of 600 seconds, the system will automatically send an upgrade alarm request to the main control platform of the regional monitoring center. This request includes the original alarm data, non-response records and a snapshot of the current switch cabinet operating status, triggering the central-level manual intervention process.

[0227] Among them, equipment abnormality alarm signal refers to the structured alarm data packet generated by the big data analysis platform; mobile terminal device of maintenance personnel refers to the authorized and bound smartphone or industrial handheld terminal; alarm location refers to the unique identifier of the abnormal monitoring point in the substation topology and its physical coordinates inside the switchgear; alarm type refers to the first-level, second-level, or third-level alarm category determined according to the triggering conditions; real-time temperature curve refers to the temperature time series data visualization chart of the monitoring point in the current and the previous two monitoring cycles; suggested handling measures refer to the standardized handling guidelines generated by the system based on historical fault cases and equipment models; preset response time refers to the maximum allowed time interval from the alarm push time to the time when maintenance personnel are required to provide feedback, which is 600 seconds; and escalation alarm request refers to the mandatory intervention instruction initiated to a higher-level monitoring system when the local alarm is not responded to in a timely manner.

[0228] It should be noted that this step transforms the system's judgment results into on-site, perceptible warning behaviors, enabling real-time visualization and audible alerts of abnormal states. The combination of color and sound alarms can quickly attract the attention of maintenance personnel, shorten response delays, and provide a local linkage starting point for subsequent advanced functions such as pushing mobile alarms and initiating preventative measures, forming a complete closed loop.

[0229] In some embodiments, S107 can also be designed to perform closed-loop optimization of the baseline heatmap, alarm threshold, and prediction model based on operation and maintenance feedback and operational data, in order to dynamically update and optimize the identification of substation switchgear anomalies. Specific operations may include the following steps:

[0230] Specifically, step S107 includes steps D1 to D6:

[0231] Step D1: Receive structured anomaly handling feedback information submitted by maintenance personnel on a mobile terminal. The feedback information includes the determination of the authenticity of the fault, retested temperature data, physical defect location, and execution status of the handling measures.

[0232] Step D2: Determine whether this abnormal event is a real fault. If the feedback information confirms that there are physical defects such as contact surface burning, loose bolts or insulation deterioration, proceed to step D3; otherwise, proceed to step D6.

[0233] Step D3: Detect the load status and environmental stability of the current switchgear during continuous operation after the anomaly handling is completed. If the load rate is within the rated range and the fluctuation of the ambient temperature inside the cabinet is lower than the stability threshold, proceed to step D4; otherwise, proceed to step D6.

[0234] Step D4: After collecting and processing infrared thermal image data for multiple consecutive monitoring cycles, perform a difference comparison between each frame of thermal image and the current baseline thermal image, and calculate the area ratio of the temperature distribution difference region.

[0235] Step D5: Determine whether the area ratio of temperature distribution difference regions in all acquisition frames does not exceed the preset ratio. If so, use the heat map data of this period to reconstruct the baseline heat map and replace the original baseline heat map. At the same time, store the complete temperature evolution sequence of this abnormal event, the spatial location of high-risk monitoring points and the final fault type into the fault mode library, and trigger the incremental training process of the temperature trend prediction model.

[0236] In this embodiment, incremental training refers to continuing to train the model using newly arrived data, based on data that the model has already learned from, thereby updating its knowledge without having to retrain from scratch using all historical data. This method is particularly useful for handling continuously flowing data streams, such as time series forecasting, sensor data, and user behavior analysis, and can effectively save computing resources and time.

[0237] Step D6: Statistically analyze the alarm records of the same type of switchgear at all levels in the most recent monitoring periods, and calculate the false alarm rate and false negative rate of Level 1 alarm, Level 2 alarm and Level 3 alarm respectively.

[0238] Furthermore, step D6 includes three judgment scenarios: D61, D62, and D63.

