Early warning method, device and equipment for cable joint and storage medium

Abstract

The invention discloses a cable joint early warning method, device and equipment and a storage medium, and the method comprises the steps: carrying out the preprocessing of a collected effective temperature data set, and constructing an initial temperature relation model; continuously acquiring real-time surface temperature data and environment temperature data of the cable joint, and calculating real-time internal temperature and temperature anomaly degree by using the initial temperature relation model and the temperature data acquired in real time; according to the temperature abnormal degree and the change trend thereof, performing multi-stage division on the abnormal condition, and generating temperature early warning information containing the abnormal degree and the position information; and when the temperature early warning information appears, retrieving a pre-established historical temperature anomaly case library, finding out historical cases similar to the current temperature anomaly condition, and generating targeted temperature anomaly processing suggestions according to the retrieved similar historical cases and the current power grid operation condition. According to the technical scheme of the invention, the potential abnormal temperature condition of the cable joint can be identified timely and accurately, and effective early warning and processing suggestions can be provided.

Description

Technical Field

[0001] The present invention relates to the technical field of cable joint early warning, and in particular to a cable joint early warning method, device, equipment and storage medium. Background Art

[0002] Cable joints are important components used to connect two sections of cables in power systems. They ensure electrical connection and mechanical strength between cables while providing necessary insulation and protection functions. The design of cable joints needs to consider multiple factors, including voltage level, current capacity, environmental conditions, etc., to ensure the reliable operation of the power system. In modern power grids, cable joints are widely used in underground cable systems, submarine cables, high-voltage transmission lines and other scenarios, and are a key link in ensuring the stability and safety of power transmission.

[0003] However, cable joints are prone to temperature anomalies during long-term operation, which is caused by a variety of factors, such as increased contact resistance, insulation aging, and changes in ambient humidity. Temperature anomalies not only reduce the service life of cable joints, but in severe cases may also cause cable joint failures, leading to power system interruptions and safety accidents. Traditional cable joint monitoring methods often rely on regular inspections or simple temperature threshold alarms, which make it difficult to identify potential temperature anomalies in a timely and accurate manner, and are unable to provide effective early warnings and treatment suggestions. Summary of the invention

[0004] The main purpose of the present invention is to solve the problem that the existing cable joint monitoring method is difficult to timely and accurately identify the potential temperature abnormality of the cable joint, and is unable to provide effective early warning and processing suggestions.

[0005] A first aspect of the present invention provides an early warning method for a cable joint, the early warning method for a cable joint comprising: Collect the internal temperature data, surface temperature data and ambient temperature data of the cable joint to obtain a valid temperature data set; Preprocessing the effective temperature data set to obtain a standardized historical temperature data set; According to the standardized historical temperature data set, an initial temperature relationship model is constructed to characterize the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint; Continuously collecting real-time surface temperature data and ambient temperature data of the cable joint, and calculating the real-time internal temperature and the degree of temperature anomaly using the initial temperature relationship model and the real-time collected temperature data; According to the abnormal temperature degree and its changing trend, the abnormal condition is divided into multiple levels, and temperature warning information including abnormal degree and location information is generated; When the temperature warning information appears, the pre-established historical temperature anomaly case library is searched to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status.

[0006] Optionally, collecting the internal temperature data, the surface temperature data and the ambient temperature data of the cable connector to obtain a valid temperature data set includes: An internal temperature sensor is implanted at the core position of the cable joint, multiple surface temperature sensors are evenly arranged along the circumference of the cable joint surface, and an ambient temperature sensor is set in the surrounding environment of the cable joint; Performing time synchronization setting on the internal temperature sensor, the surface temperature sensor and the ambient temperature sensor to obtain a unified data acquisition time series; Setting a temperature data collection frequency, and synchronously collecting the internal temperature data of the internal temperature sensor, the surface temperature data of the surface temperature sensor, and the ambient temperature data of the ambient temperature sensor according to the unified data collection time sequence; The collected temperature data were preliminarily verified, and obvious outliers were removed to obtain a valid temperature data set.

