A line overheat fault analysis method and system based on a temperature measurement protection device

Abstract

The application discloses a kind of line overheat fault analysis method and system based on temperature measuring protection device, method includes: obtaining line surface temperature data sequence and operating current data sequence collected by temperature measuring protection device;According to the theoretical temperature sequence determined by electric heating coupling model and compared with measured value, residual temperature sequence is obtained;Residual temperature sequence is statistically analyzed by sliding window, and abnormal fluctuation interval of residual continuous positive accumulation is identified;The current and temperature subsequence corresponding to the abnormal interval are intercepted to form the electric-thermal response data pair, the local variation coefficient under different time scales is extracted, and the variation coefficient sequence is obtained;The dynamic time warping distance of adjacent abnormal interval variation coefficient sequence is calculated, and if it is greater than the threshold, the suspected data pair is identified for overheat fault type;According to different fault types, corresponding second analysis strategy is used for influence range analysis. The identification precision and diagnosis efficiency of line overheat fault are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of power transmission line condition monitoring technology, and particularly relates to a method and system for analyzing line overheating faults based on a temperature measurement and protection device. Background Technology

[0002] During long-term operation, power transmission lines are prone to localized overheating due to factors such as poor contact at conductor joints, insulation aging, and line overload. If not detected and addressed promptly, overheating can develop into serious faults such as wire breakage and insulation breakdown, and may even lead to safety accidents such as fires. Therefore, monitoring overheating faults in power transmission lines is of significant engineering importance.

[0003] A temperature monitoring device is a type of temperature monitoring equipment installed on power transmission and distribution lines. It typically consists of a temperature sensor, a current transformer, a data acquisition unit, and a wireless communication module. It can collect real-time temperature and current data from the line surface and transmit the data to the monitoring master station. However, existing monitoring methods based on temperature monitoring devices mainly rely on temperature threshold judgments, i.e., issuing an alarm when the line surface temperature exceeds a set threshold. This method has significant shortcomings: firstly, fixed thresholds are difficult to adapt to seasonal environmental temperature changes and line load fluctuations, easily leading to false alarms or missed alarms; secondly, temperature response lags behind current changes due to thermal inertia, making early warning impossible by relying solely on temperature.

[0004] In recent years, some studies have attempted to combine current and temperature data for comprehensive analysis, using the correlation between the two or the establishment of simple thermal circuit models to determine the circuit status. However, most of these methods are limited to analysis on a single time scale, making it difficult to capture the multi-scale dynamic characteristics of electrothermal coupling. Furthermore, existing methods typically focus only on individual abnormal events, lacking consideration of the correlation between adjacent abnormal events, failing to effectively distinguish between different types of overheating faults (such as poor contact, insulation aging, and overload), and also struggling to assess the impact range of the fault on adjacent time periods.

[0005] Therefore, there is an urgent need for a line overheating fault analysis method that can adaptively extract multi-scale features of electrothermal coupling, accurately identify overheating fault types, and assess their impact range, so as to improve the intelligence level and early warning capability of transmission line condition monitoring. Summary of the Invention

[0006] This invention provides a method and system for analyzing line overheating faults based on a temperature measurement and protection device, aiming to solve the technical problems of inaccurate line overheating fault identification, inability to distinguish fault types, and inability to assess the scope of impact in the prior art.

[0007] In a first aspect, the present invention provides a method for analyzing circuit overheating faults based on a temperature measurement and protection device, comprising: Acquire the line surface temperature data sequence and line operating current data sequence collected by the temperature protection device during the preset monitoring period; Based on the line operating current data sequence, a preset electrothermal coupling model is used to determine the theoretical temperature sequence of the line under test, and the theoretical temperature sequence is compared point by point with the line surface temperature data sequence to obtain the residual temperature sequence. A sliding window statistical analysis was performed on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residuals continued to accumulate positively. A current subsequence corresponding to a certain abnormal fluctuation range is extracted from the line operating current data sequence, and a temperature subsequence corresponding to the certain abnormal fluctuation range is extracted from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation range. Determine the local coefficient of variation of a certain electrothermal response data pair at different time scales to obtain a coefficient of variation sequence of the certain electrothermal response data pair; Calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; Determine whether the dynamic time warp distance is greater than a preset distance threshold; If the distance exceeds a preset threshold, the first electrothermal response data pair and the second electrothermal response data pair are defined as the first suspicious data pair and the second suspicious data pair, respectively. The first suspicious data pair and the second suspicious data pair are then identified for overheating fault type according to a preset first analysis strategy to obtain the target overheating fault type identification result.

[0008] Secondly, the present invention provides a circuit overheating fault analysis system based on a temperature measurement and protection device, comprising: The acquisition module is configured to acquire the line surface temperature data sequence and the line operating current data sequence collected by the temperature protection device during a preset monitoring period. The comparison module is configured to determine the theoretical temperature sequence of the line under test based on the line operating current data sequence using a preset electrothermal coupling model, and to compare the theoretical temperature sequence with the line surface temperature data sequence point by point to obtain the residual temperature sequence. The analysis module is configured to perform sliding window statistical analysis on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residual continues to accumulate positively. The interception module is configured to intercept a current subsequence corresponding to a certain abnormal fluctuation range from the line operating current data sequence, and to intercept a temperature subsequence corresponding to the certain abnormal fluctuation range from the line surface temperature data sequence, to form a certain electrothermal response data pair for the certain abnormal fluctuation range. The determination module is configured to determine the local coefficient of variation of a certain electrothermal response data pair at different time scales, and obtain a coefficient of variation sequence of the certain electrothermal response data pair; The calculation module is configured to calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; The judgment module is configured to determine whether the dynamic time warp distance is greater than a preset distance threshold. The identification module is configured to define the first electrothermal response data pair and the second electrothermal response data pair as the first suspicious data pair and the second suspicious data pair respectively if the dynamic time warp distance is greater than a preset distance threshold, and to identify the overheating fault type of the first suspicious data pair and the second suspicious data pair according to a preset first analysis strategy, so as to obtain the target overheating fault type identification result.

[0009] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the circuit overheating fault analysis method based on a temperature measurement protection device according to any embodiment of the present invention.

[0010] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the circuit overheating fault analysis method based on a temperature measurement protection device according to any embodiment of the present invention.

[0011] The circuit overheating fault analysis method and system based on temperature measurement and protection device of this application have the following beneficial effects: By collecting real-time data on the surface temperature and operating current of the line through a temperature protection device, the theoretical temperature is determined based on an electrothermal coupling model and compared with the measured values ​​to obtain a residual temperature sequence. This effectively isolates the temperature changes under normal operating conditions, significantly reduces the interference of ambient temperature fluctuations and load changes on the detection results, and significantly improves the sensitivity and accuracy of overheating fault identification.

[0012] By detecting intervals where the residuals are continuously positive through a sliding window, and combining the dual criteria of positive cumulative area and duration to identify abnormal fluctuation intervals, the limitations of single threshold judgment are overcome, false alarms caused by instantaneous noise are effectively avoided, and the reliability of abnormal interval identification is improved. By using empirical mode decomposition, the electrothermal response is decomposed into oscillatory components at different time scales. The local coefficient of variation of the instantaneous amplitude ratio is extracted, and a multimodal coefficient of variation sequence is constructed. This allows for in-depth mining of the dynamic characteristics of the electrothermal coupling relationship in the time-frequency domain, providing rich feature inputs for comparing the similarity between adjacent intervals. The introduction of dynamic time warping distance quantifies the similarity of fluctuation patterns between adjacent intervals, providing a quantitative criterion for the identification of fault type changes; by extracting multidimensional feature vectors through sliding window cross-correlation analysis and combining them with a support vector machine classifier, it is possible to accurately distinguish between three types of line overheating faults: poor contact, insulation aging, and overload. For different fault types, causal relationship testing, monotonic trend testing, and thermal inertia analysis are used to accurately assess the impact range of overheating faults on adjacent non-abnormal intervals. The thermal hazard propagation time period, thermal hazard precursor time period, and thermal hazard continuous impact time period are marked respectively, forming a complete analysis chain from anomaly identification to type identification to impact assessment, providing direct and clear decision-making basis for line operation and maintenance. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating a method for analyzing line overheating faults based on a temperature measurement and protection device, provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of a line overheating fault analysis system based on a temperature measurement and protection device, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 The diagram shows a flowchart of a line overheating fault analysis method based on a temperature measurement and protection device according to this application.

[0017] like Figure 1 As shown, the method for analyzing circuit overheating faults based on temperature monitoring and protection devices specifically includes the following steps: Step S101: Obtain the line surface temperature data sequence and line operating current data sequence collected by the temperature protection device during the preset monitoring period.