[0239] D61: Determine if there are multiple Level 1 alarm events marked as non-real faults in the operation and maintenance feedback, and if the highest temperature at the time of their triggering does not accompany the temperature rise or temperature change characteristics of adjacent cycles. If so, it means that the preset high temperature threshold is too low, and start the high temperature threshold adjustment calibration process.

[0240] The specific steps involved in the high-temperature threshold adjustment calibration process include:

[0241] Filter all Level 1 alarm events marked as non-real faults by operation and maintenance feedback within the most recent monitoring periods, and extract the highest temperature value when they were triggered and the corresponding historical temperature sequence.

[0242] Determine whether any of these events involve a situation where the highest temperature is not accompanied by an excessive temperature rise difference between adjacent cycles and there are no characteristics of a sudden temperature change.

[0243] If such events exist and their number exceeds the preset false alarm sample lower limit, then calculate the distribution lower limit of the highest temperature value of all such events;

[0244] The current preset high temperature threshold is compared with the distribution lower limit. If the current threshold is lower than the distribution lower limit, the high temperature threshold is updated to the value of the distribution lower limit rounded up.

[0245] The updated high temperature threshold is written into the alarm rule engine and bound to the corresponding switch cabinet model and environmental compensation parameter set;

[0246] Record the calibration log for this process, including the triggering reason, number of samples, original threshold and new threshold, for subsequent auditing and model backtracking.

[0247] D62: Determine if there are multiple secondary alarms that were not triggered but were later confirmed as contact resistance deterioration, and if the highest temperature difference between adjacent monitoring cycles in their historical data shows a continuous and slow upward trend. If so, it means that the preset temperature rise threshold is too high, and the temperature rise threshold reduction calibration process is initiated.

[0248] The specific steps in the temperature rise threshold reduction calibration process include:

[0249] Retrieve all historical cases in the fault mode library that were confirmed as contact resistance deterioration but did not trigger a level 2 alarm, and extract the sequence of the highest temperature difference between adjacent monitoring cycles during their operation.

[0250] For each case, the temperature rise sequence is trend-fitted to identify whether there are situations where the temperature rise difference steadily increases for multiple consecutive cycles but never reaches the current preset temperature rise threshold.

[0251] If there are such underreported cases and their maximum continuous temperature rise is close to the current threshold, then calculate the upper bound of the maximum continuous temperature rise among all underreported cases.

[0252] The current preset temperature rise threshold is compared with the upper bound. If the current threshold is higher than the upper bound, the temperature rise threshold is updated to the value of the upper bound rounded down.

[0253] The updated temperature rise threshold is loaded into the alarm rule engine and associated with a group of switchgear of the same model and load characteristics;

[0254] Initiate a high-frequency sampling verification period, and then intensively monitor this type of equipment in the following several cycles. Once it is confirmed that the new threshold does not introduce false alarms, the calibration loop is completed.

[0255] D63: Determine if there are frequent triggering of Level 3 alarms without subsequent temperature rise or high temperature development, and if the temperature mutation value is caused by a single frame jump rather than a continuous rate of change exceeding the limit. If so, it indicates that the preset mutation threshold is too sensitive to instantaneous noise, and the enhanced process of the joint verification mechanism of mutation threshold and rate of change is initiated.

[0256] The enhanced process of the joint verification mechanism for mutation threshold and rate of change includes:

[0257] Collect samples marked as false alarms from all recent Level 3 alarm events, and extract their temperature abrupt change values ​​and corresponding temperature change rate per unit time sequences;

[0258] Determine whether these false alarms are caused by a single frame temperature jump, and whether the temperature change rate between consecutive cycles does not continuously exceed the preset change rate threshold;

[0259] If it is confirmed that the main cause of the false alarm is an isolated jump rather than a real sudden change trend, a dual-condition joint judgment mechanism is activated: a level 3 alarm is only allowed to be triggered when the temperature change value exceeds the preset change threshold and the temperature change rate exceeds the preset change rate threshold for two or more consecutive monitoring cycles.

[0260] The joint decision-making backtracking was re-executed on historical mutation events to verify whether the new mechanism can effectively suppress false alarms while retaining the ability to identify true mutations.

