A method for adaptive temperature sensing adjustment of a magnetically controlled column switch

Abstract

This application relates to a temperature-sensing adaptive adjustment method for a magnetically controlled column switch, comprising: periodically collecting the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature; preprocessing the electromagnetic coil surface temperature, the internal temperature of the operating mechanism housing, and the ambient temperature to generate a timestamp-aligned three-dimensional temperature dataset; inputting the three-dimensional temperature dataset into a pre-constructed temperature feature model to extract temperature change trend features; inputting the temperature change trend features into a trained temperature state level judgment model to obtain the temperature state level of the magnetically controlled column switch for the next cycle; and adaptively adjusting the operating parameters of the magnetically controlled column switch based on the temperature state level of the magnetically controlled column switch for the next cycle and in conjunction with preset operating parameter adjustment rules. This application significantly improves the operational reliability and equipment lifespan of the magnetically controlled column switch, while reducing the risk of overheating failures and maintenance costs.

Description

Technical Field

[0001] This application relates to the field of power system technology, and in particular to a temperature-sensing adaptive adjustment method for a magnetically controlled pole-mounted switch. Background Technology

[0002] In existing power distribution automation systems, magnetically controlled pole-mounted switches are widely used on overhead lines due to their advantages such as compact structure, reliable operation, and low maintenance. These switches typically employ electromagnetic operating mechanisms for opening and closing operations, and their performance depends on the driving capability of the electromagnetic coil and the response characteristics of the mechanical transmission system.

[0003] Because electromagnetic coils generate heat during operation, and their resistance changes with temperature, their electrical parameters vary under different ambient temperatures. To ensure proper switch operation, current technologies typically use fixed voltage or current driving methods to control the electromagnetic coil. While this control method is simple and easy to implement, it has limitations in practical applications because it doesn't consider the influence of ambient temperature on coil impedance and magnetic field strength. Specifically, in low-temperature environments, the lower resistance of the electromagnetic coil may lead to excessive starting current, increasing the power supply load and even triggering protection mechanisms. Conversely, in high-temperature environments, the increased coil resistance weakens the magnetic field strength, affecting the switch's opening and closing speed and operational reliability. Furthermore, significant differences in climate conditions across regions make fixed control strategies insufficient to meet the demands of complex and ever-changing application scenarios.

[0004] Therefore, there is an urgent need for a temperature-sensing adaptive adjustment method for magnetically controlled column switches. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, this application provides a temperature-sensing adaptive adjustment method for a magnetically controlled column switch, which solves the technical problem in the prior art that it is impossible to dynamically adjust the control parameters of the magnetically controlled column switch to adapt to temperature changes without changing the basic structure of the existing magnetically controlled column switch.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the main technical solutions adopted in this application include:

[0009] This application provides a method for adaptive temperature sensing adjustment of a magnetically controlled column switch, including:

[0010] S100: Periodically collect the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature using a multi-dimensional temperature sensor. Then, preprocess the electromagnetic coil surface temperature, the internal temperature of the operating mechanism housing, and the ambient temperature to generate a three-dimensional temperature dataset with timestamp alignment.

[0011] S200. Input the three-dimensional temperature dataset into a pre-built temperature feature model to extract temperature change trend features; and input the temperature change trend features into a trained temperature state level judgment model to obtain the temperature state level of the magnetic control column switch for the next cycle.

[0012] S300. Based on the temperature status level of the magnetically controlled column switch in the next cycle, and in conjunction with the preset operating parameter adjustment rules, the operating parameters of the magnetically controlled column switch are adaptively adjusted.

[0013] Optionally, in some embodiments of this application, the operating parameters include the driving voltage of the electromagnetic coil of the magnetically controlled column switch and the operating interval time.

[0014] Optionally, in step S100, preprocessing the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature to generate a timestamp-aligned three-dimensional temperature dataset specifically includes:

[0015] S110. Perform wavelet denoising processing on the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature to obtain the denoised temperature data.

[0016] S120. Align the denoised temperature data with the sampling timestamps of different sensors using cubic spline interpolation to obtain a timestamp-aligned three-dimensional temperature dataset.

[0017] Optionally, in some embodiments of this application, in step S300, the temperature status level includes normal state, warning state, overheating state, and dangerous state;

[0018] The preset operating parameter adjustment rules include:

[0019] No adjustments are needed when the temperature is normal.

[0020] When the temperature condition is a warning, reduce the driving voltage of the electromagnetic coil by 5% and extend the operation interval by 10%.

