A method for early warning of abnormal temperature in switchgear

Abstract

This application provides a method for early warning of abnormal temperature in switchgear, relating to the field of switchgear temperature early warning. The method includes: collecting continuous temperature data from key parts of the switchgear and preprocessing it to obtain a temperature time series; dividing the temperature time series into sliding window segments to obtain micro-window data segments; calculating the temperature rise rate score for each micro-window data segment; calculating the temperature fluctuation amplitude score for each micro-window data segment; constructing an anomaly scoring function based on the temperature rise rate score and the temperature fluctuation amplitude score to obtain a comprehensive anomaly score for each micro-window; and performing over-limit judgment based on each comprehensive anomaly score, triggering a graded early warning, and outputting structured graded early warning information. The technical solution of this application possesses an intelligent early warning mechanism that combines real-time performance, interpretability, and practicality, and has broad engineering applicability and good prospects for promotion.

Description

Technical Field

[0001] This application relates to the field of switchgear temperature early warning, and in particular to a method for early warning of abnormal switchgear temperature. Background Technology

[0002] Current mainstream temperature anomaly detection solutions still primarily rely on fixed threshold strategies. The biggest advantage of this approach is its simplicity and ease of implementation, but its core drawback is its inability to handle "trend-based" temperature rise anomalies. Extensive engineering experience shows that switchgear thermal failures are often not "sudden high temperatures," but rather a complex process of "slow temperature rise + inability to reach a stable state." Fixed thresholds can only passively respond to the result, failing to identify process-related and early warning patterns, often leading to delayed alarms or even missing critical stages of fault development.

[0003] Some improved solutions incorporate statistical features, such as moving averages, slope, or rate of change monitoring, which provide a certain degree of trend response capability. However, due to their lack of contextual modeling capabilities, they are easily affected by external variables such as load disturbances and ambient temperature differences, resulting in relatively coarse judgment logic. More seriously, the nonlinear and non-stationary characteristics of temperature changes make it difficult for static statistical rules to adapt to different cabinet types and operating conditions, hindering their widespread adoption.

[0004] To improve detection accuracy, some studies employ machine learning or deep learning models. While these models can capture complex patterns, especially those based on LSTM and Transformer architectures that can characterize long-term dependencies in time series, their training process relies on large amounts of high-quality labeled data, and the models are opaque and difficult to interpret. In industrial field deployments, their high resource consumption, difficulty in debugging, and difficulty in localization severely restrict their practical application and promotion. Furthermore, deep models struggle to identify "atypical faults" in unlabeled states, rendering them ineffective against early warning signs of risks.

[0005] Existing technologies have significant shortcomings in terms of practicality, deployability, interpretability, and forward-looking response, especially for edge computing devices or small-scale scenarios (such as low-voltage distribution cabinets and outdoor switch boxes). A lightweight, fast-responding, rule-transparent, and embeddable trend-based early warning mechanism is needed. Therefore, there is an urgent need for a dynamic adaptive scoring mechanism that lies between "full rule-based" and "full black box" approaches, possessing both controllable rule logic and the ability to respond to complex evolutionary patterns, thus filling the practical gaps in existing methods. Summary of the Invention

[0006] The purpose of this invention is to provide a method for early warning of abnormal temperature in switchgear in order to solve the problems of slow response, serious false alarms, poor interpretability and unsuitability for edge scenarios in existing switchgear temperature early warning technology.

[0007] The above-mentioned objective of this application is achieved through the following technical solution: Step S1: Collect continuous temperature data of key parts of the switchgear and preprocess it to obtain a temperature time series; Step S2: Divide the temperature time series into sliding window segments to obtain micro-window data segments; Step S3: Calculate the temperature rise rate score for each micro-window data segment; Step S4: Calculate the temperature fluctuation range score for each micro-window data segment; Step S5: Based on the temperature rise rate score and temperature fluctuation amplitude score, construct an anomaly scoring function to obtain the comprehensive anomaly score for each micro-window; Step S6: Based on the comprehensive anomaly scores, make an over-limit judgment, trigger a graded warning, and output structured graded warning information.

[0008] Optionally, step S1 includes: Key components of the switchgear include: switchgear surface, instrument transformers, circuit breaker contacts, busbar interfaces, and cable terminals; Preprocessing includes: data cleaning, noise reduction, normalization, and sequence segmentation.