[0261] If the backtracking verification passes, the joint judgment logic will be solidified into the alarm judgment module, and the calculation method of the temperature mutation value will be updated to the trend derivative based on the sliding window.

[0262] The record of this mechanism enhancement will be stored in the system configuration version library as a constraint for subsequent model training.

[0263] Therefore, this solution distinguishes between the physical causes of false alarms and false alarms, and adopts strategies such as distribution boundary analysis, trend-based false alarm backtracking, and multi-cycle joint verification to achieve refined and scenario-based dynamic adjustment of the three types of threshold parameters. This enables the alarm system to adapt to the thermal characteristics evolution of the equipment throughout its entire life cycle and resist environmental noise interference, significantly improving the robustness and engineering reliability of anomaly identification.

[0264] Among them, the lower limit of distribution refers to the lowest percentile boundary obtained by statistically analyzing the highest temperature value in non-real Level 1 alarm events according to cumulative probability; the trend derivative of the sliding window refers to the slope of local temperature change calculated based on the temperature sequence of multiple consecutive monitoring points; the fault mode library refers to a structured database storing verified fault cases and their temperature evolution characteristics; the structured anomaly handling feedback information refers to a digital record containing fault confirmation status, retest results, and handling actions, filled in and submitted by maintenance personnel on a mobile terminal according to a standard template; the temperature distribution difference area refers to the set of pixels in the current infrared thermal image whose temperature deviation exceeds the tolerance range and has spatial continuity compared to the reference thermal image; the preset ratio refers to the area percentage threshold used to determine whether the thermal image difference constitutes a significant anomaly, which can be preset according to actual conditions; the incremental training process refers to a machine learning update mechanism that uses new samples to locally fine-tune the temperature trend prediction model while retaining the original model parameters; and the alarm rule engine refers to a logic execution module that automatically matches and executes multi-level threshold judgments based on the switch cabinet model, rated current, and environmental compensation parameters.

[0265] In summary, this application proposes a method for anomaly identification in substation switchgear. By installing a miniature infrared thermal imaging device on the switchgear door, non-contact real-time temperature monitoring of the moving and stationary contacts inside the switchgear is achieved, effectively solving the problem of data partiality in traditional monitoring methods. This method can acquire real-time temperature data from multiple monitoring points inside the switchgear and upload it to a big data analysis platform via wired or wireless communication, enabling continuous monitoring of equipment operating status. In the big data analysis platform, the highest temperature of the monitoring area, the highest temperature difference between adjacent monitoring periods, and temperature abrupt changes are calculated based on the real-time temperature data. When these parameters exceed preset thresholds, an anomaly alarm signal can be generated promptly, triggering corresponding alarm strategies, thereby improving monitoring efficiency and the ability to handle complex operating conditions. Furthermore, this method further improves the accuracy and timeliness of anomaly identification by generating infrared thermal images, extracting areas of temperature distribution difference, and marking high-risk monitoring points. Simultaneously, by using a temperature trend prediction model to predict future temperature trends, early warning signals are generated in advance, and a preventative alarm mechanism is activated, providing strong support for the safe and stable operation of the power system.

[0266] like Figure 3 As shown, this is a schematic diagram of the monitoring system obtained according to this embodiment, specifically a schematic diagram of real-time infrared thermal image monitoring of the temperature monitoring column, and a schematic diagram of temperature anomaly identification parameters. Figure 4 This is a schematic diagram of real-time infrared thermal imaging monitoring on the real-time monitoring panel, and a schematic diagram of temperature anomaly identification parameters. Figure 5 These are example values ​​for specific parameters in the temperature anomaly identification parameters, such as relevant data for real-time alarms and responses, and alarm response mechanisms. These values ​​have not yet been confirmed through operation. Figure 6This is a schematic diagram after confirming the Level 2 temperature rise alarm.

[0267] Example 2: This example also provides a substation switchgear anomaly identification system, including:

[0268] The data acquisition and processing module is used to install a miniature infrared thermal imaging device on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device acquires a complete frame of infrared thermal images of the inside of the switchgear according to a preset scanning frequency; and performs spatial mapping and ambient temperature compensation processing on each frame of infrared thermal images.