[0007] Optionally, preprocessing the effective temperature data set to obtain a standardized historical temperature data set includes: Performing time series segmentation on the effective temperature data set, dividing the continuous temperature data into data segments of fixed length; Performing statistical analysis on the temperature data in each of the data segments, calculating the mean, standard deviation, maximum and minimum values ​​of the temperature, and obtaining temperature statistical characteristics; The temperature data in the data segment is smoothed by using a moving average filtering method to eliminate short-term fluctuations and obtain a smoothed temperature curve; Normalizing the smoothed temperature curve to map temperature data of different dimensions into a unified interval to obtain standardized temperature data; The temperature statistical features are combined with the standardized temperature data to form a standardized historical temperature data set containing statistical information and time series information.

[0008] Optionally, constructing an initial temperature relationship model characterizing the relationship between the internal temperature, surface temperature and ambient temperature of the cable connector according to the standardized historical temperature data set includes: Extracting features from the standardized historical temperature data set, selecting feature parameters that are closely related to temperature, and constructing a feature vector set; Performing dimensionality reduction processing on the feature vector set by using a principal component analysis method to obtain a feature data set after dimensionality reduction; A support vector regression algorithm is adopted, with the feature data set after dimension reduction as input and the internal temperature as output, to train a support vector regression model; Using a genetic algorithm to optimize the hyperparameters of the support vector regression model to obtain an optimized support vector regression model; The optimized support vector regression model is used as the initial temperature relationship model.

[0009] Optionally, the continuously collecting real-time surface temperature data and ambient temperature data of the cable joint, and calculating the real-time internal temperature and the degree of temperature anomaly using the initial temperature relationship model and the real-time collected temperature data, comprises: According to a preset sampling frequency, the real-time surface temperature data and the ambient temperature data of the cable joint are continuously collected to form a real-time temperature data stream; Performing sliding window processing on the real-time temperature data stream to divide the continuous real-time temperature data into data windows of fixed length; Performing feature extraction and standardization processing on the real-time temperature data in each of the data windows to obtain a real-time feature vector consistent with the input format of the initial temperature relationship model; Inputting the real-time feature vector into the initial temperature relationship model to calculate and obtain the predicted real-time internal temperature; The deviation between the predicted real-time internal temperature and a preset normal temperature range is calculated, and the deviation is used as a quantitative indicator of the degree of temperature abnormality.

[0010] Optionally, the abnormal condition is divided into multiple levels according to the abnormal temperature degree and its changing trend, and temperature warning information including abnormal degree and location information is generated, including: Performing a time series analysis on the temperature anomaly degree, calculating the rate of change and acceleration of the temperature anomaly degree, and obtaining the temperature anomaly trend characteristics; According to the temperature anomaly degree and the temperature anomaly trend characteristics, the abnormal condition is evaluated using a fuzzy logic algorithm to obtain a fuzzy anomaly score; Based on a preset multi-level threshold, the fuzzy anomaly score is divided into different anomaly levels, the anomaly levels including normal, slightly abnormal, moderately abnormal and severely abnormal; Acquire the global position information of the cable joint and the local position information where the abnormality occurs, and associate the global position information and the local position information with the abnormality level; Generate structured temperature warning information including abnormality degree and location information according to the abnormality level, the global location information, the local location information and the temperature abnormality trend characteristics; The structured temperature warning information is prioritized to determine the release order and notification scope of the warning information.

[0011] Optionally, when the temperature warning information appears, a pre-established historical temperature anomaly case library is retrieved to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status, including: When the temperature warning information appears, feature extraction is performed on the temperature warning information to construct a multi-dimensional feature vector including abnormal degree, abnormal location, abnormal development trend and environmental conditions; Using the multi-dimensional feature vector as a query condition, searching a pre-established historical temperature anomaly case library, and preliminarily screening out potential similar cases; Calculating the similarity between the multidimensional feature vector and the feature vector of the potential similar case, using a weighted Euclidean distance method to obtain a similarity score; According to the similarity score, select the top N historical cases with the highest similarity to form a candidate case set; Obtaining current grid operation status, including load level, grid topology, and available resources, and comparing it with historical cases in the candidate case set; Based on the comparison results of the processing methods in the candidate case set and the current power grid operation status, generate preliminary temperature anomaly processing suggestions; Conduct a feasibility assessment on the preliminary temperature anomaly handling suggestions, consider the difficulty of implementation, required resources and potential risks, and form final targeted temperature anomaly handling suggestions.