[0021] When the temperature is overheated, reduce the driving voltage of the electromagnetic coil by 15% and extend the operation interval by 30%.

[0022] If the temperature condition becomes dangerous, immediately disconnect the power supply to the electromagnetic coil.

[0023] Optionally, in some embodiments of this application, in step S200, the three-dimensional temperature dataset is input into a pre-constructed temperature feature model, and the dynamic temperature feature model extracted from the temperature change trend features is converted into a dynamic trend analysis model based on a sliding time window, including:

[0024] S210. Divide the three-dimensional temperature dataset into sliding windows and perform first-order difference within each sliding window to obtain the temperature change rate vector for each dimension.

[0025] S220. Perform second-order difference on the three-dimensional temperature data after first-order difference to obtain the temperature change acceleration characteristic index of each dimension.

[0026] S230. The three-dimensional temperature dataset is averaged to obtain the moving average temperature value for each dimension, and the moving average value is compared with the original temperature value at the end of the current window to obtain the temperature offset.

[0027] S240. The temperature change rate vector, temperature change acceleration feature index and temperature offset of each dimension are spliced ​​together to generate temperature change trend features.

[0028] Optionally, in some embodiments of this application, S240 includes:

[0029] S241. Normalize the temperature change rate vector, temperature change acceleration characteristic index and temperature offset of each dimension respectively.

[0030] S242. Principal component analysis is used to reduce the dimensionality of the normalized multidimensional features to obtain temperature change trend features containing the first principal component and the second principal component; the first principal component corresponds to the overall temperature rise trend, and the second principal component corresponds to local abnormal fluctuations.

[0031] Optionally, in some embodiments of this application, in step S200, the temperature and heat state level judgment model is a hybrid judgment model based on support vector machine combined with multi-level fuzzy logic reasoning, and the temperature and heat state level judgment model includes:

[0032] The feature input layer is used to receive temperature change trend features;

[0033] A primary classifier for support vector machines is used for nonlinear classification using radial basis functions, and outputs preliminary classification results and confidence scores.

[0034] The fuzzy inference preprocessing layer is used to map confidence levels to membership levels;

[0035] A multi-level fuzzy logic reasoning layer is used to process the transition interval between temperature and heat state levels through dynamic fuzzy rules, and to adjust the membership degree in combination with environmental parameters to obtain multi-level fuzzy reasoning results.

[0036] The decision fusion module is used to weight and fuse the preliminary classification results with the multi-level fuzzy reasoning results to output the final temperature status level.

[0037] Optionally, in some embodiments of this application, the dynamic fuzzy rules in the multi-level fuzzy logic inference layer include multiple parallel fuzzy rule nodes, each corresponding to a transition interval between different temperature state levels, including:

[0038] The normal-warning transition node is defined as follows: the first principal component is greater than or equal to 0.3 and less than 0.5, and the rate of change of the first principal component is ≤0.1.

[0039] Warning - Overheating transition node: the first principal component is greater than or equal to 0.5 and the second principal component is greater than 0.5;

[0040] Overheating-dangerous transition point, the second principal component is greater than 0.6 and lasts for more than 2 cycles.

[0041] Optionally, in some embodiments of this application, each of the plurality of parallel fuzzy rule nodes is also dynamically adjusted in conjunction with the environmental humidity parameter:

[0042] When the ambient humidity is greater than the preset humidity threshold, the threshold value of the principal component in the membership trigger condition corresponding to each transition node is reduced by 0.1.

[0043] Optionally, in some embodiments of this application, the method further includes:

[0044] When the temperature condition is dangerous in the next cycle, after cutting off the power supply, continue to execute steps S100-S300. When the temperature condition drops to the normal state and lasts for three cycles, restore the power supply. When the temperature condition lasts for three cycles and is in a dangerous state, generate equipment maintenance alarm information and send it to maintenance personnel.

[0045] (III) Beneficial Effects

[0046] This application provides a temperature-sensing adaptive adjustment method for a magnetically controlled column switch. It employs a multi-dimensional temperature sensor for real-time monitoring, a temperature feature model, and a hybrid temperature state level judgment model based on support vector machines and fuzzy logic. Compared with existing technologies, it can accurately predict the temperature state level of the magnetically controlled column switch and adaptively adjust the operating parameters, thereby improving the stability of equipment operation, reducing the risk of overheating failure, and extending service life. Attached Figure Description

[0047] Figure 1 This is a schematic flowchart of a temperature-sensing adaptive adjustment method for a magnetically controlled column switch according to an embodiment of this application. Detailed Implementation

[0048] To better explain and facilitate understanding of this application, the following detailed description of the application is provided in conjunction with the accompanying drawings and specific embodiments.