[0269] The structured processing module is used to process the acquired real-time temperature data using a physical model-driven environmental reflection component stripping method, store the data, assign a unique serial number, and upload it to the big data analysis platform.

[0270] The equipment anomaly alarm module is used by the big data analysis platform to calculate the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature mutation value based on the uploaded real-time temperature data. When the highest temperature of the monitoring area exceeds a preset high temperature threshold, or the difference between the highest temperatures of adjacent monitoring periods exceeds a preset temperature rise threshold, or the temperature mutation value exceeds a preset mutation threshold, an equipment anomaly alarm signal is generated. The highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring period. The difference between the highest temperatures of adjacent monitoring periods refers to the numerical difference between the highest temperature of the current period and the highest temperature of the previous period. The temperature mutation value refers to the temperature difference corresponding to the rate of temperature change exceeding a preset rate of change threshold per unit time. The preset rate of change threshold is a critical rate of change used to determine whether a sudden temperature rise has occurred.

[0271] The alarm execution module is used to collect alarm information records of the same type of switchgear at all levels within the most recent monitoring periods, and to calculate and execute the judgment of level 1 alarm, level 2 alarm and level 3 alarm respectively.

[0272] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0273] This embodiment also provides an electronic device, which can be a terminal, and its internal structure diagram can be as follows. Figure 2As shown, the electronic device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for identifying abnormalities in a substation switchgear. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the device's casing, or an external keyboard, touchpad, or mouse.

[0274] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0275] A miniature infrared thermal imaging device is installed on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device is used to perform non-contact real-time temperature monitoring of the moving and stationary contacts inside the switchgear.

[0276] Real-time temperature data of multiple monitoring points inside the switch cabinet is obtained through a miniature infrared thermal imaging device.

[0277] Real-time temperature data is uploaded to the big data analysis platform via wired or wireless communication.

[0278] In the big data analysis platform, based on real-time temperature data, the highest temperature of the monitored area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature abrupt change value are calculated.

[0279] When the highest temperature exceeds the preset high temperature threshold, or the difference between the highest temperatures in adjacent monitoring cycles exceeds the preset temperature rise threshold, or the temperature change value exceeds the preset change threshold, an equipment abnormality alarm signal is generated.

[0280] In response to an equipment malfunction alarm signal, a high-temperature color alarm and / or audible alarm are triggered.

[0281] For illustrative purposes, the foregoing description has been made with reference to specific embodiments. However, the foregoing illustrative discussions are not intended to be exhaustive or to limit the present disclosure to the precise forms disclosed. Numerous modifications and variations are possible in accordance with the foregoing teachings. These embodiments were chosen and described in order to best illustrate the principles of the present disclosure and its practical application, thereby enabling those skilled in the art to best utilize the disclosure and to employ various embodiments with different modifications to suit a particular intended application.