[0009] Optionally, step S2 includes: Construct a sliding window and set the duration of a single window to 1. The step length is Window length Take a time interval between 10 and 60 seconds, with a step size of [missing information]. Less than or equal to 1 / 2 of the window length; Given a preprocessed temperature-time series, denoted as Each sampling point For the first Temperature observation value at that time; By using a sliding window operation, the temperature time series can be divided into a series of lengths. The subsequences, each representing a micro-window data segment; The starting index of the sliding window is constructed as follows:

[0010] in For the first The starting position of each window The number of windows that can be generated, satisfying the constraints. ; For each valid window, it can be represented as:

[0011] in Indicates the first Each micro-window data segment; each window This constitutes a continuous temperature change segment; Display window The first temperature sample value.

[0012] Optionally, step S3 includes: For micro-window data segments The rate of temperature rise is defined as the difference between the temperature at the end of the window and the initial temperature divided by the window duration, i.e.:

[0013] in, For the first The temperature rise rate score for each window; A linear curve is fitted using least squares within the window, and its slope is calculated as a score for the rate of temperature rise, such as:

[0014] in, This represents a function that extracts the slope parameter of the fitted line; This represents the linear least squares fitting operation; Set a reference threshold for the rate of temperature rise scoring ,when Exceeding this threshold can be used as an indicator of a rapid increase in temperature, as follows:

[0015] in and These represent the mean and standard deviation of the historical window rate, respectively. The adjustment parameters are used to control sensitivity.

[0016] Optionally, step S4 includes: Micro Window Data Segment The temperature fluctuation range score is defined as follows:

[0017] in For the first The temperature fluctuation range score for each window represents the maximum temperature deviation within that window. and Indicates the maximum and minimum temperatures within the window; Set a volatility threshold for volatility rating Only when Only when a certain time is the state considered unstable, and the fluctuation threshold is set using a dynamic distribution modeling approach:

[0018] in and These represent the mean and standard deviation of the historical window temperature fluctuation range scores, respectively. To adjust the parameters.

[0019] Optionally, step S5 includes: A weighted linear scoring structure is used to construct an anomaly scoring function for the first... For each micro-window data segment, the anomaly scoring is defined as follows:

[0020] in For the first The abnormal rating values ​​of each window, and Scoring for adjustable temperature rise rate Score based on temperature fluctuation amplitude Weighting coefficients; By reference threshold and fluctuation threshold The scoring results are then corrected to obtain the final comprehensive anomaly score, as follows:

[0021] in Indicates the first A comprehensive anomaly score for each micro-window.

[0022] Optionally, step S6 includes: If the abnormal score values ​​of several consecutive windows all exceed the set threshold, it is considered that the current device has a real abnormal trend, triggering an alert; let the score threshold for the alert be... The number of consecutive windows is The alarm judgment conditions are as follows:

[0023] in For indicator functions, it means the first... Does the score of each window exceed the threshold, and does the cumulative number of windows that continuously meet the conditions reach [a certain threshold]? Timely triggering of warnings; The warning signals are divided into a multi-level response mode, i.e., a graded warning system, and are output according to the scoring range or the degree of continuous exceedance. Specifically, this includes: Define multiple rating level ranges: If the score satisfies: If this occurs, a Level 1 warning will be triggered, indicating that there are signs of temperature rise in key parts of the switchgear. like: If the temperature rises rapidly, it is a Level 2 warning, which means that the critical parts of the switchgear are heating up rapidly. like: This triggers a Level 3 warning, indicating a severe temperature anomaly in critical components of the switchgear; among which... and To further classify and differentiate thresholds; The warning reset setting is described as follows:

[0024] in To reset the threshold, To maintain the length of the decline window, which represents the time range within which the score declines continuously and stably.

[0025] An electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform a switch cabinet temperature anomaly early warning method.

[0026] A computer-readable storage medium storing instructions that, when executed, perform a method for early warning of abnormal temperature in a switchgear.

[0027] The beneficial effects of the technical solution provided in this application are: Based on this, this invention proposes a "micro-window dynamic scoring mechanism." By dividing the temperature time series into sliding window segments and calculating the temperature rise rate and local stability index within each micro-window, a multi-dimensional precursor scoring model is constructed to dynamically assess whether there is an "atypical" temperature rise trend in the current operating state. Compared with traditional threshold determination or global model prediction, this method has the following advantages: First, by constructing a sliding window, it can track the temperature evolution process in a fine-grained manner over time, avoiding dependence on single-point abrupt changes and improving the response sensitivity to early, slow temperature rises. Second, by fusing two indicators, the rate factor and fluctuation amplitude, it constructs a more discriminative anomaly scoring function, which can identify both sudden temperature increases and potential contact defects characterized by "drastic temperature fluctuations but unchanged mean." Third, the scoring function parameters can be dynamically adjusted based on historical data, enabling the model to adapt to different ambient temperatures and load conditions. Finally, the entire method has clear rules and transparent logic, and can be deployed in embedded edge devices to achieve local early warning without relying on complex models and cloud resources, thus possessing extremely high practical value. Attached Figure Description