[0018] In this step, the temperature protection device is installed at the monitoring point of the transmission line (such as conductor joints, tension clamps, current-carrying plates, and other parts prone to heat generation). The temperature protection device includes a temperature sensor (such as a thermocouple or infrared temperature sensor), a current transformer (or Rogowski coil), a data acquisition unit, and a wireless communication module. All sensors are synchronized through a unified clock source to ensure that the time alignment accuracy of the data from each channel is better than 10 milliseconds. The sampling frequency is set according to the dynamic characteristics of the line. For slow heating processes, it can be set to once per minute, and for fast heating processes, it can be set to once per second or higher. The sampling duration covers a complete operating cycle of the line, usually not less than 24 hours.

[0019] Step S102: Based on the line operating current data sequence, a preset electrothermal coupling model is used to determine the theoretical temperature sequence of the line under test, and the theoretical temperature sequence is compared point by point with the line surface temperature data sequence to obtain the residual temperature sequence.

[0020] In this step, a multi-order thermal network model of the circuit under test is established. The multi-order thermal network model is composed of multiple thermal resistance elements and multiple thermal capacity elements connected in series or in parallel according to the actual heat transfer path of the circuit under test. The historical current data sequence and the corresponding historical temperature data sequence of the circuit under test under normal operating conditions are obtained as calibration samples. Using the historical current data sequence as input and the historical temperature data sequence as output, a parameter estimation method is used to determine the thermal resistance value of each thermal resistance element and the thermal capacity value of each thermal capacity element in the multi-order thermal network model, so that the deviation between the temperature data sequence output by the multi-order thermal network model under the excitation of the current data sequence and the historical temperature data sequence in the calibration sample is minimized, thereby completing the parameter identification of the multi-order thermal network model and obtaining the electrothermal coupling model. The current values ​​at each moment in the line operating current data sequence are sequentially input into the electrothermal coupling model according to the time sequence. The electrothermal coupling model simulates the process of heat being transferred from the heat source to the line surface step by step according to the thermal resistance and heat capacity values. The temperature of each heat capacity element is calculated recursively from the initial moment. The temperature value of the heat capacity element located at the end of the heat transfer path of the multi-order thermal network model at the current moment is taken as the theoretical temperature value at the current moment. By traversing all sampling moments, the theoretical temperature value corresponding to each sampling moment is obtained. All theoretical temperature values ​​are arranged in chronological order to form the theoretical temperature sequence.

[0021] In one specific embodiment, a multi-order thermal network model of the circuit under test is first established. The multi-stage thermal network model is composed of multiple thermal resistance elements and multiple thermal capacity elements connected in series or parallel according to the actual heat transfer path of the circuit under test. Taking a typical overhead conductor as an example, its heat transfer path from the inside to the outside is as follows: conductor core (heat source) → conductor-insulation interface → insulation layer → insulation layer-sheath interface → sheath layer → sheath layer-surface interface → surface layer → surface-environment interface → environment. Along this heat transfer path, the conductor core, insulation layer, sheath layer, and surface layer are each taken as a thermal capacity node, forming a total of 4 thermal capacity elements. The heat transfer resistance between adjacent nodes is taken as a thermal resistance element, forming a total of 3 thermal resistance elements. Each element is connected in series according to the heat transfer path. In this thermal network model, the heat source is located at the innermost conductor thermal capacity node, and its heat generation power is determined by the circuit operating current and the circuit equivalent resistance. A convection thermal resistance element is also set between the outermost surface layer thermal capacity node and the environment to simulate the process of surface heat loss to the environment. The initial temperature of each thermal capacity element is set to the ambient temperature.

[0022] Secondly, the historical current data sequence and corresponding historical temperature data sequence of the line under test under normal operating conditions are obtained as calibration samples. The historical normal operating conditions refer to the operating conditions of the line under test when there are no known overheating faults or insulation abnormalities. At least one complete load cycle (e.g., 24 hours) of data is selected from the historical operation and maintenance records. During this cycle, the line has experienced a complete load change process from light load to heavy load and back to light load, covering the thermal response behavior under different current levels, ensuring that the current data and temperature data are synchronized in time and the sampling interval is consistent.

[0023] Then, using the historical current data sequence as input and the historical temperature data sequence as output, a parameter estimation method is used to determine the thermal resistance value of each thermal resistance element and the heat capacity value of each heat capacity element in the multi-order thermal network model. Specifically, the current values ​​at each moment in the historical current data sequence are input into the multi-order thermal network model in chronological order, and the temperature of each heat capacity element is calculated according to the following recursive method: For each heat capacity element, its temperature change rate is equal to the net heat power flowing into that element divided by its heat capacity. Taking a series-connected fourth-order thermal network model as an example, nodes 1 to 4 are numbered sequentially from the heat source (core) to the surface, with node temperatures ranging from T1 to T4. The thermal resistances between nodes are R1 (between node 1 and node 2), R2 (between node 2 and node 3), R3 (between node 3 and node 4), and R4 (between node 4 and the environment). The heat power P is composed of the current I and the equivalent resistance R. eq Decision: P = I²·R eq The temperature recurrence relation for each node is then: Node 1 (core): C1·dT1 / dt=P-(T1-T2) / R1; Node 2 (insulation layer): C2·dT2 / dt=(T1-T2) / R1-(T2-T3) / R2; Node 3 (Sheath layer): C3·dT3 / dt=(T2-T3) / R2-(T3-T4) / R3; Node 4 (surface layer): C4·dT4 / dt=(T3-T4) / R3-(T4-T env ) / R4; Where C1 to C4 are the heat capacities of each heat-capacity element, and T env The ambient temperature.

[0024] Discretizing the above continuous differential equation in the time domain, the temperature recursion formula for node i at the k-th sampling time is: T i (k+1)=T i (k)+(Δt / Ci)·[Inflow heat power - Outflow heat power] Among them, T i (k+1) represents the temperature value of the i-th heat capacity node at the (k+1)-th sampling time, T i (k) represents the temperature value of the i-th heat capacity node at the k-th sampling time, Ci represents the heat capacity value of the i-th heat capacity element, and Δt represents the sampling time interval. With this recursive formula, given the initial temperature, the temperature value of each node can be calculated time by time.

[0025] Using historical current data sequences as input and historical temperature data sequences (i.e., the temperature at node 4, i.e., the surface temperature) as output, a least-squares optimization method is employed to adjust the thermal resistance values ​​R1 to R4 and the heat capacity values ​​C1 to C4, minimizing the sum of squared deviations between the model's output surface temperature T4 and the measured temperature in the calibration samples. After optimization, a set of optimal thermal resistance and heat capacity values ​​is obtained, completing parameter identification. The model at this point is the electrothermal coupling model.

[0026] Next, the current values ​​at each moment in the line operating current data sequence are sequentially input into the electrothermal coupling model. The electrothermal coupling model then calculates the temperature of each thermal element hourly from the initial moment based on the identified thermal resistance and thermal capacity values. The recursive process is the same as that used during parameter identification, but the thermal resistance and thermal capacity values ​​are now fixed.

[0027] For each calculation moment, the temperature value of the heat capacity element (i.e., heat capacity element node 4 corresponding to the surface layer) located at the end of the heat transfer path in the multi-order thermal network model at the current moment is taken as the theoretical temperature value at that moment. This temperature value represents the expected surface temperature response generated by the current current excitation under the assumption that there are no overheating faults in the line.

[0028] By iterating through all sampling times, the theoretical temperature value corresponding to each sampling time is obtained. All theoretical temperature values ​​are arranged in chronological order to form the theoretical temperature sequence.

[0029] Finally, the theoretical temperature sequence and the line surface temperature data sequence are subtracted point by point at the same time, that is, the measured temperature is subtracted from the theoretical temperature to obtain the residual temperature sequence.

[0030] In summary, by establishing a multi-level thermal network model and combining it with historical normal operation data for parameter identification, the electrothermal coupling model can adaptively match the actual thermal dynamic characteristics of different lines without relying on idealized material parameter assumptions. This significantly improves the accuracy of theoretical temperature calculations. Compared with directly applying threshold judgments to measured temperatures, the theoretical temperature sequence calculated based on the electrothermal coupling model represents the temperature response that should be presented under ideal conditions assuming no thermal hazards in the line. Therefore, the residual temperature sequence can effectively isolate temperature changes caused by load fluctuations and ambient temperature variations under normal operating conditions, highlighting abnormal temperature rises caused by potential hazards such as poor contact, insulation aging, and overload. The residual temperature sequence obtained by comparing the theoretical temperature sequence with the line surface temperature data sequence point by point significantly reduces the interference of ambient temperature fluctuations and load variations on the detection results, and significantly improves the sensitivity and accuracy of subsequent abnormal fluctuation interval identification. Furthermore, the multi-level thermal network model can distinguish the differences in thermal inertia on different heat transfer paths, enabling the model to finely characterize the thermal response characteristics of different parts of the line, providing an accurate data foundation for subsequent thermal hazard monitoring based on residual sequences.