[0049] Magnetically controlled pole-mounted switches are high-voltage power distribution switchgear based on electromagnetic drive principles. They mainly consist of core components such as electromagnetic coils, operating mechanisms, and arc-extinguishing devices. They achieve rapid closing operations through electromagnetic force, featuring fast operation and high reliability. They are used in power distribution line automation, industrial power systems, new energy power generation grid connection, and harsh environments such as high temperature and humidity. However, in actual operation, the electromagnetic coils and operating mechanisms of magnetically controlled pole-mounted switches are prone to overheating under frequent operation or high current conditions. Current technologies rely solely on fixed threshold protection, lacking dynamic adjustment capabilities. This is particularly problematic in harsh environments such as high temperature and humidity, potentially leading to equipment damage such as insulation aging and contact sintering, and also resulting in high maintenance costs due to outdated monitoring methods. To address these issues, this application proposes a temperature-sensing adaptive adjustment method for magnetically controlled column switches. This method addresses problems in existing technologies, such as overheating damage and low operational reliability caused by incomplete temperature monitoring and delayed adjustment. It utilizes multi-dimensional temperature sensors to collect real-time temperature data from the electromagnetic coil surface, the inside of the operating mechanism housing, and the environment. A three-dimensional temperature dataset is constructed using wavelet denoising and timestamp alignment. Temperature change features are extracted based on sliding window dynamic trend analysis and principal component analysis, and a hybrid model combining support vector machine and multi-level fuzzy logic inference is employed to predict the temperature state level. Finally, operating parameters are adaptively adjusted according to different levels. This method achieves accurate prediction and proactive intervention of the temperature rise trend of the magnetically controlled column switch, significantly improving the stability and lifespan of the equipment while reducing the risk of overheating failures and maintenance costs.

[0050] To better understand the above technical solutions, exemplary embodiments of this application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application can be understood more clearly and thoroughly, and that the scope of this application can be fully conveyed to those skilled in the art.

[0051] Figure 1 This is a schematic flowchart illustrating a temperature-sensing adaptive adjustment method for a magnetically controlled column switch according to an embodiment of this application. Figure 1 As shown, the temperature-sensing adaptive adjustment method includes:

[0052] Step S100: Periodically collect the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature using a multi-dimensional temperature sensor. Then, preprocess the data of the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature to generate a three-dimensional temperature dataset with timestamp alignment.

[0053] In the specific implementation process, a high-precision, high-temperature resistant K-type thermocouple sensor is selected to collect the surface temperature of the electromagnetic coil. Thermally conductive silicone is used to tightly attach the thermocouple sensor to the surface of the electromagnetic coil, and high-temperature resistant insulating tape is used for fixation to ensure full contact between the sensor and the coil surface and to obtain accurate surface temperature data. At the same time, considering that the internal space of the operating mechanism box is relatively enclosed, a platinum resistance temperature sensor is selected. The sensor is fixed inside the box by a bracket to ensure that its installation is firm and does not affect the normal operation of the operating mechanism. Similarly, in order to obtain accurate ambient temperature data, a digital temperature sensor is installed at a certain distance from the switch on the magnetic control column in a well-ventilated location free from heat sources.

[0054] When collecting temperature data, a uniform periodic sampling frequency of 1 minute is set.

[0055] Furthermore, the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature are preprocessed to generate a timestamp-aligned three-dimensional temperature dataset, specifically including:

[0056] Step S110: Perform wavelet denoising processing on the surface temperature of the electromagnetic coil, the internal temperature of the operating mechanism housing, and the ambient temperature to obtain the denoised temperature data.

[0057] When performing wavelet denoising, the db4 wavelet basis function is selected based on the characteristics of the temperature signal, specifically including the following steps:

[0058] Determination of the number of decomposition layers: Through trial and error combined with signal spectrum analysis, the number of decomposition layers was determined to be 4 layers, ensuring that high-frequency noise can be effectively removed without losing the characteristic information of temperature changes;

[0059] Thresholding: A soft thresholding method is used to process the wavelet coefficients of each layer. The threshold calculation formula is as follows:

[0060]

[0061] Where σ is the noise standard deviation, N is the signal length, and λ is the threshold;

[0062] Signal reconstruction: Perform inverse transform on the processed wavelet coefficients to reconstruct the denoised temperature signal.