[0012] A second aspect of the present invention provides an early warning device for a cable joint, the early warning device for the cable joint comprising: A data acquisition unit is used to collect the internal temperature data, surface temperature data and ambient temperature data of the cable connector to obtain a valid temperature data set; A data preprocessing unit, used for preprocessing the effective temperature data set to obtain a standardized historical temperature data set; A model building unit, used to build an initial temperature relationship model that characterizes the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint according to the standardized historical temperature data set; A real-time monitoring unit, used to continuously collect real-time surface temperature data and ambient temperature data of the cable joint, and calculate the real-time internal temperature and the degree of temperature anomaly by using the initial temperature relationship model and the real-time collected temperature data; An early warning generating unit, configured to divide the abnormal condition into multiple levels according to the abnormal temperature degree and its changing trend, and generate temperature early warning information including abnormal degree and location information; The decision support unit is used to retrieve a pre-established historical temperature anomaly case library when the temperature warning information appears, find historical cases similar to the current temperature anomaly condition, and generate targeted temperature anomaly handling suggestions based on the retrieved similar historical cases and the current power grid operation status.

[0013] The third aspect of the present invention provides an early warning device for a cable joint, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected through a line; the at least one processor calls the instructions in the memory so that the early warning device for the cable joint executes the steps of the above-mentioned early warning method for the cable joint.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the above-mentioned cable joint early warning method.

[0015] The cable joint early warning method proposed in the present invention effectively solves the problem that traditional monitoring methods are difficult to timely and accurately identify temperature anomalies through multi-dimensional data collection and in-depth analysis. First, the method collects the internal temperature, surface temperature and ambient temperature data of the cable joint, constructs a comprehensive temperature data set, and provides a rich information basis for anomaly detection. By preprocessing and standardizing these data, the method eliminates data noise and inconsistency, and improves the accuracy of subsequent analysis. Based on the standardized historical temperature data set, the method constructs an initial temperature relationship model that characterizes the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint. This model can capture the complex relationship of temperature changes and provide a reliable theoretical basis for real-time internal temperature estimation and anomaly detection. By continuously collecting real-time temperature data and analyzing it using the established model, the method can quickly calculate the real-time internal temperature and the degree of temperature anomaly, greatly improving the real-time and accuracy of anomaly detection.

[0016] In addition, the multi-level division and intelligent early warning mechanism of this method further enhance the accuracy and operability of the early warning. By analyzing the degree of temperature anomaly and its changing trend, the method can make a detailed multi-level division of abnormal conditions and generate structured early warning information containing the degree of anomaly and location information. This refined early warning mechanism can not only accurately reflect the severity of the anomaly, but also provide clear location guidance for operation and maintenance personnel, greatly improving the efficiency of fault location and processing. More importantly, when the early warning information appears, the method can automatically retrieve the historical temperature anomaly case library, find similar cases, and generate targeted temperature anomaly processing suggestions in combination with the current power grid operation status. This function makes full use of historical experience and current situations, provides scientific and effective decision-making support for operation and maintenance personnel, and significantly improves the efficiency and accuracy of abnormal processing. In general, this method realizes accurate early warning and intelligent processing of cable joint temperature anomalies through data-driven and intelligent analysis, providing a strong technical guarantee for the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0018] Figure 1 It is a schematic diagram of an embodiment of an early warning method for a cable joint in an embodiment of the present invention; Figure 2 A schematic diagram of an embodiment of an early warning device for a cable joint in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of an early warning device for a cable joint in an embodiment of the present invention.