[0031] Step S103: Perform sliding window statistical analysis on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residual continues to accumulate positively.

[0032] In this step, a continuous decision window of a preset size slides on the residual temperature sequence with a preset sliding step size. Whenever the continuous decision window is completely located in the region where the residual value is greater than zero, the continuous decision window is marked as a positive residual window. Scan along the time direction to obtain all consecutive positive residual windows in the residual temperature sequence, merge adjacent positive residual windows into a continuous positive segment, the start time of the continuous positive segment is the start time of the first positive residual window in the continuous positive segment, and the end time of the continuous positive segment is the end time of the last positive residual window in the continuous positive segment. For each consecutive positive segment, the starting time of the consecutive positive segment is taken as the starting point of integration, and the ending time of the consecutive positive segment is taken as the ending point of integration. The residual values ​​at each time point within the consecutive positive segment are integrated to obtain the positive cumulative area of ​​each consecutive positive segment. Obtain the time span between the end time and the start time of each consecutive positive segment, and use it as the duration of each consecutive positive segment; When the cumulative positive area of ​​a certain continuous positive segment is greater than a preset area threshold and the duration of the certain continuous positive segment is greater than a preset time threshold, the certain continuous positive segment is determined as the certain abnormal fluctuation interval.

[0033] In one specific embodiment, firstly, a continuous decision window of a preset size slides along the residual temperature sequence using a preset sliding step size. The size of the continuous decision window is set according to the sampling frequency and the thermal inertia characteristics of the line. The length of the continuous decision window is greater than the duration of a single noise pulse and less than one-tenth of the line thermal response time constant, to ensure that the window can effectively filter out high-frequency noise without missing continuous abnormal temperature rises. For example, when the sampling frequency is one sampling point per minute, the length of the continuous decision window is set to 10 sampling points (i.e., 10 minutes), and the sliding step size is set to 1 sampling point. Whenever the continuous decision window slides to a new position, it is checked whether the residual values ​​of all sampling points in the window are greater than zero. If the residual values ​​of all sampling points in the window are greater than zero, the continuous decision window is marked as a positive residual window; if the residual value of any sampling point in the window is less than or equal to zero, the window is marked as a non-positive residual window and does not participate in subsequent processing.

[0034] Secondly, the residual temperature sequence is scanned along the time direction to obtain all continuous positive residual windows. Since the sliding step size is smaller than the size of the continuous decision window, there is an overlapping area between adjacent positive residual windows, which are continuous in time. All temporally adjacent and overlapping positive residual windows are merged into a continuous positive segment. The specific merging method is as follows: starting from the first positive residual window, it is checked whether the subsequent positive residual window is continuous or overlapping with the previous positive residual window in time. If so, it is included in the current continuous positive segment; otherwise, the current continuous positive segment ends and a new continuous positive segment begins. The start time of the continuous positive segment is the start time of the first positive residual window in the continuous positive segment, and the end time of the continuous positive segment is the end time of the last positive residual window in the continuous positive segment. Through this step, the discrete positive residual windows are merged into several continuous positive segments with a longer time span. Each continuous positive segment corresponds to a time interval in which the residual is continuously positive.

[0035] Next, for each consecutive positive segment, the residual values ​​at each time point within the consecutive positive segment are integrated, with the start time of the segment as the integration starting point and the end time as the integration ending point. This integration yields the positive cumulative area of ​​the consecutive positive segment. The integration employs a numerical integration method. In one specific embodiment, the trapezoidal integration method is used: the residual curve between two adjacent sampling points within the consecutive positive segment is approximated as a straight line. The area of ​​the trapezoid between this straight line and the time axis is calculated. The areas of all trapezoids are summed to obtain the positive cumulative area of ​​the consecutive positive segment. The positive cumulative area reflects the total amount of positive offset of the residual relative to the zero line within this time interval, and its physical meaning is "the degree of accumulation of abnormal temperature rise within this time period".

[0036] Meanwhile, the time span between the end time and the start time of each consecutive positive segment is obtained as the duration of each consecutive positive segment. This duration reflects the length of time the residual continues to shift positively, and its physical meaning is "the length of time the abnormal temperature rise continues".

[0037] Finally, when the cumulative positive area of ​​a continuous positive segment is greater than a preset area threshold and the duration of the continuous positive segment is greater than a preset time threshold, the continuous positive segment is determined as an abnormal fluctuation interval. The preset area threshold is determined based on the statistical distribution of the residual sequence under normal line operation. In one specific embodiment, the 95th percentile of the cumulative positive area of ​​all continuous positive segments in the historical normal period is taken as the preset area threshold. The preset time threshold is set based on the thermal response characteristics of the line. In one specific embodiment, it is set to twice the size of the continuous judgment window to ensure that the time span of the abnormal fluctuation interval covers at least two windows and avoids misjudgment due to the accidental fluctuation of a single window. The abnormal fluctuation interval can only be determined when both the cumulative positive area and the duration conditions are met simultaneously. Meeting only one condition does not constitute a judgment.

[0038] In summary, by combining sliding window detection with continuous positive segment merging, discrete windows that are continuously positive in the residual sequence are grouped into continuous time intervals. This overcomes the defect of point-by-point detection being easily interfered with by single noise points, significantly improving the noise resistance of abnormal interval identification. On this basis, continuous positive segments are screened using a dual criterion of positive cumulative area and duration. The positive cumulative area reflects the cumulative energy of the abnormal temperature rise, while the duration reflects the persistence of the abnormal temperature rise. The combination of the two can effectively distinguish between short-term random fluctuations and true persistent overheating hazards, avoiding false alarms and false negatives caused by a single threshold judgment. Specifically, even if the residual is briefly positive, the positive cumulative area of ​​short-term random fluctuations is limited and the duration is short, so it cannot simultaneously satisfy the two threshold conditions; while the positive offset of the residual caused by persistent overheating hazards has a long duration and a large cumulative area, which can be reliably identified. This method does not rely on a fixed temperature threshold and can adaptively adapt to the residual change characteristics under different line and environmental conditions, improving the accuracy and robustness of abnormal fluctuation interval identification.

[0039] Step S104: Extract the current subsequence corresponding to a certain abnormal fluctuation range from the line operating current data sequence, and extract the temperature subsequence corresponding to the certain abnormal fluctuation range from the line surface temperature data sequence, to form a certain electrothermal response data pair for the certain abnormal fluctuation range.

[0040] In this step, the start and end time points of a certain abnormal fluctuation range are determined, and a certain time length of the certain abnormal fluctuation range is calculated. Based on a certain time length, a certain forward expansion length and a certain backward expansion length are determined using a preset interval expansion coefficient, wherein the certain forward expansion length and the certain backward expansion length are positively correlated with the certain time length; A certain interception starting point is obtained by extending the forward extension length forward from the starting time node, and a certain interception ending point is obtained by extending the backward extension length backward from the ending time node; All current data points between a certain starting point and a certain ending point are extracted from the line operating current data sequence to form a current subsequence; From the line surface temperature data sequence, all temperature data points between a certain starting point and a certain ending point are extracted to form a certain temperature subsequence; The phase calibration of a certain current subsequence and a certain temperature subsequence is performed by cross-correlation analysis, so that the calibrated current subsequence is aligned with the certain temperature subsequence in time; By combining a certain current subsequence and a certain temperature subsequence after alignment, a certain electrothermal response data pair is obtained for a certain abnormal fluctuation range.

[0041] In one specific embodiment, taking a certain abnormal fluctuation range identified in step S103 as the processing object, the subsequences corresponding to the abnormal fluctuation range are extracted from the line operating current data sequence and the line surface temperature data sequence, and the two are combined into an electrothermal response data pair.

[0042] First, determine the start and end time nodes of a certain abnormal fluctuation interval, and calculate the time length of the certain abnormal fluctuation interval. The time length is equal to the end time node minus the start time node, and the unit is the number of sampling intervals or actual time units (such as minutes or seconds). If step S103 identifies multiple abnormal fluctuation intervals, this step is performed independently for each abnormal fluctuation interval, and each abnormal fluctuation interval generates its own electrothermal response data pair.