[0063] Step S120: Align the denoised temperature data with the sampling timestamps of different sensors using cubic spline interpolation to obtain a timestamp-aligned three-dimensional temperature dataset.

[0064] Specifically, using the UTC time provided by a high-precision clock server as the benchmark, the local timestamps of each sensor are unified into a unified time coordinate system. When performing cubic spline interpolation, a cubic spline algorithm library with high interpolation accuracy and computational efficiency, such as SciPy's CubicSpline, is selected. Piecewise cubic polynomials are constructed for each sensor data to ensure the smoothness of the interpolation curve and the continuity of the first derivative. At the same time, in regions with drastic temperature changes, such as before and after switching actions, the density of interpolation nodes is increased to improve local accuracy.

[0065] Furthermore, the local outlier factor algorithm is applied to identify and label outliers in the interpolation process, and robust interpolation strategies, such as weighted spline interpolation, are used to replace outliers. Then, the interpolated three-dimensional temperature data are aligned according to a unified timestamp to form a structured dataset.

[0066] Through the above steps, the final generated three-dimensional temperature dataset will have the characteristics of accurate timestamp alignment, reliable data quality, and standardized structure, providing a high-quality data foundation for subsequent temperature feature extraction and state assessment.

[0067] Step S200: Input the three-dimensional temperature dataset into the pre-constructed temperature feature model to extract temperature change trend features; and input the temperature change trend features into the trained temperature state level judgment model to obtain the temperature state level of the magnetic control column switch for the next cycle.

[0068] The temperature and heat status levels include normal status, warning status, overheating status, and dangerous status.

[0069] In step S200, the three-dimensional temperature dataset is input into a pre-constructed temperature feature model, and the dynamic temperature feature model extracted from the temperature change trend features is transformed into a dynamic trend analysis model based on a sliding time window, including:

[0070] Step S210: Divide the three-dimensional temperature dataset into sliding windows and perform first-order difference within each sliding window to obtain the temperature change rate vector for each dimension.

[0071] Specifically, the continuously acquired three-dimensional temperature data is divided into fixed time windows, such as every 5 minutes, and an overlapping sliding method is used, with each slide lasting 3 minutes, to capture the continuity and local fluctuations of temperature changes. First-order differencing is performed on the temperature sequence for each dimension to obtain the temperature change rate vector.

[0072] Step S220: Perform second-order difference on the three-dimensional temperature data after first-order difference to obtain the temperature change acceleration characteristic index of each dimension.

[0073] After performing the first-order difference, a second-order difference is performed to calculate the rate of change of temperature, i.e., the acceleration of temperature change in each dimension. The inflection points of the temperature change trend are captured, and the statistical characteristics of the acceleration are extracted: mean, standard deviation, peak value, and number of zero crossings. The acceleration / deceleration phases of the temperature rise / fall trend are also distinguished. Positive acceleration corresponds to a rapid temperature rise, negative acceleration corresponds to a slowdown in temperature rise, and zero acceleration corresponds to a steady-state operating condition.

[0074] Step S230: Average the three-dimensional temperature dataset to obtain the moving average temperature value for each dimension, and compare the moving average value with the original temperature value at the end of the current window to obtain the temperature offset.

[0075] Step S230 specifically includes:

[0076] Implement a three-level moving average:

[0077] Short-term average (window size = 5): Captures instantaneous fluctuations;

[0078] Medium-term average (window size = 10): reflects the trend of normal operation;

[0079] Long-term average (window size = 20): Identifies seasonal or aging effects;

[0080] An exponentially weighted moving average is used to enhance sensitivity to recent changes;

[0081] Calculate the deviation between the original temperature value and the moving average value at each scale.

[0082] Step S240: Perform multi-dimensional feature splicing on the temperature change rate vector, temperature change acceleration feature index and temperature offset of each dimension to generate temperature change trend features.

[0083] Step S240 includes:

[0084] Step S241: Normalize the temperature change rate vector, temperature change acceleration characteristic index, and temperature offset of each dimension respectively.

[0085] Specifically, this includes standardizing the temperature change rate vector using Z-score, normalizing the temperature change acceleration characteristic index using quantiles to enhance robustness, and applying hyperbolic tangent transform to the temperature offset to compress the influence range of outliers.

[0086] Step S242: Principal component analysis is used to reduce the dimensionality of the normalized multidimensional features to obtain temperature change trend features containing the first principal component and the second principal component; the first principal component corresponds to the overall temperature rise trend, and the second principal component corresponds to local abnormal fluctuations.