[0028] The present application will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a step diagram of an embodiment of this application; Figure 2This is a raw temperature sequence diagram from an embodiment of this application; Figure 3 This is a graph showing the scoring results of the sliding window in an embodiment of this application; Figure 4 This is a mapping diagram showing the score values ​​and warning levels in the embodiments of this application; Figure 5 This is a superimposed graph of temperature curves and anomaly scores in the embodiments of this application; Figure 6 This is a heatmap of the response of the scoring function to parameter changes in the embodiments of this application; Figure 7 This is a diagram showing the center position of the abnormal window in the embodiments of this application; Figure 8 This is a schematic diagram of the electronic device structure in the embodiments of this application. Detailed Implementation

[0029] To provide a clearer understanding of the technical features, objectives, and effects of this application, the specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0030] The embodiments of this application provide a method for early warning of abnormal temperature in switchgear.

[0031] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating the steps of a switchgear temperature anomaly early warning method according to an embodiment of this application, including: Step S1: Collect continuous temperature data of key parts of the switchgear and preprocess it to obtain a temperature time series; Step S2: Divide the temperature time series into sliding window segments to obtain micro-window data segments; Step S3: Calculate the temperature rise rate score for each micro-window data segment; Step S4: Calculate the temperature fluctuation range score for each micro-window data segment; Step S5: Based on the temperature rise rate score and temperature fluctuation amplitude score, construct an anomaly scoring function to obtain the comprehensive anomaly score for each micro-window; Step S6: Based on the comprehensive anomaly scores, make an over-limit judgment, trigger a graded warning, and output structured graded warning information.

[0032] This application provides an embodiment as follows: by implementing sliding window segmentation on the temperature time series, two key indicators, the temperature rise rate and the temperature fluctuation amplitude, are calculated in each micro-window to construct a precursor scoring function for assessing the degree of abnormality in the operating state, and multi-level early warning output is realized based on the continuity of the scoring results and threshold judgment.

[0033] The core technical feature of this invention lies in its use of a micro-window structure to achieve fine-grained local analysis of temperature changes. By combining temperature rise rate and stability scoring to construct trend judgment logic, it eliminates the dependence on absolute temperature values ​​and enables early detection of atypical faults such as abnormal temperature rise trends, periodic fluctuations, and hidden contact defects. Furthermore, the scoring function employs linear or piecewise logic, with transparent and adjustable parameters, providing excellent interpretability and engineering controllability, making it suitable for deployment in lightweight edge nodes or embedded hardware systems.

[0034] As one example, thermal failures in switchgear are often not "sudden high temperatures," but rather a complex process of "slow temperature rise + inability to reach a stable state." The technical solution of this application is an important technological innovation for switchgear operation safety scenarios.

[0035] As one example, it boasts several advantages: Fast response speed: Through continuous scoring within a local window, it avoids reliance on long-term data modeling, enabling real-time capture of early temperature rise trends and achieving early warning at the precursor level; Transparent and logically clear criteria: The warning scoring function is constructed based on physical meaning (rate + fluctuation amplitude), facilitating understanding, debugging, and subsequent engineering adjustments, and possessing strong interpretability; Lightweight and highly deployable algorithm: The overall logic lacks a complex model structure, allowing direct deployment on edge acquisition devices or gateway devices without training samples; Strong adaptability: The scoring function supports a dynamic threshold update mechanism, automatically adjusting the judgment criteria based on operating conditions such as ambient temperature and current load, thus improving its applicability; Effectively addresses the shortcomings of existing methods: It overcomes the bottleneck of traditional methods failing to identify slow temperature rises or "non-over-temperature" anomalies, enhancing operational foresight and system security capabilities.

[0036] Step S1 includes: Key components of the switchgear include: switchgear surface, instrument transformers, circuit breaker contacts, busbar interfaces, and cable terminals; Preprocessing includes: data cleaning, noise reduction, normalization, and sequence segmentation.

[0037] In one embodiment, this step primarily involves acquiring and preliminarily processing temperature data from key components of the switchgear during operation, providing structured input for subsequent sliding window construction and dynamic scoring analysis. The system continuously collects temperature data at fixed time intervals using temperature sensors deployed at locations prone to thermal faults, such as busbars, circuit breaker contacts, and cable terminals, forming a time series. Before being input into the scoring module, the collected data undergoes basic preprocessing operations, including removing missing points, smoothing outliers, temperature normalization, and time alignment, to improve the robustness and accuracy of subsequent analysis.