[0043] Secondly, based on a certain time length, the forward expansion length and backward expansion length are determined using a preset interval expansion coefficient. The interval expansion coefficient is a constant greater than 0 and less than 1. In a specific embodiment, it is set to 0.2. The forward expansion length is equal to the time length multiplied by the interval expansion coefficient, that is, forward expansion length = time length × 0.2; the backward expansion length is equal to the time length multiplied by the interval expansion coefficient, that is, backward expansion length = time length × 0.2. The expansion length is positively correlated with the time length. Its physical meaning is that: the abnormal fluctuation interval usually corresponds to the period when the residual continues to accumulate positively. However, there is a thermal inertia lag between the current change and the temperature response. The current has already changed before the abnormality occurs, and the temperature is still responding to the current change after the abnormality ends. Therefore, it is necessary to retain an additional data segment on both sides of the abnormal interval so that the electrothermal response data captured later can completely cover the transition process before and after the abnormality occurs, and avoid information loss due to the capture boundary being close to the abnormal interval.

[0044] The starting point of the interception is obtained by extending the forward extension length forward from the starting time node, i.e., the starting point of the interception = the starting time node - the forward extension length; the ending point of the interception is obtained by extending the backward extension length backward from the ending time node, i.e., the ending point of the interception = the ending time node + the backward extension length. If the starting point of the interception is earlier than the monitoring start time or the ending point of the interception is later than the monitoring end time, then the starting point of the interception is corrected to the monitoring start time and the ending point of the interception is corrected to the monitoring end time.

[0045] In the line operating current data sequence, all current data points from the starting point to the ending point are extracted to form a current subsequence; in the line surface temperature data sequence, all temperature data points from the starting point to the ending point are extracted to form a temperature subsequence. The current subsequence and the temperature subsequence have the same time axis and the same length.

[0046] Then, cross-correlation analysis is used to perform phase calibration on the current and temperature subsequences. Due to thermal inertia, the change in line temperature lags behind the change in current; that is, the temperature response does not change synchronously with the current but begins to change after a certain period of time. Therefore, phase calibration is needed to align them in time. Specifically, the current subsequence is shifted backward by several different time offsets, ranging from 0 to a preset maximum delay. For each time offset, the correlation between the overlapping data of the shifted current and temperature subsequences is calculated to obtain the cross-correlation coefficient corresponding to that offset. The time offset corresponding to the maximum cross-correlation coefficient is selected as the optimal offset. The current subsequence is then shifted backward by the optimal offset, so that the correspondence between the shifted current and temperature subsequences on the time axis reflects the true causal relationship of "current change first, temperature response later," resulting in the phase-calibrated current subsequence. At this point, the temperature subsequence remains unchanged, the current subsequence is delayed, and the two are aligned in time.

[0047] Finally, the aligned current subsequence and temperature subsequence are combined and stored as a binary sequence in chronological order to obtain the electrothermal response data pair for the abnormal fluctuation range. Each time point in the electrothermal response data pair contains both current and temperature values, and their temporal relationship correctly reflects the causal order of electrothermal coupling.

[0048] In summary, by extending the interval and retaining additional data on both sides of the abnormal fluctuation interval, the extracted electrothermal response data fully covers the current change process before the anomaly and the temperature transition process after the anomaly. This avoids information loss caused by the extraction boundary being too close to the abnormal interval, providing a complete data foundation for subsequent electrothermal coupling feature extraction. The completeness of the extracted data directly determines the accuracy of subsequent empirical mode decomposition and local variation coefficient extraction. If normal state data before the anomaly is missing, a baseline reference for the electrothermal response cannot be established; if transition process data after the anomaly is missing, the decay characteristics of thermal inertia cannot be captured. Based on this, phase calibration of the current subsequence and temperature subsequence is performed using cross-correlation analysis. Utilizing the physical characteristic that the temperature response lags behind the current change, the optimal offset that maximizes the correlation between the two is found, eliminating the temporal misalignment caused by thermal inertia. The corrected current and temperature accurately correspond on the time axis, ensuring that the current and temperature data at the same moment in the subsequent empirical mode decomposition reflect the state of the same heat transfer process, avoiding mode pairing errors caused by temporal misalignment. Cross-correlation analysis quantitatively calculates the correlation coefficient between current and temperature under different offsets, upgrading the thermal response delay time from an empirical value to a data-driven optimal value. Compared with empirical correction methods with fixed delay times, it has higher accuracy and robustness, and can adapt to the differentiated delay characteristics under different load levels and thermal inertia conditions. After phase calibration, the current subsequence and temperature subsequence are aligned in time, and the combined electrothermal response data pair can truly reflect the dynamic behavior of electrothermal coupling of the line, providing reliable input data for the subsequent extraction of multi-scale local variation coefficients.

[0049] Step S105: Determine the local coefficient of variation of a certain electrothermal response data pair at different time scales to obtain a certain coefficient of variation sequence of the certain electrothermal response data pair.

[0050] In this step, empirical mode decomposition is performed on the temperature subsequence in a certain electrothermal response data pair to obtain several temperature eigenmode functions. Each temperature eigenmode function represents the oscillation component of the temperature signal at different time scales. Empirical mode decomposition is performed on the current subsequence in a certain electrothermal response data pair to obtain several current eigenmode functions. Each current eigenmode function represents the oscillation component of the current signal at different time scales. According to the order of frequency from high to low, the temperature eigenmode functions and current eigenmode functions of the same order are paired to obtain multiple temperature-current mode pairs; The instantaneous amplitude of a certain temperature-current mode pair is calculated using Hilbert transform, resulting in a sequence of instantaneous amplitudes for a certain temperature and a sequence of instantaneous amplitudes for a certain current. The ratio of the instantaneous amplitude sequence of a certain temperature to the instantaneous amplitude sequence of a certain current is calculated point by point to obtain the instantaneous amplitude ratio sequence of the certain temperature-current mode pair; For a given instantaneous amplitude ratio sequence, a sliding window is used to calculate the ratio of the standard deviation to the mean of the amplitude ratio data within the window. This ratio is used as the local coefficient of variation at the center of the sliding window. The local coefficient of variation is CV(t) = σ(t) / μ(t), where σ(t) is the standard deviation of the data within the window and μ(t) is the mean of the data within the window. The local coefficients of variation of all temperature-current mode pairs at various times are aligned according to time to construct a multimodal coefficient of variation matrix. The rows of the multimodal coefficient of variation matrix correspond to different modal orders, and the columns correspond to different time nodes. The multimodal coefficient of variation matrix is ​​concatenated into a one-dimensional sequence in row order to obtain a coefficient of variation sequence for a certain electrothermal response data pair.

[0051] Step S106: Calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals.

[0052] In this step, the first coefficient of variation sequence is divided into multiple non-overlapping first windows of equal length, the first arithmetic mean of all data points in each first window is calculated, and the first arithmetic means are sorted to obtain the first aggregate sequence. The second coefficient of variation sequence is divided into multiple non-overlapping equal-length second windows. The second arithmetic mean of all data points in each second window is calculated, and the second arithmetic means are sorted to obtain the second aggregate sequence. Construct a distance matrix and use a dynamic programming algorithm to search for the optimal regular path in the distance matrix, wherein the element in the i-th row and j-th column of the distance matrix is ​​the Euclidean distance between the i-th mean of the first aggregation sequence and the j-th mean of the second aggregation sequence; In the recursive process of dynamic programming, a cumulative cost matrix D is defined, where the element D(i,j) in the cumulative cost matrix D represents the minimum cumulative cost from the top left corner (1,1) of the matrix to the point (i,j). After obtaining the cumulative cost matrix D through recursive calculation, backtracking is started from the minimum cumulative cost D(m,n) from the upper left corner (1,1) to the lower right corner (m,n) of the matrix to obtain all points on the optimal regularization path, and the weighted cumulative sum of all points on the optimal regularization path is calculated to obtain the dynamic time regularization distance.

[0053] In a specific embodiment, the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair obtained in step S105 are used as processing objects, and the dynamic time warping distance between them is calculated. The first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two time-adjacent abnormal fluctuation intervals.

[0054] First, the first coefficient of variation sequence is divided into multiple non-overlapping first windows of equal length. For example, the length of the first window is determined based on the total length of the first coefficient of variation sequence. The larger value between one percent of the total length of the sequence and the preset minimum window length is taken to ensure that the window length is between 5 and 20 sampling points. This preserves the local morphological features of the sequence while effectively suppressing high-frequency noise. The arithmetic mean of all data points in each first window is calculated to obtain the first arithmetic mean of that window. The first arithmetic means of all the first windows are arranged in chronological order on the time axis to obtain the first aggregate sequence. The length of the first aggregate sequence is equal to the length of the first coefficient of variation sequence divided by the length of the first window and rounded up.