Claims

1. A substation switchgear abnormality identification method, characterized by, The method includes: A miniature infrared thermal imaging device is installed on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device acquires a complete frame of infrared thermal images of the inside of the switchgear according to a preset scanning frequency. Spatial mapping and ambient temperature compensation processing are performed on each frame of infrared thermal images. The acquired real-time temperature data is structured using a physical model-driven environmental reflection component stripping method, then stored, assigned a unique serial number, and uploaded to the big data analysis platform. The big data analysis platform calculates the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature abrupt change value based on the uploaded real-time temperature data. When the highest temperature of the monitoring area exceeds a preset high temperature threshold, or the difference between the highest temperatures of adjacent monitoring periods exceeds a preset temperature rise threshold, or the temperature abrupt change value exceeds a preset abrupt change threshold, an equipment abnormality alarm signal is generated. The highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring period. The difference between the highest temperatures of adjacent monitoring periods refers to the numerical difference between the highest temperature of the current period and the highest temperature of the previous period. The temperature abrupt change value refers to the temperature difference corresponding to the rate of temperature change exceeding a preset rate of change threshold per unit time. The preset rate of change threshold is a critical rate of change used to determine whether a sudden temperature rise has occurred. Collect alarm information records of the same type of switchgear at all levels within the most recent monitoring periods, and calculate and execute the judgment of first-level alarm, second-level alarm and third-level alarm respectively; The process of performing spatial mapping and ambient temperature compensation on each frame of infrared thermal image includes: Based on the pre-defined spatial coordinate mapping relationship, the position of each pixel in the infrared thermal image is mapped to the specific physical location of the moving contact, stationary contact and their connection structure inside the switch cabinet; Non-uniformity correction and ambient temperature drift compensation are performed on the original infrared radiation signal to eliminate measurement deviations caused by air convection, metal reflection, or background heat sources inside the cabinet. Ambient temperature drift compensation refers to dynamically correcting the infrared temperature measurement results based on the reference temperature of the environment inside the cabinet to eliminate the impact of environmental fluctuations on measurement accuracy. Based on the temperature distribution of the current pixel and its neighborhood, the representative temperature value of the specific physical location corresponding to the pixel is extracted and marked as a valid monitoring point; the valid monitoring point refers to the monitoring location located at the moving contact, stationary contact or key connection part and whose temperature data has been verified by noise filtering and jump rejection. The temperature values ​​of all valid monitoring points, along with their spatial coordinates and acquisition timestamps, are encapsulated to form a structured real-time temperature dataset for subsequent uploading and analysis. The structured real-time temperature dataset refers to a standardized set of data units containing triples of temperature value, spatial coordinates, and timestamp. The physical model-driven environmental reflection component stripping method includes: The system receives real-time temperature data from a miniature infrared thermal imaging device and converts it into a raw temperature matrix. It then performs neighborhood consistency checks on the temperature value of each monitoring point. If the temperature difference between a monitoring point and its neighboring monitoring points exceeds a preset jump tolerance threshold, the point is marked as a suspicious anomaly. Determine whether the number of suspicious anomalies in the current scanning cycle exceeds the preset upper limit of the total number of monitoring points. If it does not exceed the limit, perform environmental temperature drift compensation on all non-suspicious anomalies, correct the infrared temperature measurement deviation based on the reading of the reference temperature sensor in the cabinet, and generate a pre-cleaned temperature dataset. Otherwise, discard all temperature data in the current scanning cycle, record the invalid data event log, and trigger the self-test process of the miniature infrared thermal imaging device to check for lens contamination, power supply fluctuations, or communication interruption. After completing the self-test, wait for the next scanning cycle to re-acquire data. To centralize the temperature data after initial cleaning, each temperature value is bound to a collection timestamp and a spatial location code. The timestamp comes from the high-precision real-time clock inside the device, and the spatial location code is generated based on the pre-calibrated mapping relationship between infrared image pixels and the internal structure of the switch cabinet. Temperature data with spatiotemporal tags is encapsulated into structured temperature data packets and written to the tail of the first-in-first-out queue of the local non-volatile storage module. After assigning a unique sequence number, the local cache management program is started. Detect the current communication link status. If the link is stable, prepare to upload in batches. If the link is interrupted, retain the data packets until the link is restored. The process of performing environmental temperature drift compensation on all non-suspicious anomalies includes: At least one high-stability ambient reference temperature sensor is fixedly installed in the non-heating area inside the switch cabinet. The surface of the sensor is coated with a low emissivity material and shielded from direct thermal radiation, and is used to collect the background ambient temperature inside the cabinet in real time. Simultaneously acquire the raw infrared radiation intensity matrix output by the miniature infrared thermal imaging device and the reading of the environmental reference temperature sensor, and align the timestamps of both to the same scanning cycle; Based on the infrared thermal imaging physical model, the theoretical radiation intensity of an ideal blackbody at the current ambient temperature is calculated and compared with the actual infrared radiation intensity to derive the proportion of environmental reflection components. For each pixel in the original infrared radiation intensity matrix, the spurious radiation value contributed by the environmental reflection component is subtracted, and the true component that comes only from the thermal radiation of the measured contact itself is retained. The corrected infrared radiation intensity is converted into surface temperature values ​​to generate a temperature matrix after environmental drift compensation.