[0038] Step S2 includes: Construct a sliding window and set the duration of a single window to 1. The step length is Window length Take a time interval between 10 and 60 seconds, with a step size of [missing information]. Less than or equal to 1 / 2 of the window length; Given a preprocessed temperature-time series, denoted as Each sampling point For the first Temperature observation value at that time; By using a sliding window operation, the temperature time series can be divided into a series of lengths. The subsequences, each representing a micro-window data segment; The starting index of the sliding window is constructed as follows:

[0039] in For the first The starting position of each window The number of windows that can be generated, satisfying the constraints. ; For each valid window, it can be represented as:

[0040] in Indicates the first Each micro-window data segment; each window This constitutes a continuous temperature change segment; Display window The first temperature sample value.

[0041] As one embodiment, window length The window length should be selected based on the actual temperature rise characteristics. Too short a window can lead to unstable scoring, while too long a window may mask short-term anomalies. In this invention, the window length is generally between 10 and 60 seconds, and the step size... It is generally set to 1 / 2 or less of the window length to ensure coverage density and enhance detection sensitivity.

[0042] As one embodiment, to adapt to changes in data density under different operating conditions, the sliding window of this invention is defined in units of timestamps or sampling points, supporting dynamic adjustment of the window size to adapt to the unified scheduling logic of high-sampling-rate and low-sampling-rate devices. Window segmentation operations can be completed at the acquisition node or edge gateway, achieving low-latency data structure conversion and constructing a micro-segment sequence data structure that supports continuous scoring calculation.

[0043] As one embodiment, this step aims to divide the temperature time series into multiple micro-window segments according to a set length and step size to achieve local trend modeling of the temperature change process. In actual operation, the switchgear temperature signal exhibits nonlinearity, non-stationarity, and phased characteristics, making it difficult for global analysis to capture early trend anomalies. This invention, through the sliding structure of the micro-windows, enables the model to respond to sudden changes, fluctuations, and instability in temperature rise on a shorter time scale, thereby enhancing the sensitivity and real-time performance of the early warning mechanism.

[0044] Step S3 includes: For micro-window data segments The rate of temperature rise is defined as the difference between the temperature at the end of the window and the initial temperature divided by the window duration, i.e.:

[0045] in, For the first The temperature rise rate score for each window; In one embodiment, the numerator is the temperature change within the window, and the denominator is the window duration; the ratio of the two measures the heat accumulation efficiency during that period. In some cases, temperature changes may exhibit slight fluctuations or short-term fluctuations. To avoid rate distortion caused by abnormal peaks, this invention also supports using the slope of a linear fitting within the window as a rate estimation method. This involves fitting a linear curve using least squares within the window and calculating its slope as the temperature rise rate score. This method is similar to the first-to-last difference method when the equipment is stable or experiencing minimal fluctuations, but it provides a more stable score when high-frequency disturbances or induced interference exist.

[0046] A linear curve is fitted using least squares within the window, and its slope is calculated as a score for the rate of temperature rise, such as:

[0047] in, This represents a function that extracts the slope parameter of the fitted line; This represents the linear least squares fitting operation; Set a reference threshold for the rate of temperature rise scoring ,when Exceeding this threshold can be used as an indicator of a rapid increase in temperature, as follows:

[0048] in and These represent the mean and standard deviation of the historical window rate, respectively. The adjustment parameters are used to control sensitivity.

[0049] As one embodiment, the present invention allows setting a reference threshold for the temperature rise rate score. ,when Exceeding this threshold can be used as an indicator of a rapid temperature rise. This threshold can be a fixed value or adaptively updated using the historical window mean and standard deviation. This dynamic threshold strategy is well-adapted to scenarios with drastic temperature baseline fluctuations or significant seasonal environmental changes. Finally, this step outputs a temperature rise rate score for each window. This is used to synthesize anomaly scoring functions with stability scores in subsequent tests.

[0050] As one embodiment, after the sliding window is constructed, this step performs a rate analysis of temperature change within each micro-window to capture the speed at which the temperature rises within the switchgear components per unit time. This rate information reflects the rate of heat source energy release, the effectiveness of the heat dissipation path, and changes in contact conditions, and is an important physical quantity for judging early signs of faults. Compared to absolute temperature values, the temperature rise rate has a higher dynamic response capability and can detect potential abnormal trends before the temperature exceeds the warning line.