[0055] The second coefficient of variation sequence is divided into multiple non-overlapping, equal-length second windows, with the length of the second window being the same as that of the first window. The arithmetic mean of all data points within each second window is calculated to obtain the second arithmetic mean of that window. The second arithmetic means of all the second windows are arranged in chronological order on the time axis to obtain the second aggregated sequence. The length of the second aggregated sequence is equal to the length of the second coefficient of variation sequence divided by the length of the second window and rounded up. Through segmented aggregation approximation, the number of sampling points of the two sequences is reduced to the same order of magnitude, while preserving the overall trend and main fluctuation pattern of the sequences.

[0056] Then, a distance matrix is ​​constructed, with the number of rows equal to the length m of the first aggregation sequence and the number of columns equal to the length n of the second aggregation sequence. The element in the i-th row and j-th column of the distance matrix is ​​the Euclidean distance between the i-th mean of the first aggregation sequence and the j-th mean of the second aggregation sequence, i.e., d(i,j)=|ai-bj|, where ai is the i-th value of the first aggregation sequence and bj is the j-th value of the second aggregation sequence. Each element of this distance matrix reflects the local numerical difference between the two aggregation sequences at corresponding positions.

[0057] A dynamic programming algorithm is used to search for the optimal regularization path in the distance matrix. The optimal regularization path starts from the top left corner (1,1) and ends at the bottom right corner (m,n) of the distance matrix. It is used to establish a non-linear correspondence between the first and second aggregation sequences, aligning points with similar shapes in the two sequences. Each step of the optimal regularization path can only move to the right, down, or to the lower right, and both the row and column indices are monotonically increasing during the movement to ensure that the correspondence does not violate the time order.

[0058] In the recursive process of dynamic programming, we define a cumulative cost matrix D, where D(i,j) represents the minimum cumulative cost from the top-left corner (1,1) to point (i,j). We initialize D(1,1) = d(1,1). For the first row, D(1,j) = d(1,j) + D(1,j-1) + p right , where p right The penalty coefficient for shifting to the right is D(i,1) = d(i,1) + D(i-1,1) + p for the first column. down , where p down This represents the penalty coefficient for moving downwards. For other positions (i from 2 to m, j from 2 to n), the recursive formula is: D(i,j)=d(i,j)+min{D(i-1,j)+p down ,D(i,j-1)+p right ,D(i-1,j-1)+p diag} Where p diag The penalty coefficient for moving to the lower right, for example, p down =1.2, p right =1.2, p diag =1.0, the penalty coefficient for moving to the lower right is less than the penalty coefficients for moving to the right and moving down, so that the optimal regular path is extended along the diagonal direction first, avoiding excessive stretching or compression of the time axis due to local noise.

[0059] After obtaining the cumulative cost matrix D through recursive calculation, backtracking starts from D(m,n) to obtain all points on the optimal regularized path. The backtracking rule is: starting from the current point (i,j), compare D(i-1,j)+p down D(i,j-1)+p right and D(i-1,j-1)+p diag Given three candidate values, select the position that minimizes the cumulative cost as the previous point on the path, move to that point, and repeat this process until backtracking to (1,1). Record the coordinates of all points on the path.

[0060] The weighted cumulative sum of all points on the optimal regularization path is calculated to obtain the dynamic time regularization distance. The smaller the distance value, the more similar the fluctuation patterns of the first and second coefficient of variation sequences are; the larger the distance value, the more significant the difference in fluctuation patterns between the two sequences.

[0061] In summary, by quantifying the similarity of electrothermal response fluctuation patterns between adjacent abnormal fluctuation intervals using dynamic time warping distance, a quantitative criterion is provided for determining whether two adjacent abnormal intervals are caused by the same fault source. First, a piecewise aggregation approximation is used to reduce the dimensionality and smooth the coefficient of variation sequence, which not only preserves the overall trend and main fluctuation patterns of the sequence, but also significantly reduces the number of data points, thereby reducing the computational complexity of dynamic programming. At the same time, it plays a role in smoothing and denoising, improving the robustness of subsequent matching. The Dynamic Time Warping (VTW) algorithm, through nonlinear time axis alignment, allows two sequences to be stretched and compressed on the time axis for matching. This enables the finding of the optimal correspondence between two anomalous intervals even if their durations differ or their fluctuation patterns have local offsets on the time axis. This overcomes the limitation of Euclidean distance, which requires two sequences to be of equal length and strictly aligned on the time axis, making it more suitable for comparing anomalous intervals with unequal durations in actual monitoring. The algorithm introduces a directional penalty coefficient in the dynamic programming recursion, prioritizing the path extension along the diagonal direction. This avoids excessive distortion caused by local noise or short-term fluctuations, making the matching results more consistent with physical reality—that is, the correspondence between the electrothermal response waveforms of two adjacent anomalous intervals on the time axis should not be drastically misaligned. The VTW distance comprehensively reflects the overall differences between two sequences in three dimensions: fluctuation amplitude, fluctuation pattern, and time axis correspondence. A smaller distance indicates more similar electrothermal response characteristics between the two intervals, possibly caused by the same fault source. A larger distance indicates significant differences in characteristics between the two intervals, suggesting a possible change in fault type or the addition of a new fault. This provides a quantitative basis for subsequent fault type identification.

[0062] Step S107: Determine whether the dynamic time warping distance is greater than a preset distance threshold.

[0063] In this step, when the distance is not greater than a preset distance threshold, it indicates that the fluctuation patterns of the first and second coefficient of variation sequences are similar, the electrothermal response characteristics of the two adjacent abnormal fluctuation intervals are consistent, they may be caused by the same fault source and the nature of the fault has not changed significantly. At this time, only the first electrothermal response data pair is subjected to overheating fault type identification according to the first analysis strategy in the subsequent step S108, and the overheating fault type identification result corresponding to the first electrothermal response data pair is obtained. The identification result is directly used as the overheating fault type identification result corresponding to the second electrothermal response data pair. The reason for adopting this processing method is that since the electrothermal response fluctuation patterns of the two adjacent abnormal fluctuation intervals are similar, the underlying physical fault mechanisms are the same. Identifying them separately will produce the same identification results. Therefore, it is not necessary to repeatedly perform the computationally intensive feature extraction and classification identification steps for the second electrothermal response data pair. Under the premise of ensuring identification accuracy, the computational complexity is significantly reduced and the overall analysis efficiency is improved.

[0064] Step S108: If the distance is greater than a preset distance threshold, the first electrothermal response data pair and the second electrothermal response data pair are defined as the first suspicious data pair and the second suspicious data pair, respectively. The first suspicious data pair and the second suspicious data pair are then identified for overheating fault type according to a preset first analysis strategy to obtain the target overheating fault type identification result.

[0065] In this step, a sliding window cross-correlation analysis is performed on the temperature subsequence and current subsequence after phase calibration in the first questionable data pair. Multiple sliding windows of different lengths are set, and the maximum cross-correlation coefficient and the corresponding time delay are calculated in each window to obtain the maximum cross-correlation coefficient sequence and the time delay sequence. The mean of the maximum cross-correlation coefficient sequence is calculated as a first correlation strength index, and the standard deviation of the maximum cross-correlation coefficient sequence is calculated as a first correlation stability index. The mean of the time delay sequence is calculated as a first response lag index, and the standard deviation of the time delay sequence is calculated as a first lag volatility index. The first correlation strength index, the first correlation stability index, the first response lag index, and the first lag fluctuation index are combined to form the first feature vector of the first questionable data pair. A multi-class support vector machine model is pre-constructed. The support vector machine model adopts a radial basis kernel function and is trained through historical fault samples. The historical fault samples include feature vectors corresponding to known overheating fault types, including poor contact type, insulation aging type, and overload type. The first feature vector is input into the multi-class support vector machine model, and the multi-class support vector machine model outputs the overheating fault type corresponding to the first suspicious data pair; The second feature vector corresponding to the second suspicious data pair is input into the multi-class support vector machine model, and the multi-class support vector machine model outputs the overheating fault type corresponding to the second suspicious data pair.