[0087] In this embodiment, after normalizing the temperature change rate vector, temperature change acceleration characteristic index, and temperature offset, principal component analysis (PCA) is used to reduce the dimensionality of these multidimensional features. PCA first preprocesses and enhances these normalized multidimensional features by eliminating the influence of mean shift through feature centering and reducing the correlation between features through feature whitening, making the data distribution more suitable for the application conditions of PCA.

[0088] Optionally, to adapt to dynamic changes in equipment status, an incremental PCA algorithm is employed for online learning. Through adaptive learning rate and forgetting factor mechanisms, the principal components are ensured to quickly capture real-time changes in the temperature feature space. The number of principal components is determined using a combination of scree plot analysis and variability deformation thresholding, aiming to retain 95% of data variability. The optimal dimensionality reduction dimension is automatically identified, balancing data simplification and information preservation.

[0089] The first principal component PC1 after dimensionality reduction corresponds to the overall temperature rise trend, mainly driven by features such as the average temperature, long-term rate of change, and moving average. Positive and negative values ​​of PC1 represent the overall temperature rise or fall trend, respectively, and its absolute value reflects the strength of the trend, which can be directly used to predict the future temperature trend of the equipment. For example, a positive value with an increasing absolute value indicates an increase in equipment load or a decrease in heat dissipation efficiency, which may lead to a continuous rise in temperature; a negative value may indicate a decrease in load or that the heat dissipation system is effective.

[0090] The second principal component, PC2, corresponds to localized abnormal fluctuations, dominated by characteristics such as temperature change acceleration, short-term offset, and peak factor. High values ​​of PC2 indicate the presence of localized abnormal heating or cooling, and its fluctuation frequency reflects the periodicity of the anomaly. By analyzing the correlation between PC2 and PC1, the system can distinguish between systemic temperature rises and localized faults. For example, when PC1 is normal but PC2 is abnormally elevated, it is more likely to indicate localized poor contact or heat dissipation blockage.

[0091] Through the above refined implementation, the use of principal component analysis not only effectively compresses the data dimensions, but more importantly, it extracts temperature features with clear physical meaning, enabling the system to simultaneously monitor the overall temperature rise trend and local abnormal fluctuations. This greatly improves the accuracy and timeliness of fault warnings, providing solid technical support for the intelligent operation and maintenance of magnetically controlled pole-mounted switches.

[0092] Furthermore, the temperature and heat state level judgment model is a hybrid judgment model based on support vector machine combined with multi-level fuzzy logic reasoning, and the temperature and heat state level judgment model includes:

[0093] The feature input layer is used to receive temperature change trend features;

[0094] A primary classifier for support vector machines is used for nonlinear classification using radial basis functions, and outputs preliminary classification results and confidence scores.

[0095] The fuzzy inference preprocessing layer is used to map confidence levels to membership levels;

[0096] A multi-level fuzzy logic reasoning layer is used to process the transition interval between temperature and heat state levels through dynamic fuzzy rules, and to adjust the membership degree in combination with environmental parameters to obtain multi-level fuzzy reasoning results.

[0097] The decision fusion module is used to weight and fuse the preliminary classification results with the multi-level fuzzy reasoning results to output the final temperature status level.

[0098] In the multi-level fuzzy logic inference layer, the dynamic fuzzy rules include multiple parallel fuzzy rule nodes, each corresponding to a transition interval between different temperature state levels, including:

[0099] The normal-warning transition node is defined as follows: the first principal component is greater than or equal to 0.3 and less than 0.5, and the rate of change of the first principal component is ≤0.1.

[0100] Warning - Overheating transition node: the first principal component is greater than or equal to 0.5 and the second principal component is greater than 0.5;

[0101] Overheating-dangerous transition point, the second principal component is greater than 0.6 and lasts for more than 2 cycles.

[0102] In the multiple parallel fuzzy rule nodes, each fuzzy rule node is also dynamically adjusted in conjunction with the environmental humidity parameter:

[0103] When the ambient humidity is greater than the preset humidity threshold, the threshold of the principal component value in the membership trigger condition corresponding to each transition node is reduced by 0.1. The preset humidity threshold is 75%.

[0104] In the specific implementation process, the feature input layer, as the "entry point" of the model, is responsible for receiving the temperature change trend features extracted by the temperature feature model, namely the first and second principal component data after principal component analysis. Before receiving the data, this layer preprocesses the input features, including data format verification, outlier detection and correction, to ensure the standardization and validity of the input data and provide a reliable basis for subsequent processing.