[0051] Step S4 includes: As one example, the temperature stability calculation method is based on the temperature range within a micro-window, that is, the difference between the maximum and minimum temperatures within the window. The larger the temperature fluctuation amplitude score, the more drastic the temperature change during that period, the worse the stability, and the potential for system instability; conversely, it indicates that the system is in a thermal steady state or a stable load phase.

[0052] Micro Window Data Segment The temperature fluctuation range score is defined as follows:

[0053] in For the first The temperature fluctuation range score for each window represents the maximum temperature deviation within that window. and Indicates the maximum and minimum temperatures within the window; As one example, considering that high fluctuations do not always represent risk (such as short-term load switching or air-cooled startup, which can also cause temperature jumps), this invention sets a tolerance threshold for fluctuation amplitude scoring. Only when Only when certain conditions are met is the state considered unstable. This strategy is particularly suitable for scenarios with drastic changes in operating conditions or where different types of equipment coexist, automatically adjusting sensitivity to reduce false alarms.

[0054] Set a volatility threshold for volatility rating Only when Only when a certain time is the state considered unstable, and the fluctuation threshold is set using a dynamic distribution modeling approach:

[0055] in and These represent the mean and standard deviation of the historical window temperature fluctuation range scores, respectively. To adjust the parameters.

[0056] As one embodiment, to improve the accuracy of stability assessment in complex scenarios, this invention also supports the use of statistical indicators such as standard deviation, root mean square deviation, or median absolute deviation (MAD) as alternatives or supplements for further analysis of the temperature fluctuation structure within the window. Standard deviation, root mean square deviation, or median absolute deviation are used as alternatives or supplements to the temperature fluctuation amplitude scoring. Finally, this step outputs a score for the temperature fluctuation range of each window. This will be compared with the aforementioned rate of temperature rise score. Together, they form the input to the comprehensive anomaly scoring function for the next step, in order to achieve multi-dimensional fault precursor perception capabilities.

[0057] As one embodiment, after obtaining the temperature rise rate score for each micro-window, this step constructs a stability score index reflecting the temperature fluctuation amplitude within the window to further identify unstable temperature segments with potential fault characteristics. This index is used to determine whether the temperature of components inside the switch cabinet is in a relatively stable state over a certain period of time. Studies have shown that many abnormal processes caused by poor contact or partial discharge often manifest as continuous temperature fluctuations over a short period of time, rather than a single rise. Especially before reaching absolute high temperatures, this type of "unstable but not exceeding limit" temperature behavior is highly indicative of potential problems.

[0058] Step S5 includes: A weighted linear scoring structure is used to construct an anomaly scoring function for the first... For each micro-window data segment, the anomaly scoring is defined as follows:

[0059] in For the first The abnormal rating values ​​of each window, and Scoring for adjustable temperature rise rate Score based on temperature fluctuation amplitude Weighting coefficients; This scoring function has the following significant characteristics: First, the scoring calculation logic is simple and transparent, and the meaning of each physical quantity is clear, making it easy for on-site engineers to understand and adjust parameters. Through adjustment... and The proportional relationship can control the model's sensitivity to rate anomalies or fluctuation anomalies. For example, in high-load scenarios, the ratio can be appropriately increased. This enhances the ability to monitor the rate of temperature rise; in harsh environments or scenarios where contact points are aging, it can improve... This enhances the ability to respond to temperature instability. Secondly, the function structure supports extension to nonlinear combinational forms, allowing for the introduction of exponential, threshold truncation, piecewise functions, and other methods to correct the scoring results.

[0060] By reference threshold and fluctuation threshold The scoring results are then corrected to obtain the final comprehensive anomaly score, as follows:

[0061] in Indicates the first A comprehensive anomaly score for each micro-window.

[0062] As one embodiment, the above variant can improve the robustness of the model near the error boundary and reduce misjudgments caused by small fluctuations. After integrating the score numerical distribution, further normalization or upper limit settings can be applied to facilitate use by the graded early warning module.

[0063] Furthermore, this scoring mechanism is dynamically adaptive. If the system enables the adaptive threshold mechanism, then... and The system can make real-time corrections based on historical score distributions, allowing the scoring function to gradually approach the personalized behavioral characteristics of the device over long-term operation, thereby improving the accuracy and stability of the local model.

[0064] Ultimately, this step provides a scoring basis for subsequent early warning triggering logic and achieves unified modeling of different types of anomaly characteristics. (Scoring) It can be used not only for single-window risk assessment, but also supports combination with multi-window aggregation strategies for multi-time period trend fusion and fault level inference.