[0066] In one specific embodiment, after obtaining the target overheating fault type identification result, based on the target overheating fault type identification result, a preset second analysis strategy is used to analyze the overheating fault impact range of the third electrothermal response data pair corresponding to the previous non-abnormal fluctuation interval adjacent to the first suspicious data pair and the fourth electrothermal response data pair corresponding to the subsequent non-abnormal fluctuation interval adjacent to the second suspicious data pair. The second analysis strategy specifically includes: When the overheating fault type of the first suspected data pair is identified as poor contact, a causal relationship test is performed on the third electrothermal response data pair, specifically including: subtracting the mean of the current subsequence and temperature subsequence in the third electrothermal response data pair from their respective mean values ​​to obtain a zero-mean current sequence and a zero-mean temperature sequence. Pair the zero-mean current sequence with the zero-mean temperature sequence at the same time to obtain at least one first pairing, and calculate the Pearson correlation coefficient between the two values ​​in the at least one first pairing to obtain at least one simultaneous correlation coefficient. The value at a certain moment in the zero-mean current sequence is paired with the values ​​at several consecutive moments after a certain moment in the zero-mean temperature sequence to obtain at least one second pairing. The Pearson correlation coefficient between the two values ​​in the at least one second pairing is calculated to obtain at least one delayed correlation coefficient. Determine whether a certain delayed correlation coefficient is greater than any simultaneous correlation coefficient, and whether a certain delayed correlation coefficient is greater than a preset correlation coefficient threshold; If a certain delay correlation coefficient is greater than any simultaneous correlation coefficient, and a certain delay correlation coefficient is greater than a preset correlation coefficient threshold, then the continuous time period that is closest in time to the first questionable data pair in the third electrothermal response data pair is marked as the thermal hazard conduction time period. When the overheating fault type of the first suspicious data pair is identified as insulation aging type, a monotonicity test is performed on the temperature subsequence of the third electrothermal response data pair. The monotonicity test specifically includes: The temperature subsequence in the third electrothermal response data pair is divided into a first temperature sub-segment and a second temperature sub-segment in chronological order. The first mean of the first temperature sub-segment and the second mean of the second temperature sub-segment are calculated respectively. The difference between the first mean and the second mean is calculated to obtain the mean difference. The temperature subsequence in the third electrothermal response data pair is linearly fitted using the least squares method to obtain the linear fitting slope, where the linear fitting slope is the slope of the fitted straight line of temperature changing with time. Determine whether the mean difference is greater than the mean difference threshold and whether the linear fitting slope is greater than the fitting slope threshold; If the mean difference is greater than the mean difference threshold and the linear fitting slope is greater than the fitting slope threshold, then the entire time interval of the third electrothermal response data pair is marked as a period of early signs of thermal hazards. When the overheating fault type of the second suspicious data pair is identified as overload type, thermal inertia analysis is performed on the fourth electrothermal response data pair. The thermal inertia analysis specifically includes: The current subsequence in the fourth electrothermal response data pair is squared to obtain the current squared sequence. The current subsequence in the fourth electrothermal response data pair and the current square sequence are normalized to obtain the normalized current subsequence and the normalized current square sequence. The time delay at which the cross-correlation function between the normalized current subsequence and the normalized current square sequence reaches its maximum value is determined and used as the thermal response delay time. The thermal response delay time is compared with the baseline thermal response delay time of the equipment under normal operating conditions. If the current thermal response delay time exceeds a preset multiple of the baseline thermal response delay time, the entire time interval of the fourth electrothermal response data pair is marked as the period of continuous impact of thermal hazards.

[0067] Electrothermal coupling characteristics of questionable data pairs were extracted from multiple time scales using sliding window cross-correlation analysis. Correlation strength and stability indices characterized the overall electrothermal coupling from the perspectives of amplitude and fluctuation, respectively. Response hysteresis and hysteresis fluctuation indices characterized thermal inertia from the perspective of time delay. These four complementary dimensions comprehensively reflect the differentiated performance of different fault types in electrothermal coupling relationships. The radial basis function kernel support vector machine model projects feature vectors into a high-dimensional space through nonlinear mapping, effectively handling nonlinear relationships between features. Compared to linear classifiers, it has stronger classification capabilities, and the distinction boundaries between the three fault types are clear. The electrothermal coupling relationship of poor contact faults exhibits nonlinear distortion, the correlation strength of insulation aging faults shows a slow decreasing trend, and the response hysteresis index of overload faults increases significantly. The four-dimensional feature vectors effectively capture these differences, enabling the support vector machine model to achieve accurate classification.

[0068] In terms of the second analysis strategy, a causal relationship test is used for poor contact type, which uses the comparison of delayed correlation coefficient and instantaneous correlation coefficient to identify the causal effect of current on temperature, accurately locating the range of heat conduction influence and avoiding misjudging normal fluctuations as potential hazards. For insulation aging type, a monotonic trend test is used, which uses the dual criteria of segmented mean difference and linear fitting slope to focus on both the overall rise before and after the segment and the continuous upward trend, enabling early detection of the precursor characteristics of slow temperature rise and providing a time window for preventive maintenance. For overload type, thermal inertia analysis is used, which quantifies the decline in heat dissipation performance into a measurable indicator by comparing the thermal response delay time with the baseline value. The three types of influence range analysis output "heat hazard conduction time period", "heat hazard precursor time period" and "heat hazard continuous influence time period" respectively, with clear labeling methods, which facilitates maintenance personnel to quickly understand the scope of hazard influence and provides a direct basis for subsequent maintenance decisions, forming a complete analysis closed loop from anomaly identification → type identification → impact assessment.

[0069] In summary, the method of this application acquires the line surface temperature data sequence and operating current data sequence collected by the temperature measurement and protection device; determines the theoretical temperature sequence based on the electrothermal coupling model and compares it with the measured values ​​to obtain the residual temperature sequence; performs sliding window statistical analysis on the residual temperature sequence to identify abnormal fluctuation intervals where the residuals continuously accumulate positively; extracts the current and temperature subsequences corresponding to the abnormal intervals to form electrothermal response data pairs, extracts the local coefficients of variation at different time scales to obtain the coefficient of variation sequence; calculates the dynamic time warping distance of the coefficient of variation sequences of adjacent abnormal intervals, and if it is greater than a threshold, identifies the overheating fault type of the suspicious data pairs; and uses the corresponding second analysis strategy to analyze the impact range according to different fault types; effectively improving the identification accuracy and diagnostic efficiency of line overheating faults.

[0070] Please see Figure 2 The diagram shows a structural block diagram of a line overheating fault analysis system based on a temperature measurement and protection device according to this application.

[0071] like Figure 2 As shown, the line overheating fault analysis system 200 includes an acquisition module 210, a comparison module 220, an analysis module 230, an interception module 240, a determination module 250, a calculation module 260, a judgment module 270, and an identification module 280.

[0072] The acquisition module 210 is configured to acquire the line surface temperature data sequence and the line operating current data sequence collected by the temperature protection device during a preset monitoring period. The comparison module 220 is configured to determine the theoretical temperature sequence of the line under test based on the line operating current data sequence using a preset electrothermal coupling model, and compare the theoretical temperature sequence with the line surface temperature data sequence point by point to obtain the residual temperature sequence. The analysis module 230 is configured to perform sliding window statistical analysis on the residual temperature sequence to identify at least one abnormal fluctuation interval where the residual continuously accumulates positively. The extraction module 240 is configured to extract the current subsequence corresponding to a certain abnormal fluctuation interval from the line operating current data sequence and the temperature subsequence corresponding to the certain abnormal fluctuation interval from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation interval. The determination module 250 is configured to determine the certain... The local coefficient of variation of the electrothermal response data pairs at different time scales is used to obtain a coefficient of variation sequence for a certain electrothermal response data pair; the calculation module 260 is configured to calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; the judgment module 270 is configured to determine whether the dynamic time warping distance is greater than a preset distance threshold; the identification module 280 is configured to, if the dynamic time warping distance is greater than the preset distance threshold, define the first electrothermal response data pair and the second electrothermal response data pair as the first suspicious data pair and the second suspicious data pair, respectively, and perform overheating fault type identification on the first suspicious data pair and the second suspicious data pair according to a preset first analysis strategy to obtain the target overheating fault type identification result.

[0073] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0074] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the circuit overheating fault analysis method based on a temperature measurement and protection device in any of the above method embodiments. In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows: Acquire the line surface temperature data sequence and line operating current data sequence collected by the temperature protection device during the preset monitoring period; Based on the line operating current data sequence, a preset electrothermal coupling model is used to determine the theoretical temperature sequence of the line under test, and the theoretical temperature sequence is compared point by point with the line surface temperature data sequence to obtain the residual temperature sequence. A sliding window statistical analysis was performed on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residuals continued to accumulate positively. A current subsequence corresponding to a certain abnormal fluctuation range is extracted from the line operating current data sequence, and a temperature subsequence corresponding to the certain abnormal fluctuation range is extracted from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation range. Determine the local coefficient of variation of a certain electrothermal response data pair at different time scales to obtain a coefficient of variation sequence of the certain electrothermal response data pair; Calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; Determine whether the dynamic time warp distance is greater than a preset distance threshold; If the distance exceeds a preset threshold, the first electrothermal response data pair and the second electrothermal response data pair are defined as the first suspicious data pair and the second suspicious data pair, respectively. The first suspicious data pair and the second suspicious data pair are then identified for overheating fault type according to a preset first analysis strategy to obtain the target overheating fault type identification result.