[0019] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0022] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, which must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0023] An embodiment of the present application provides an early warning method for a cable joint. Figure 1 A flowchart of a cable joint early warning method provided in an embodiment of the present application. The method comprises the following steps: See also Figure 1 , collect the internal temperature data, surface temperature data and ambient temperature data of the cable connector to obtain a valid temperature data set; In one embodiment of the present invention, the internal temperature data, surface temperature data and ambient temperature data of the cable connector are collected to obtain a valid temperature data set, including: implanting an internal temperature sensor at the core position of the cable connector, evenly arranging multiple surface temperature sensors along the circumferential direction on the surface of the cable connector, and setting an ambient temperature sensor in the surrounding environment of the cable connector; performing time synchronization setting on the internal temperature sensor, the surface temperature sensor and the ambient temperature sensor to obtain a unified data collection time series; setting the temperature data collection frequency, and synchronously collecting the internal temperature data of the internal temperature sensor, the surface temperature data of the surface temperature sensor and the ambient temperature data of the ambient temperature sensor according to the unified data collection time series; performing preliminary verification on the collected temperature data, eliminating obvious abnormal values, and obtaining a valid temperature data set.

[0024] Specifically, it is necessary to first install temperature sensors at different locations in the cable joint. Specifically, an internal temperature sensor is implanted in the core of the cable joint, which is usually achieved by reserving the sensor installation position during the cable joint manufacturing process. For example, a micro thermocouple or fiber grating sensor can be buried at the interface between the cable core and the insulation layer. This method can directly measure the actual temperature inside the cable joint, which is crucial for early detection of potential hot spot problems.

[0025] Next, multiple surface temperature sensors are evenly arranged along the circumference of the cable joint surface. This can be achieved by pre-designing multiple sensor installation points on the cable joint shell, usually 4 to 8 points are selected to cover the entire joint surface 360 ​​degrees. For example, a thermal resistor or infrared sensor can be used and fixed to the joint surface by gluing or embedding. This arrangement can fully monitor the temperature distribution on the joint surface and help detect local overheating.

[0026] At the same time, an ambient temperature sensor is set in the environment around the cable joint. This is usually done by installing an independent temperature sensor at a certain distance (such as 50cm) from the cable joint to measure the temperature of the surrounding air. The measurement of ambient temperature is very important for distinguishing the heating of the cable joint itself from the influence of the external environment.

[0027] In order to ensure the time consistency of the data, the internal temperature sensor, surface temperature sensor and ambient temperature sensor need to be synchronized. This can be achieved by using a unified clock source or the Network Time Protocol (NTP). For example, a central controller can be set up to send synchronization signals to all sensors regularly to ensure that their clocks remain consistent. The purpose of this is to obtain a unified data acquisition time series, making subsequent data analysis more accurate and reliable.

[0028] After completing the installation and synchronization of the sensor, you need to set an appropriate temperature data acquisition frequency. The selection of this frequency requires a balance between data accuracy and system load. It can usually be set to collect data once per second or once per minute, depending on the working environment and monitoring requirements of the cable joint. For example, for high-voltage cable joints, a higher acquisition frequency is required to capture rapid temperature changes.

[0029] According to the set unified data collection time sequence, the system will synchronously collect the internal temperature data of the internal temperature sensor, the surface temperature data of the surface temperature sensor, and the ambient temperature data of the ambient temperature sensor. This synchronous collection ensures the consistency of temperature data at different locations in time, providing a reliable data basis for subsequent temperature relationship analysis.

[0030] Finally, the collected temperature data is preliminarily verified and obvious outliers are removed to obtain a valid temperature data set. This step can be achieved by setting a reasonable temperature range threshold. For example, if the reading of a sensor suddenly jumps to 300°C, far exceeding the normal operating temperature range, it can be regarded as an outlier and removed. In addition, statistical methods, such as standard deviation tests, can be used to identify and remove abnormal data points. The purpose of this step is to improve data quality and avoid interference from outliers in subsequent analysis.