[0065] This application provides an embodiment as follows, in which the temperature rise rate scoring is completed. Temperature fluctuation score After the calculations, the core task of this step is to fuse the two types of scoring results to construct an anomaly scoring function that can quantify the intensity of fault trends within the micro-window. This function should comprehensively reflect two types of phenomena: on the one hand, the risk of thermal anomalies caused by rapid temperature rise; on the other hand, structural anomalies such as poor contact and heat dissipation fluctuations manifested by temperature instability. By constructing a unified scoring structure, not only can the accuracy of the overall system judgment be improved, but also good engineering support can be provided in terms of interpretability and parameter controllability.

[0066] Step S6 includes: If the abnormal score values ​​of several consecutive windows all exceed the set threshold, it is considered that the current device has a real abnormal trend, triggering an alert; let the score threshold for the alert be... The number of consecutive windows is The alarm judgment conditions are as follows:

[0067] in For indicator functions, it means the first... Does the score of each window exceed the threshold, and does the cumulative number of windows that continuously meet the conditions reach [a certain threshold]? Timely triggering of warnings; This application provides an embodiment as follows. The above strategy avoids false alarms caused by a single score peak and can also promptly identify early warning signs when minor and continuous anomalies occur in multiple windows. To improve the practicality and engineering controllability of the method, this invention further divides the warning signal into multi-level response modes and outputs them according to the score amplitude or the degree of continuous exceedance. This strategy enables the system to output different levels of risk warnings based on the score intensity, facilitating hierarchical response processing by maintenance personnel.

[0068] The warning signals are divided into a multi-level response mode, i.e., a graded warning system, and are output according to the scoring range or the degree of continuous exceedance. Specifically, this includes: Define multiple rating level ranges: If the score satisfies: If this occurs, a Level 1 warning will be triggered, indicating that there are signs of temperature rise in key parts of the switchgear. like: If the temperature rises rapidly, it is a Level 2 warning, which means that the critical parts of the switchgear are heating up rapidly. like: This triggers a Level 3 warning, indicating a severe temperature anomaly in critical components of the switchgear; among which... and To further classify and differentiate thresholds; The warning reset setting is described as follows:

[0069] in To reset the threshold, To maintain the length of the decline window, which represents the time range within which the score declines continuously and stably.

[0070] This application provides an embodiment as follows. Furthermore, the early warning module of this invention supports a "determination-hold-reset" state management logic, meaning that once an alarm state is triggered, it must meet certain "score decline" conditions to be deactivated. This avoids system oscillations caused by repeated triggering and deactivation of alarm states due to score fluctuations. Ultimately, this step maps the continuous score signal sequence into a clear and executable early warning signal output, realizing a closed-loop anomaly discrimination mechanism of "early warning identification → temperature rise monitoring → anomaly response → state holding → condition reset," providing the system with robust trend anomaly response capabilities.

[0071] After obtaining the anomaly score value for each micro-window, this step aims to construct a reliable alarm triggering mechanism based on the score sequence to address interference from transient anomalies, random fluctuations, or equipment errors that may exist in actual temperature data. Since temperature signals in industrial scenarios are often accompanied by disturbances or short-term abrupt changes, relying solely on a single score value to determine whether to trigger an alarm may lead to false alarms. Therefore, this invention proposes an early warning triggering mechanism based on a continuous score over-limit criterion, improving the stability and effectiveness of the early warning through a multi-window integral evaluation method.

[0072] This application provides an embodiment as follows, in which the output graded early warning information includes: timestamp, window start and end index, temperature rise rate score, temperature fluctuation score, comprehensive anomaly score, early warning level, and sensor number.

[0073] This application provides an embodiment as follows: After completing continuous scoring and alarm status judgment, the task of this step is to output the identified abnormal events in a structured form and label them according to the scoring intensity to support subsequent system linkage, fault diagnosis, and operation and maintenance intervention. This part, as the terminal output module of the scoring mechanism of this invention, plays the role of the feedback interface in the "perception-feedback-control" closed loop. The visualization, interpretability, and traceability of its output content are of great significance to the practicality of the entire system.

[0074] First, for any window sequence that is judged to be abnormal This invention outputs a structured early warning information record, the content of which includes, but is not limited to, the following fields: Warning timestamp (window end time): Corresponding window rating value: Temperature rise rate rating: Temperature fluctuation rating: ; Start and end index of the window: Warning level label: such as "Level 1 Warning", "Level 2 Warning", "Level 3 Warning", etc., indicating the sensor number or physical location of the alarm source. All the above fields can be output as JSON format or a structure, facilitating uploading to the cloud platform, alarm device linkage, or local data storage by the embedded system. This structured design also supports subsequent data mining modules such as "warning event cluster analysis" and "device health trend identification".

[0075] Secondly, this invention introduces a multi-level early warning level mapping rule to divide continuous scoring results into multiple response levels for differentiated processing.