2. The substation switchgear abnormality identification method of claim 1, wherein, The big data analysis platform calculates the highest temperature in the monitored area, the difference in highest temperature between adjacent monitoring periods, and temperature abrupt changes based on the uploaded real-time temperature data, including: Obtain the real-time temperature values ​​of each monitoring point at the current moment, and determine the maximum value as the highest temperature; Obtain the historical highest temperature value of each monitoring point in the previous monitoring period, and calculate the difference between the current highest temperature and the historical highest temperature as the highest temperature difference between adjacent monitoring periods; Time series analysis is performed on the temperature data within the current monitoring period to calculate the rate of temperature change per unit time. When the rate of change is greater than a preset rate of change threshold, the temperature difference corresponding to that rate of change is determined as the temperature abrupt change value. 3.The substation switchgear abnormality identification method of claim 1, wherein, The big data analysis platform calculates temperature fluctuation values ​​based on the uploaded real-time temperature data, including: Structured temperature data of the same monitoring point in the most recent consecutive monitoring periods are extracted from the local cache to construct the temperature time series of that point; Using a fixed time interval as the step size, the ratio of the temperature difference to the time difference between two adjacent points is calculated to obtain the instantaneous temperature change rate per unit time. The calculated rate of change at each moment is compared with a preset rate of change threshold. If any rate of change exceeds the threshold, the actual temperature difference within the corresponding time period is recorded as a candidate mutation value. The largest candidate mutation value among all monitoring points is selected as the temperature mutation value for this monitoring cycle for subsequent alarm judgment.

4. The substation switchgear abnormality identification method of claim 3, wherein, The statistical analysis of alarm information records at all levels for the same type of switchgear within the most recent monitoring periods, and the calculation and execution of judgments for Level 1, Level 2, and Level 3 alarms, includes: Collect historical temperature data of the same type of switchgear under rated load, full load and slight overload conditions during continuous operation, and extract the highest temperature distribution range of the moving contact and stationary contact areas under thermal stability conditions. Determine whether the distribution range covers at least thirty complete monitoring cycles. If it does, calculate the high percentile value of the highest temperature distribution range and round it up by a fixed temperature rise step as the preset high temperature threshold. Ensure that this threshold is higher than the upper limit of the normal thermal steady state of the equipment but lower than the thermal failure critical point of the insulation material. Otherwise, continue to collect data until the cycle requirement is met. Based on the sequence of highest temperature differences between adjacent monitoring periods in historical temperature data, the maximum temperature rise difference under normal load fluctuation conditions is statistically analyzed. Determine if there are any records of non-abnormal load change events caused by external scheduling. If so, remove the temperature rise difference for the corresponding time period, recalculate the maximum value, multiply the corrected maximum temperature rise difference by a safety factor greater than one and round it down to the nearest integer as the preset temperature rise threshold. This threshold is used to distinguish between abnormal temperature rise caused by reasonable load changes and contact resistance degradation. Otherwise, directly multiply the corrected maximum temperature rise difference by a safety factor greater than one and round it down to the nearest integer as the preset temperature rise threshold. This threshold is used to distinguish between abnormal temperature rise caused by reasonable load changes and contact resistance degradation. Retrieve all confirmed temperature mutation events from the historical fault case database and extract the temperature change rate from normal operation to the last effective monitoring period before the fault occurred. Determine whether each event has complete pre-monitoring data and no communication interruption. If it does, calculate the minimum temperature change rate at the onset of the mutation in all valid events, set the minimum value as the preset change rate threshold, and calculate the corresponding temperature difference value in combination with the current system monitoring cycle time interval as the preset mutation threshold. Otherwise, remove the event and continue to process the next case. The preset high temperature threshold, preset temperature rise threshold, and preset sudden change threshold are written into the alarm rule engine of the big data analysis platform and bound to the switch cabinet model, rated current, and ambient temperature compensation coefficient. Once the newly connected switchgear completes its initial identification, the system automatically matches its model and operating parameters, loads the corresponding high temperature threshold, preset temperature rise threshold and preset sudden change threshold parameter set, and performs real-time alarm judgment. After the system is put into operation, a threshold verification process is initiated once every quarter. The historical statistical distribution is updated using newly added operational data to determine whether the false alarm rate in the current alarm record exceeds the preset false alarm limit or the missed alarm rate exceeds the preset missed alarm limit.