[0105] The Support Vector Machine (SVM) primary classifier utilizes the powerful nonlinear mapping capability of the radial basis function (RBF) to perform preliminary classification of input temperature change trend features. During the training phase, the model is optimized using a large amount of historical temperature data and corresponding state labels, adjusting parameters of the RBF, such as kernel width, to improve classification accuracy. After training, the model classifies new input feature data, outputting a preliminary judgment of the temperature state level, along with a confidence score for this judgment. The confidence score reflects the model's certainty regarding the classification result, providing an important reference for subsequent processing.

[0106] The fuzzy inference preprocessing layer receives the confidence data output from the support vector machine and converts it into membership degrees that the fuzzy logic system can process. This layer uses a pre-defined nonlinear mapping function to assign membership degrees to data at different temperature levels based on the confidence level. For example, a higher confidence level will make the membership degree of the data closer to 1 at the corresponding temperature level, while a lower confidence level will result in a relatively lower membership degree, thus achieving the conversion from numerical confidence to fuzzy membership degrees.

[0107] The multi-level fuzzy logic inference layer is the core decision-making unit, processing the transition intervals between temperature and heat status levels through multiple parallel fuzzy rule nodes. At the normal-warning transition node, when the first principal component is between 0.3 and 0.5 and its rate of change does not exceed 0.1, it indicates that the equipment temperature has begun to rise but has not yet reached a significant abnormal level, potentially triggering a warning state judgment. The warning-overheating transition node focuses on situations where the first principal component is greater than or equal to 0.5 and the second principal component is greater than 0.5. At this point, the equipment not only shows a significant overall temperature rise trend but also exhibits local abnormal fluctuations, indicating a possible overheating state. The overheating-danger transition node is even more stringent; when the second principal component is greater than 0.6 and persists for more than two cycles, it indicates that the local abnormality of the equipment is severe and has not been alleviated, potentially facing a dangerous state. Furthermore, this layer fully considers the impact of environmental factors on the equipment state. When the ambient humidity exceeds a preset threshold, it automatically lowers the threshold value of the corresponding principal component for each transition node (by 0.1). Because high humidity environments may affect the equipment's heat dissipation performance, making the equipment more prone to temperature anomalies, dynamically adjusting the threshold allows the model to more accurately reflect the true state of the equipment.

[0108] The decision fusion module integrates the preliminary classification results from the support vector machine (SVM) and the results from multi-level fuzzy logic inference, employing a weighted fusion strategy to output the final temperature and thermal status level. This module assigns appropriate weights to the two results based on historical data and the actual application scenario. For example, in the early stages of equipment operation, the classification results from the SVM based on data features may be more valuable; while in the aging stage of the equipment, the judgment based on multi-level fuzzy logic inference combined with environmental factors may be more reliable. Through weighted fusion, the model can leverage the strengths of both methods, avoiding the limitations of a single approach, and ultimately outputting an accurate and reliable temperature and thermal status level for the magnetically controlled column switch over the next cycle, providing crucial information for adjusting equipment operating parameters.

[0109] Step S300: Based on the temperature status level of the magnetically controlled column switch in the next cycle, and in conjunction with preset operating parameter adjustment rules, adaptively adjust the operating parameters of the magnetically controlled column switch. The operating parameters include the driving voltage of the electromagnetic coil of the magnetically controlled column switch and the operating interval time.

[0110] The preset operating parameter adjustment rules include:

[0111] No adjustments are needed when the temperature is normal.

[0112] When the temperature condition is a warning, reduce the driving voltage of the electromagnetic coil by 5% and extend the operation interval by 10%.

[0113] When the temperature is overheated, reduce the driving voltage of the electromagnetic coil by 15% and extend the operation interval by 30%.

[0114] If the temperature condition becomes dangerous, immediately disconnect the power supply to the electromagnetic coil.

[0115] During adaptive adjustment, this process, based on preset operating parameter adjustment rules, employs a strategy combining graded response, smooth transition, and effect verification to construct a closed-loop control system, specifically including:

[0116] First, the temperature and heat status levels are accurately identified, and the four status levels—normal, warning, overheating, and danger—are mapped to different control strategies. Each status level corresponds to a specific temperature characteristic range. By real-time monitoring of the values ​​and rates of change of the first principal component PC1 and the second principal component PC2, combined with the membership degree calculation using fuzzy logic inference, the most probable status level and its confidence level are determined.

[0117] Optionally, to enhance robustness, this embodiment also employs a sliding window voting mechanism, where parameter adjustment is triggered only if the state determination is consistent within two consecutive sampling periods.