[0076] This mapping mechanism combines temperature change behavior with scoring intensity to achieve classification decisions, enhancing the refinement and adaptability of alarm outputs. Users can adjust the threshold distribution according to the importance of the equipment, risk preference, and system scale, ensuring that the early warning output is both highly sensitive and avoids frequent triggering of unnecessary alarms.

[0077] In terms of output methods, early warning information can be delivered simultaneously through multiple channels: outputting mandatory alarm signals to local audible and visual alarms; pushing event notifications to SCADA, intelligent operation and maintenance platforms, or mobile terminals; persistently storing it in local logs or historical databases for later traceability; and integrating it with other temperature, current, and voltage signals to participate in multivariate decision-making. Ultimately, the conversion of temperature anomaly scoring results into early warning response signals is completed, and the signals are output to the system control module in a hierarchical structure, realizing a closed loop from signal recognition to system intervention, demonstrating good engineering practicality and deployment flexibility.

[0078] Figure 2 The original temperature sequence collected during equipment operation is displayed. A slow upward trend can be observed overall, with a sudden temperature spike around point 60, which continues until point 75 before stabilizing at a high temperature. This temperature rise may be caused by factors such as load changes, poor contact, or heat dissipation failure. The original sequence provides the basic input data for subsequent sliding window processing and scoring calculations.

[0079] Figure 3 The figure shows the scoring results for each sliding window, where the orange curve represents the temperature rise rate score, green represents the temperature fluctuation score, and red represents the overall anomaly score. As can be seen from the figure, the scoring curves show a significant increase within the 60th to 80th window segment, accurately reflecting the onset and duration of the temperature anomaly. This figure verifies the ability of the scoring mechanism of this invention to identify early anomaly trends.

[0080] Figure 4The scoring values ​​are mapped to warning levels, with each window's warning level identified by color. It can be seen that before the anomaly occurs, the score is primarily green (Level 1). As the temperature rises, the score quickly enters the orange (Level 2) and red (Level 3) ranges, completing the risk grading response. This figure clearly demonstrates that the scoring model of this invention has multi-level judgment capabilities, adapting to the classification requirements of fault precursors of varying severity.

[0081] Figure 5 The temperature curve is overlaid with the anomaly score, with temperature on the left axis and the score value on the right axis. As can be seen from the graph, the anomaly score shows an upward trend before the temperature rises significantly, especially reaching the level two warning threshold before the temperature inflection point. This indicates that the scoring mechanism has an advanced warning capability, enabling proactive identification of abnormal trends before the temperature exceeds the limit, an advantage that traditional fixed threshold methods cannot achieve.

[0082] Figure 6 This is a heatmap showing the response of the scoring function to parameter changes, where the horizontal axis represents the window number, the vertical axis represents the rate scoring weight, and the color indicates the final score value. The graph shows that under different weight combinations, the overall trend of the scoring results remains consistent, but it is more sensitive to parameter settings at certain window locations, indicating that this scoring model has a certain degree of flexibility in parameter tuning. Users can choose weight configurations that focus more on rate or fluctuation based on the specific characteristics of their devices, reflecting the adjustability and portability of this invention.

[0083] Figure 7 The center of the identified abnormal window is marked on the temperature curve (red dashed line). As can be seen from the figure, the abnormal scores are concentrated in the rapidly rising temperature zone and the stable high-temperature zone, with no false alarms occurring in the normal stable zone, verifying that this method has good localization accuracy and false alarm suppression capability. This figure also visually illustrates the system's ability to perceive continuous abnormal segments, making it suitable for trend-based fault monitoring.

[0084] This application also discloses an electronic device. (See reference...) Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 500 may include: at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0085] The communication bus 502 is used to enable communication between these components.

[0086] The user interface 503 may include a display screen, and optionally, the user interface 503 may also include a standard wired interface or a wireless interface.

[0087] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0088] This application also discloses a computer-readable storage medium storing multiple instructions adapted for loading by a processor to execute the aforementioned method for early warning of abnormal temperature in a switchgear.

[0089] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure.

[0090] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for early warning of abnormal temperature in switchgear, characterized in that, The method includes the following steps: Step S1: Collect continuous temperature data of key parts of the switchgear and preprocess it to obtain a temperature time series; Step S2: Divide the temperature time series into sliding window segments to obtain micro-window data segments; Step S3: Calculate the temperature rise rate score for each micro-window data segment; Step S4: Calculate the temperature fluctuation range score for each micro-window data segment; Step S5: Based on the temperature rise rate score and temperature fluctuation amplitude score, construct an anomaly scoring function to obtain the comprehensive anomaly score for each micro-window; Step S6: Based on the comprehensive anomaly scores, make an over-limit judgment, trigger a graded warning, and output structured graded warning information.