[0075] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the line overheating fault analysis system based on the temperature protection device. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely disposed relative to a processor, which can be connected to the line overheating fault analysis system based on the temperature protection device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0076] Figure 3This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the circuit overheating fault analysis method based on the temperature protection device described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the circuit overheating fault analysis system based on the temperature protection device. The output device 340 may include a display screen or other display device.

[0077] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0078] In one implementation, the above-described electronic device is applied to a line overheating fault analysis system based on a temperature measurement and protection device, for a client, and includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Acquire the line surface temperature data sequence and line operating current data sequence collected by the temperature protection device during the preset monitoring period; Based on the line operating current data sequence, a preset electrothermal coupling model is used to determine the theoretical temperature sequence of the line under test, and the theoretical temperature sequence is compared point by point with the line surface temperature data sequence to obtain the residual temperature sequence. A sliding window statistical analysis was performed on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residuals continued to accumulate positively. A current subsequence corresponding to a certain abnormal fluctuation range is extracted from the line operating current data sequence, and a temperature subsequence corresponding to the certain abnormal fluctuation range is extracted from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation range. Determine the local coefficient of variation of a certain electrothermal response data pair at different time scales to obtain a coefficient of variation sequence of the certain electrothermal response data pair; Calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; Determine whether the dynamic time warp distance is greater than a preset distance threshold; If the distance exceeds a preset threshold, the first electrothermal response data pair and the second electrothermal response data pair are defined as the first suspicious data pair and the second suspicious data pair, respectively. The first suspicious data pair and the second suspicious data pair are then identified for overheating fault type according to a preset first analysis strategy to obtain the target overheating fault type identification result.

[0079] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing circuit overheating faults based on temperature measurement and protection devices, characterized in that, include: Acquire the line surface temperature data sequence and line operating current data sequence collected by the temperature protection device during the preset monitoring period; Based on the line operating current data sequence, a preset electrothermal coupling model is used to determine the theoretical temperature sequence of the line under test, and the theoretical temperature sequence is compared point by point with the line surface temperature data sequence to obtain the residual temperature sequence. A sliding window statistical analysis was performed on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residuals continued to accumulate positively. A current subsequence corresponding to a certain abnormal fluctuation range is extracted from the line operating current data sequence, and a temperature subsequence corresponding to the certain abnormal fluctuation range is extracted from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation range. Determine the local coefficient of variation of a certain electrothermal response data pair at different time scales to obtain a coefficient of variation sequence of the certain electrothermal response data pair; Calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; Determine whether the dynamic time warp distance is greater than a preset distance threshold; If the distance exceeds a preset threshold, the first electrothermal response data pair and the second electrothermal response data pair are defined as the first suspicious data pair and the second suspicious data pair, respectively. The first suspicious data pair and the second suspicious data pair are then identified for overheating fault type according to a preset first analysis strategy to obtain the target overheating fault type identification result.

2. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, The step of determining the theoretical temperature sequence of the line under test using a preset electrothermal coupling model based on the line operating current data sequence includes: A multi-order thermal network model of the circuit under test is established. The multi-order thermal network model is composed of multiple thermal resistance elements and multiple thermal capacity elements connected in series or in parallel according to the actual heat transfer path of the circuit under test. The historical current data sequence and the corresponding historical temperature data sequence of the circuit under test under normal operating conditions are obtained as calibration samples. Using the historical current data sequence as input and the historical temperature data sequence as output, a parameter estimation method is used to determine the thermal resistance value of each thermal resistance element and the thermal capacity value of each thermal capacity element in the multi-order thermal network model, so that the deviation between the temperature data sequence output by the multi-order thermal network model under the excitation of the current data sequence and the historical temperature data sequence in the calibration sample is minimized, thereby completing the parameter identification of the multi-order thermal network model and obtaining the electrothermal coupling model. The current values ​​at each moment in the line operating current data sequence are sequentially input into the electrothermal coupling model according to the time sequence. The electrothermal coupling model simulates the process of heat being transferred from the heat source to the line surface step by step according to the thermal resistance and heat capacity values. The temperature of each heat capacity element is calculated recursively from the initial moment. The temperature value of the heat capacity element located at the end of the heat transfer path of the multi-order thermal network model at the current moment is taken as the theoretical temperature value at the current moment. By traversing all sampling moments, the theoretical temperature value corresponding to each sampling moment is obtained. All theoretical temperature values ​​are arranged in chronological order to form the theoretical temperature sequence.

3. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, The step of performing sliding window statistical analysis on the residual temperature sequence to identify at least one abnormal fluctuation interval in which the residual continuously accumulates positively includes: According to the preset size of the continuous judgment window, the window slides on the residual temperature sequence with a preset sliding step size. Whenever the continuous judgment window is completely located in the region where the residual value is greater than zero, the continuous judgment window is marked as a positive residual window. Scan along the time direction to obtain all consecutive positive residual windows in the residual temperature sequence, merge adjacent positive residual windows into a continuous positive segment, the start time of the continuous positive segment is the start time of the first positive residual window in the continuous positive segment, and the end time of the continuous positive segment is the end time of the last positive residual window in the continuous positive segment. For each consecutive positive segment, the starting time of the consecutive positive segment is taken as the starting point of integration, and the ending time of the consecutive positive segment is taken as the ending point of integration. The residual values ​​at each time point within the consecutive positive segment are integrated to obtain the positive cumulative area of ​​each consecutive positive segment. Obtain the time span between the end time and the start time of each consecutive positive segment, and use it as the duration of each consecutive positive segment; When the cumulative positive area of ​​a certain continuous positive segment is greater than a preset area threshold and the duration of the certain continuous positive segment is greater than a preset time threshold, the certain continuous positive segment is determined as the certain abnormal fluctuation interval.

4. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, The step of extracting a current subsequence corresponding to a certain abnormal fluctuation range from the line operating current data sequence and extracting a temperature subsequence corresponding to the certain abnormal fluctuation range from the line surface temperature data sequence to form a certain electrothermal response data pair for the certain abnormal fluctuation range includes: Determine the start and end time points of a certain abnormal fluctuation range, and calculate a certain time length of the certain abnormal fluctuation range; Based on a certain time length, a certain forward expansion length and a certain backward expansion length are determined using a preset interval expansion coefficient, wherein the certain forward expansion length and the certain backward expansion length are positively correlated with the certain time length; A certain interception starting point is obtained by extending the forward extension length forward from the starting time node, and a certain interception ending point is obtained by extending the backward extension length backward from the ending time node; All current data points between a certain starting point and a certain ending point are extracted from the line operating current data sequence to form a current subsequence; From the line surface temperature data sequence, all temperature data points between a certain starting point and a certain ending point are extracted to form a certain temperature subsequence; The phase calibration of a certain current subsequence and a certain temperature subsequence is performed by cross-correlation analysis, so that the calibrated current subsequence is aligned with the certain temperature subsequence in time; By combining a certain current subsequence and a certain temperature subsequence after alignment, a certain electrothermal response data pair is obtained for a certain abnormal fluctuation range.

5. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, Determining the local coefficients of variation of a certain electrothermal response data pair at different time scales, and obtaining a sequence of coefficients of variation for a certain electrothermal response data pair, includes: Empirical mode decomposition is performed on the temperature subsequence in a certain electrothermal response data pair to obtain several temperature eigenmode functions. Each temperature eigenmode function represents the oscillation component of the temperature signal at different time scales. Empirical mode decomposition is performed on the current subsequence in a certain electrothermal response data pair to obtain several current eigenmode functions. Each current eigenmode function represents the oscillation component of the current signal at different time scales. According to the order of frequency from high to low, the temperature eigenmode functions and current eigenmode functions of the same order are paired to obtain multiple temperature-current mode pairs; The instantaneous amplitude of a certain temperature-current mode pair is calculated using Hilbert transform, resulting in a sequence of instantaneous amplitudes for a certain temperature and a sequence of instantaneous amplitudes for a certain current. The ratio of the instantaneous amplitude sequence of a certain temperature to the instantaneous amplitude sequence of a certain current is calculated point by point to obtain the instantaneous amplitude ratio sequence of the certain temperature-current mode pair; For a given instantaneous amplitude ratio sequence, a sliding window is used to calculate the ratio of the standard deviation to the mean of the amplitude ratio data within the window, which is then used as the local coefficient of variation at the center of the sliding window. The local coefficients of variation of all temperature-current mode pairs at various times are aligned according to time to construct a multimodal coefficient of variation matrix. The rows of the multimodal coefficient of variation matrix correspond to different modal orders, and the columns correspond to different time nodes. The multimodal coefficient of variation matrix is ​​concatenated into a one-dimensional sequence in row order to obtain a coefficient of variation sequence for a certain electrothermal response data pair.

6. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, The calculation of the dynamic time-normalized distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair includes: The first coefficient of variation sequence is divided into multiple non-overlapping equal-length first windows. The first arithmetic mean of all data points in each first window is calculated, and the first arithmetic means are sorted to obtain the first aggregate sequence. The second coefficient of variation sequence is divided into multiple non-overlapping equal-length second windows. The second arithmetic mean of all data points in each second window is calculated, and the second arithmetic means are sorted to obtain the second aggregate sequence. Construct a distance matrix and use a dynamic programming algorithm to search for the optimal regular path in the distance matrix, wherein the element in the i-th row and j-th column of the distance matrix is ​​the Euclidean distance between the i-th mean of the first aggregation sequence and the j-th mean of the second aggregation sequence; In the recursive process of dynamic programming, a cumulative cost matrix D is defined, where the element D(i,j) in the cumulative cost matrix D represents the minimum cumulative cost from the top left corner (1,1) of the matrix to the point (i,j). After obtaining the cumulative cost matrix D through recursive calculation, backtracking is started from the minimum cumulative cost D(m,n) from the upper left corner (1,1) to the lower right corner (m,n) of the matrix to obtain all points on the optimal regularization path, and the weighted cumulative sum of all points on the optimal regularization path is calculated to obtain the dynamic time regularization distance.

7. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, Based on a preset first analysis strategy, the first suspicious data pair and the second suspicious data pair are subjected to overheating fault type identification, and the target overheating fault type identification results are obtained, including: A sliding window cross-correlation analysis was performed on the temperature subsequence and current subsequence after phase calibration in the first questionable data pair. Multiple sliding windows of different lengths were set, and the maximum cross-correlation coefficient and the corresponding time delay were calculated in each window to obtain the maximum cross-correlation coefficient sequence and the time delay sequence. The mean of the maximum cross-correlation coefficient sequence is calculated as a first correlation strength index, and the standard deviation of the maximum cross-correlation coefficient sequence is calculated as a first correlation stability index. The mean of the time delay sequence is calculated as a first response lag index, and the standard deviation of the time delay sequence is calculated as a first lag volatility index. The first correlation strength index, the first correlation stability index, the first response lag index, and the first lag fluctuation index are combined to form the first feature vector of the first questionable data pair. A multi-class support vector machine model is pre-constructed. The support vector machine model adopts a radial basis kernel function and is trained through historical fault samples. The historical fault samples include feature vectors corresponding to known overheating fault types, including poor contact type, insulation aging type, and overload type. The first feature vector is input into the multi-class support vector machine model, and the multi-class support vector machine model outputs the overheating fault type corresponding to the first suspicious data pair; The second feature vector corresponding to the second suspicious data pair is input into the multi-class support vector machine model, and the multi-class support vector machine model outputs the overheating fault type corresponding to the second suspicious data pair.

8. The method for analyzing circuit overheating faults based on a temperature measurement and protection device according to claim 1, characterized in that, After obtaining the target overheating fault type identification result, the method further includes: Based on the target overheating fault type identification result, a preset second analysis strategy is used to analyze the overheating fault impact range of the third electrothermal response data pair corresponding to the previous non-abnormal fluctuation range adjacent to the first suspicious data pair and the fourth electrothermal response data pair corresponding to the subsequent non-abnormal fluctuation range adjacent to the second suspicious data pair. Specifically, the second analysis strategy includes: When the overheating fault type of the first suspicious data pair is identified as poor contact, a causal relationship test is performed on the third electrothermal response data pair. When the overheating fault type of the first suspicious data pair is identified as insulation aging type, a monotonic trend test is performed on the temperature subsequence of the third electrothermal response data pair. When the overheating fault type of the second suspicious data pair is identified as overload type, thermal inertia analysis is performed on the fourth electrothermal response data pair.

9. The method for analyzing line overheating faults based on a temperature measurement and protection device according to claim 8, characterized in that, The causal relationship test specifically includes: subtracting the mean values ​​of the current subsequence and temperature subsequence in the third electrothermal response data pair from their respective mean values ​​to obtain a zero-mean current sequence and a zero-mean temperature sequence; Pair the zero-mean current sequence with the zero-mean temperature sequence at the same time to obtain at least one first pairing, and calculate the Pearson correlation coefficient between the two values ​​in the at least one first pairing to obtain at least one simultaneous correlation coefficient. The value at a certain moment in the zero-mean current sequence is paired with the values ​​at several consecutive moments after a certain moment in the zero-mean temperature sequence to obtain at least one second pairing. The Pearson correlation coefficient between the two values ​​in the at least one second pairing is calculated to obtain at least one delayed correlation coefficient. Determine whether a certain delayed correlation coefficient is greater than any simultaneous correlation coefficient, and whether a certain delayed correlation coefficient is greater than a preset correlation coefficient threshold; If a certain delay correlation coefficient is greater than the correlation coefficient at any given moment, and a certain delay correlation coefficient is greater than a preset correlation coefficient threshold, then the continuous time period that is closest in time to the first questionable data pair in the third electrothermal response data pair is marked as the thermal hazard conduction time period. The monotonicity test specifically includes: The temperature subsequence in the third electrothermal response data pair is divided into a first temperature sub-segment and a second temperature sub-segment in chronological order. The first mean of the first temperature sub-segment and the second mean of the second temperature sub-segment are calculated respectively. The difference between the first mean and the second mean is calculated to obtain the mean difference. The temperature subsequence in the third electrothermal response data pair is linearly fitted using the least squares method to obtain the linear fitting slope, where the linear fitting slope is the slope of the fitted straight line of temperature changing with time. Determine whether the mean difference is greater than the mean difference threshold and whether the linear fitting slope is greater than the fitting slope threshold; If the mean difference is greater than the mean difference threshold and the linear fitting slope is greater than the fitting slope threshold, then the entire time interval of the third electrothermal response data pair is marked as a period of early signs of thermal hazards. The thermal inertia analysis specifically includes: The current subsequence in the fourth electrothermal response data pair is squared to obtain the current squared sequence. The current subsequence in the fourth electrothermal response data pair and the current square sequence are normalized to obtain the normalized current subsequence and the normalized current square sequence. The time delay at which the cross-correlation function between the normalized current subsequence and the normalized current square sequence reaches its maximum value is determined and used as the thermal response delay time. The thermal response delay time is compared with the baseline thermal response delay time of the equipment under normal operating conditions. If the current thermal response delay time exceeds a preset multiple of the baseline thermal response delay time, the entire time interval of the fourth electrothermal response data pair is marked as the period of continuous impact of thermal hazards.

10. A line overheating fault analysis system based on a temperature measurement and protection device, used to implement the line overheating fault analysis method according to any one of claims 1-9, characterized in that, The system includes: The acquisition module is configured to acquire the line surface temperature data sequence and the line operating current data sequence collected by the temperature protection device during a preset monitoring period. The comparison module is configured to determine the theoretical temperature sequence of the line under test based on the line operating current data sequence using a preset electrothermal coupling model, and to compare the theoretical temperature sequence with the line surface temperature data sequence point by point to obtain the residual temperature sequence. The analysis module is configured to perform sliding window statistical analysis on the residual temperature sequence to identify at least one abnormal fluctuation range in which the residual continues to accumulate positively. The interception module is configured to intercept a current subsequence corresponding to a certain abnormal fluctuation range from the line operating current data sequence, and to intercept a temperature subsequence corresponding to the certain abnormal fluctuation range from the line surface temperature data sequence, to form a certain electrothermal response data pair for the certain abnormal fluctuation range. The determination module is configured to determine the local coefficient of variation of a certain electrothermal response data pair at different time scales, and obtain a coefficient of variation sequence of the certain electrothermal response data pair; The calculation module is configured to calculate the dynamic time warping distance between the first coefficient of variation sequence of the first electrothermal response data pair and the second coefficient of variation sequence of the second electrothermal response data pair, wherein the first electrothermal response data pair and the second electrothermal response data pair are electrothermal response data pairs corresponding to two temporally adjacent abnormal fluctuation intervals; The judgment module is configured to determine whether the dynamic time warp distance is greater than a preset distance threshold; The identification module is configured to define the first electrothermal response data pair and the second electrothermal response data pair as the first suspicious data pair and the second suspicious data pair respectively if the dynamic time warp distance is greater than a preset distance threshold, and to identify the overheating fault type of the first suspicious data pair and the second suspicious data pair according to a preset first analysis strategy, so as to obtain the target overheating fault type identification result.

Images