[0118] When in a warning state, a PID controller is used to reduce the driving voltage of the electromagnetic coil by 5%, and closed-loop feedback is used to ensure that the voltage is stable at the target value; the operation interval is extended by 10%, and a time slice dynamic allocation algorithm is used to ensure that the critical operation response is not affected while the interval is extended.

[0119] When the system enters an overheating state, the drive voltage will be reduced by 15%, employing a multi-step gradual adjustment strategy, with each step reducing the voltage by 3% to avoid voltage surges impacting the equipment. The operation interval will be extended by 30%, and an operation priority ranking mechanism will be implemented to ensure that critical operations are executed first. Simultaneously, auxiliary cooling equipment (such as fans or water cooling systems) will be automatically activated, and the cooling power will be dynamically adjusted according to temperature changes.

[0120] When a dangerous condition is detected, the power supply to the electromagnetic coil is immediately cut off via a fast solid-state relay with a response time of less than 10ms. At the same time, the temperature data, operation records, and system parameters for 30 minutes before and after the fault are completely saved, and a detailed fault report is generated.

[0121] In addition, embodiments of this application also include:

[0122] When adjusting the drive voltage, an S-curve control is used, with a slower adjustment rate at the beginning and end and a faster adjustment rate in the middle. The operation interval is extended gradually, with each cycle not exceeding 5% of the current value, and the equipment response is monitored during the adjustment process.

[0123] Through the above design, step S300 achieves complete closed-loop control from state perception to parameter adjustment, enabling the magnetically controlled column switch to optimize operating parameters according to temperature conditions, maximizing operating efficiency while ensuring equipment safety. Simultaneously, the introduction of environmental compensation, smooth transition, and adaptive learning mechanisms significantly improves control accuracy and adaptability, while multiple safety mechanisms ensure the system's reliability under various complex operating conditions.

[0124] Furthermore, this application also includes: when the temperature state is dangerous in a future cycle, after cutting off the power supply, continue to execute steps S100-S300; when the temperature state drops to a normal state and lasts for three cycles, restore the power supply; when the temperature state lasts for three cycles and is a dangerous state, generate equipment maintenance alarm information and send it to maintenance personnel.

[0125] This application discloses a temperature-sensing adaptive adjustment method for a magnetically controlled pole-mounted switch. By employing multi-dimensional temperature measurement and scientific data preprocessing, the accuracy and reliability of the collected data are ensured. Simultaneously, this application utilizes a dynamic trend analysis model and principal component analysis to uncover key temperature features, and combines a hybrid model of support vector machine and fuzzy logic reasoning to reliably determine the temperature state level. Based on the state-level response, operating parameters are adjusted, and environmental compensation and smooth transitions are considered to ensure safe and efficient equipment operation. This improves equipment operational stability, reduces the risk of overheating failures, and extends service life. Furthermore, it achieves automated operation and maintenance, reducing manual intervention, improving operation and maintenance efficiency and management level, and promoting the intelligent upgrading of power equipment operation and maintenance.

[0126] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0127] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0128] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first and second features are in direct contact, or that they are in indirect contact through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0129] In the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0130] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make modifications, alterations, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for adaptive temperature sensing adjustment of a magnetically controlled column switch, characterized in that, The method comprises the following steps: S100, periodically collecting electromagnetic coil surface temperature, operating mechanism box internal temperature and environment temperature through multi-dimensional temperature sensor, and performing data preprocessing on the electromagnetic coil surface temperature, operating mechanism box internal temperature and environment temperature to generate a timestamp-aligned three-dimensional temperature data set; S200, inputting the three-dimensional temperature data set into a pre-constructed temperature feature model to extract temperature change trend features; and inputting the temperature change trend features into a trained thermal state level judgment model to obtain the thermal state level of the magnetic control column switch in the future period; S300, according to the thermal state level of the magnetic control column switch in the future period, combining a preset operation parameter adjustment rule, and adaptively adjusting the operation parameters of the magnetic control column switch; In S200, the thermal state level judgment model is a hybrid judgment model based on support vector machine combined with multi-level fuzzy logic reasoning, and the thermal state level judgment model comprises: a feature input layer for receiving temperature change trend features; a support vector machine primary classifier for performing nonlinear classification using a radial basis function and outputting a preliminary classification result and a confidence level; a fuzzy reasoning preprocessing layer for mapping the confidence level to a membership degree; a multi-level fuzzy logic reasoning layer for processing the transition interval between thermal state levels through dynamic fuzzy rules and adjusting the membership degree in combination with environmental parameters to obtain a multi-level fuzzy reasoning result; a decision fusion module for weighting and fusing the preliminary classification result and the multi-level fuzzy reasoning result to output the final thermal state level.