2. The switchgear temperature anomaly early warning method as described in claim 1, characterized in that, Step S1 includes: Key components of the switchgear include: switchgear surface, instrument transformers, circuit breaker contacts, busbar interfaces, and cable terminals; Preprocessing includes: data cleaning, noise reduction, normalization, and sequence segmentation.

3. The switchgear temperature anomaly early warning method as described in claim 1, characterized in that, Step S2 includes: Construct a sliding window and set the duration of a single window to 1. The step length is Window length Take a time interval between 10 and 60 seconds, with a step size of [missing information]. Less than or equal to 1 / 2 of the window length; Given a preprocessed temperature-time series, denoted as Each sampling point For the first Temperature observation value at that time; By using a sliding window operation, the temperature time series can be divided into a series of lengths. The subsequences, each representing a micro-window data segment; The starting index of the sliding window is constructed as follows: in For the first The starting position of each window The number of windows that can be generated, satisfying the constraints. ; For each valid window, it can be represented as: in Indicates the first Each micro-window data segment; each window This constitutes a continuous temperature change segment; Display window The first temperature sample value.

4. The switchgear temperature anomaly early warning method as described in claim 1, characterized in that, Step S3 includes: For micro-window data segments The rate of temperature rise is defined as the difference between the temperature at the end of the window and the initial temperature divided by the window duration, i.e.: in, For the first The temperature rise rate score for each window; A linear curve is fitted using least squares within the window, and its slope is calculated as a score for the rate of temperature rise, such as: in, This represents a function that extracts the slope parameter of the fitted line; This represents the linear least squares fitting operation; Set a reference threshold for the rate of temperature rise scoring ,when Exceeding this threshold can be used as an indicator of a rapid increase in temperature, as follows: in and These represent the mean and standard deviation of the historical window rate, respectively. The adjustment parameters are used to control sensitivity.

5. The switchgear temperature anomaly early warning method as described in claim 4, characterized in that, Step S4 includes: Micro Window Data Segment The temperature fluctuation range score is defined as follows: in For the first The temperature fluctuation range score for each window represents the maximum temperature deviation within that window. and Indicates the maximum and minimum temperatures within the window; Set a volatility threshold for volatility rating Only when Only when a certain time is the state considered unstable, and the fluctuation threshold is set using a dynamic distribution modeling approach: in and These represent the mean and standard deviation of the historical window temperature fluctuation range scores, respectively. To adjust the parameters.

6. The switchgear temperature anomaly early warning method as described in claim 5, characterized in that, Step S5 includes: A weighted linear scoring structure is used to construct an anomaly scoring function for the first... For each micro-window data segment, the anomaly scoring is defined as follows: in For the first The abnormal rating value of each window, and Scoring for adjustable temperature rise rate Score based on temperature fluctuation amplitude Weighting coefficients; By reference threshold and fluctuation threshold The scoring results are then corrected to obtain the final comprehensive anomaly score, as follows: in Indicates the first A comprehensive anomaly score for each micro-window.

7. The switchgear temperature anomaly early warning method as described in claim 1, characterized in that, Step S6 includes: If the abnormal score values ​​of several consecutive windows all exceed the set threshold, it is considered that the current device has a real abnormal trend, triggering an alert; let the score threshold for the alert be... The number of consecutive windows is The alarm judgment conditions are as follows: in For indicator functions, it means the first... Does the score of each window exceed the threshold, and does the cumulative number of windows that continuously meet the conditions reach [a certain threshold]? Timely triggering of warnings; The warning signals are divided into multi-level response modes, i.e., graded warnings, and output is classified according to the score range or the degree of continuous exceedance. Specifically, this includes: Define multiple rating level ranges: If the score satisfies: If this occurs, a Level 1 warning will be triggered, indicating that there are signs of temperature rise in key parts of the switchgear. like: If the temperature rises rapidly, it is a Level 2 warning, which means that the critical parts of the switchgear are heating up rapidly. like: This triggers a Level 3 warning, indicating a severe temperature anomaly in critical components of the switchgear; among which... and To further classify and differentiate thresholds; The warning reset setting is described as follows: in To reset the threshold, To maintain the length of the decline window, which represents the time range within which the score declines continuously and stably.

8. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to enable the electronic device to perform the switch cabinet temperature anomaly early warning method as described in any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a computer, perform the switch cabinet temperature anomaly early warning method as described in any one of claims 1-7.

Images