5. The substation switchgear abnormality identification method of claim 4, wherein, The method also includes: closed-loop optimization of the baseline heatmap, alarm threshold, and prediction model based on operation and maintenance feedback and operational data, specifically including the following steps: Step 1: Receive structured anomaly handling feedback information submitted by maintenance personnel on mobile terminals. The feedback information includes fault authenticity determination, retested temperature data, physical defect location, and execution status of handling measures. Step 2: Determine whether this abnormal event is a real fault. If the feedback information confirms that there are physical defects such as contact surface burning, loose bolts or insulation deterioration, proceed to step 3; otherwise, proceed to step 6. Step 3: Detect the load status and environmental stability of the switchgear during continuous operation after the anomaly is handled. If the load rate is within the rated range and the fluctuation of the ambient temperature inside the cabinet is lower than the stability threshold, proceed to step 4; otherwise, proceed to step 6. Step 4: After collecting and processing infrared thermal image data for multiple consecutive monitoring cycles, perform a difference comparison between each frame of thermal image and the current baseline thermal image, and calculate the area ratio of the temperature distribution difference region. Step 5: Determine whether the area ratio of temperature distribution difference regions in all acquisition frames does not exceed the preset ratio. If so, use the heat map data of this period to reconstruct the baseline heat map and replace the original baseline heat map. At the same time, store the complete temperature evolution sequence of this abnormal event, the spatial location of high-risk monitoring points and the final fault type into the fault mode library, and trigger the incremental training process of the temperature trend prediction model. Step 6: Analyze the alarm records of the same type of switchgear at all levels in the most recent monitoring periods, and calculate the false alarm rate and false negative rate of Level 1 alarm, Level 2 alarm and Level 3 alarm respectively.

6. An abnormality identification system based on the substation switch cabinet abnormality identification method of claim 1, characterized in that, The system includes: The data acquisition and processing module is used to install a miniature infrared thermal imaging device on the cabinet door of the substation switchgear. The miniature infrared thermal imaging device acquires a complete frame of infrared thermal images of the inside of the switchgear according to a preset scanning frequency; and performs spatial mapping and ambient temperature compensation processing on each frame of infrared thermal images. The structured processing module is used to process the acquired real-time temperature data using a physical model-driven environmental reflection component stripping method, store the data, assign a unique serial number, and upload it to the big data analysis platform. The equipment anomaly alarm module is used by the big data analysis platform to calculate the highest temperature of the monitoring area, the difference between the highest temperatures of adjacent monitoring periods, and the temperature mutation value based on the uploaded real-time temperature data. When the highest temperature of the monitoring area exceeds a preset high temperature threshold, or the difference between the highest temperatures of adjacent monitoring periods exceeds a preset temperature rise threshold, or the temperature mutation value exceeds a preset mutation threshold, an equipment anomaly alarm signal is generated. The highest temperature refers to the maximum real-time temperature among all monitoring points in the current monitoring period. The difference between the highest temperatures of adjacent monitoring periods refers to the numerical difference between the highest temperature of the current period and the highest temperature of the previous period. The temperature mutation value refers to the temperature difference corresponding to the rate of temperature change exceeding a preset rate of change threshold per unit time. The preset rate of change threshold is a critical rate of change used to determine whether a sudden temperature rise has occurred. The alarm execution module is used to collect alarm information records of the same type of switchgear at all levels within the most recent monitoring periods, and to calculate and execute the judgment of level 1 alarm, level 2 alarm and level 3 alarm respectively.

Images