2. The method of warm induction adaptive adjustment of a magnetic column on switch according to claim 1, characterized in that, The operation parameters include the driving voltage of the electromagnetic coil of the magnetic control column switch and the operation interval time.

3. The method of warm induction self-adaptive adjustment of the magnetic control column on-switch according to claim 1, characterized in that, In S100, the data preprocessing of the electromagnetic coil surface temperature, operating mechanism box internal temperature and environment temperature to generate a timestamp-aligned three-dimensional temperature data set comprises: S110, performing wavelet denoising processing on the electromagnetic coil surface temperature, operating mechanism box internal temperature and environment temperature to obtain denoised temperature data; S120, aligning the sampling timestamps of different sensors by using a cubic spline interpolation method to obtain a timestamp-aligned three-dimensional temperature data set.

4. The method of claim 2, wherein the magnetic column on switch is a magnetic tunnel junction (MTJ) on switch. In S300, the thermal state level includes normal state, warning state, overheating state and dangerous state; The preset operation parameter adjustment rule includes: when the thermal state is normal, no adjustment is made; when the thermal state is warning, the driving voltage of the electromagnetic coil is reduced by 5%, and the operation interval time is extended by 10%; when the thermal state is overheating, the driving voltage of the electromagnetic coil is reduced by 15%, and the operation interval time is extended by 30%; when the thermal state is dangerous, the power supply of the electromagnetic coil is immediately cut off.

5. The method of warm induction self-adaptive adjustment of the magnetic control column on-switch according to claim 1, characterized in that, In S200, the temperature dynamic feature model in the temperature change trend features input into the pre-constructed temperature feature model is a dynamic trend analysis model based on a sliding time window, comprising: S210, the three-dimensional temperature data set is divided by a sliding window, and first-order differentiation is performed in each sliding window to obtain a temperature change rate vector of each dimension; S220, second-order differentiation is performed on the three-dimensional temperature data after first-order differentiation to obtain a temperature change acceleration characteristic index of each dimension; S230, the three-dimensional temperature data set is processed by averaging to obtain a temperature moving average value of each dimension, and the moving average value is compared with an original temperature value at the end of the current window to obtain a temperature offset; S240, the temperature change rate vector of each dimension, the temperature change acceleration characteristic index, and the temperature offset are spliced to generate a temperature change trend feature.

6. The method of warm induction self-adaptive adjustment of the magnetic control column on-switch according to claim 5, characterized in that, The S240 includes: S241, the temperature change rate vector of each dimension, the temperature change acceleration characteristic index, and the temperature offset are normalized respectively; S242, principal component analysis is used to reduce the dimensionality of the normalized multi-dimensional features to obtain a temperature change trend feature containing a first principal component and a second principal component; the first principal component corresponds to an overall temperature rise trend, and the second principal component corresponds to a local abnormal fluctuation.

7. The method of warm induction self-adaptive adjustment of the magnetic column on- switch according to claim 1, characterized in that, In the multi-level fuzzy logic reasoning layer, the dynamic fuzzy rule includes a plurality of parallel fuzzy rule nodes corresponding to transition intervals between different thermal state grades, including: a normal-warning transition node, the first principal component is greater than or equal to 0.3 and less than 0.5, and the first principal component change rate is less than or equal to 0.1; a warning-overheating transition node, the first principal component is greater than or equal to 0.5 and the second principal component is greater than 0.5; an overheating-dangerous transition node, the second principal component is greater than 0.6 and lasts for more than 2 cycles.

8. The method of warm induction self-adaptive adjustment of the magnetic column on- switch according to claim 7, characterized in that, In the plurality of parallel fuzzy rule nodes, each fuzzy rule node further combines the environmental humidity parameter for dynamic adjustment: When the environmental humidity is greater than a preset humidity threshold, the threshold value of the principal component number in the membership degree trigger condition corresponding to each transition node is reduced by 0.

1.

9. The method of warm induction self-adaptive adjustment of the magnetic column on- switch according to claim 4, characterized in that, The method further includes: When the thermal state is dangerous in the next cycle, steps S100-S300 are continuously performed after the power supply is cut off, when the thermal state drops to the normal state and lasts for three cycles, the power supply is restored; when the thermal state lasts for three cycles as the dangerous state, device maintenance alarm information is generated and sent to the maintenance personnel.

Images