[0031] Through the above steps, not only can the temperature data of the cable joint be collected comprehensively and accurately, but also the time consistency and validity of the data can be ensured. This multi-point, multi-dimensional temperature monitoring method provides a solid data foundation for subsequent temperature anomaly warnings, and significantly improves the accuracy and reliability of temperature anomaly detection. At the same time, by reasonably setting the acquisition frequency and data verification mechanism, it not only ensures the real-time nature of the data, but also avoids the interference caused by data redundancy and abnormal values, providing strong technical support for the safe operation monitoring of cable joints.

[0032] Please continue reading Figure 1 , preprocessing the effective temperature data set to obtain a standardized historical temperature data set; In one embodiment of the present invention, the effective temperature data set is preprocessed to obtain a standardized historical temperature data set, including: time series segmenting the effective temperature data set to divide continuous temperature data into data segments of fixed length; statistically analyzing the temperature data in each of the data segments to calculate the mean, standard deviation, maximum and minimum values ​​of the temperature to obtain temperature statistical characteristics; smoothing the temperature data in the data segment using a moving average filter to eliminate short-term fluctuations to obtain a smoothed temperature curve; normalizing the smoothed temperature curve to map temperature data of different dimensions into a unified interval to obtain standardized temperature data; combining the temperature statistical characteristics with the standardized temperature data to form a standardized historical temperature data set containing statistical information and time series information.

[0033] Specifically, first, the effective temperature data set is segmented into time series, and the continuous temperature data is divided into data segments of fixed length. This step can be achieved by setting a fixed time window, such as every 4 hours or every 24 hours as a data segment. Specifically, if there is a temperature data set as long as one month, it can be divided into 30 24-hour data segments. The advantage of this segmentation method is that it can capture the periodic characteristics of temperature changes, such as daily changes or load cycle changes, and it also facilitates subsequent parallel processing and improves the efficiency of data analysis.

[0034] Next, the temperature data in each data segment is statistically analyzed to calculate the mean, standard deviation, maximum and minimum values ​​of the temperature, thereby obtaining the temperature statistical features. This step can be achieved through simple mathematical calculations. For example, for a 24-hour data segment, assuming that data is collected once a minute, there will be 1440 data points. The average value of these 1440 data points is calculated as the mean, the maximum and minimum values ​​are directly selected from the data, and the standard deviation is obtained by calculating the average of the sum of the squares of the difference between each data point and the mean and then taking the square root. These statistical features provide an overall overview of the temperature changes in the data segment and help to quickly identify abnormal temperature changes. For example, if the maximum value of a data segment is suddenly much higher than usual, or the standard deviation is abnormally large, it indicates a potential temperature anomaly.

[0035] Then, the temperature data in the data segment is smoothed using the moving average filter method to eliminate short-term fluctuations and obtain a smoothed temperature curve. The moving average filter can be implemented by setting a sliding window, such as a window of 10 data points, and then calculating the average value of the data in the window as the new value of the current point. This window will move gradually to cover the entire data segment. For example, if the original data is [20, 22, 19, 21, 23, 20, 18, 21, 22, 20]℃, after using a 5-point moving average, the new value of the middle point will be (19+21+23+20+18) / 5=20.2℃. This smoothing process can effectively remove random noise and short-term fluctuations in the data, highlight the main trend of temperature changes, and provide a more reliable basis for subsequent anomaly detection.

[0036] Normalize the smoothed temperature curve and map the temperature data of different dimensions to a unified interval to obtain standardized temperature data. This step usually uses the minimum-maximum normalization method to map the data to the [0, 1] interval. The specific calculation formula is: , where X is the original data, and are the minimum and maximum values ​​in the data segment, respectively. For example, if the temperature range of a data segment is 20°C to 40°C, then 30°C will be normalized to (30-20) / (40-20)=0.5. The advantage of normalization is that it eliminates the scale differences of data from sensors in different locations or in different time periods, making the data comparable, which is beneficial for subsequent model training and anomaly detection.

[0037] Finally, the temperature statistical features are combined with the standardized temperature data to form a standardized historical temperature data set containing statistical information and time series information. This step can be done by storing the statistical features (mean, standard deviation, maximum, minimum) of each data segment as additional feature columns in parallel with the standardized time series data. For example, for each 24-hour data segment, in addition to the 1440 standardized temperature data points, 4 statistical feature values ​​are added. This combination not only retains detailed time series information, but also provides a summary of the overall characteristics of the data segment, providing a rich information basis for subsequent temperature relationship model construction and anomaly detection.

[0038] The advantage of this entire preprocessing process is that it comprehensively considers the time series characteristics and statistical characteristics of temperature data, captures periodic changes through time series segmentation, extracts key features through statistical analysis, eliminates noise interference through smoothing, and achieves data standardization through normalization, ultimately forming an information-rich, easy-to-analyze standardized historical temperature data set. This multi-angle, multi-level data preprocessing method significantly improves the accuracy and reliability of subsequent temperature anomaly detection, laying a solid data foundation for intelligent monitoring and early warning of cable joints.

[0039] Please continue reading Figure 1 , constructing an initial temperature relationship model characterizing the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint according to the standardized historical temperature data set; In one embodiment of the present invention, the initial temperature relationship model characterizing the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint is constructed based on the standardized historical temperature data set, including: extracting features from the standardized historical temperature data set, selecting feature parameters closely related to the temperature relationship, and constructing a feature vector set; using a principal component analysis method to reduce the dimension of the feature vector set to obtain a reduced-dimensional feature data set; using a support vector regression algorithm, with the reduced-dimensional feature data set as input and the internal temperature as output, to train a support vector regression model; using a genetic algorithm to optimize the hyperparameters of the support vector regression model to obtain an optimized support vector regression model; and using the optimized support vector regression model as the initial temperature relationship model.

[0040] Specifically, it is necessary to first extract features from the standardized historical temperature data set, select feature parameters that are closely related to temperature, and construct a feature vector set. Specifically, select surface temperature, ambient temperature, load current, time (hour, date, season), etc. as feature parameters. For example, for a specific moment, the feature vector can be [0.6, 0.4, 0.8, 12, 182, 2], which respectively represent the standardized surface temperature, ambient temperature, load current, hour (12 o'clock), date (the 182nd day of the year), and season (2 represents summer). This feature selection fully considers the multiple factors that affect the internal temperature of the cable joint and provides a comprehensive information basis for subsequent modeling.

[0041] Next, the principal component analysis (PCA) method is used to reduce the dimensionality of the feature vector set to obtain the feature data set after dimensionality reduction. The principle of PCA is to project the original high-dimensional data into a new coordinate system so as to maximize the variance of the data in the new coordinate system. In the specific implementation, the covariance matrix of the feature vector set is first calculated, and then the eigenvalues ​​and eigenvectors of the matrix are solved, and the eigenvectors corresponding to the largest eigenvalues ​​are selected as new basis vectors. For example, if the original eigenvector is 6-dimensional, it will be reduced to 3 dimensions through PCA, retaining more than 80% of the information of the original data. The advantage of this step is that it reduces the redundancy of the data and the complexity of subsequent model training, while retaining the main information of the data and improving the generalization ability of the model.

[0042] Then, the support vector regression (SVR) algorithm is used to train the support vector regression model with the reduced feature data set as input and the internal temperature as output. The core idea of ​​SVR is to construct an optimal hyperplane in high-dimensional space so that the distance from all samples to this hyperplane is minimized. The mathematical expression of the SVR model is: .

[0043] in, is the predicted output value (in this case, the internal temperature), x is the input feature vector (such as surface temperature, ambient temperature, etc.), is the support vector in the training set (i.e., the sample points that have an important influence on the model), and is the Lagrange multiplier (indicating the importance of each support vector), is the kernel function (used to map the input space to a high-dimensional feature space), and b is the bias term.

[0044] In practical applications, commonly used kernel functions include linear kernels, polynomial kernels, and radial basis function (RBF) kernels. For example, when using the RBF kernel, , where γ is the kernel function parameter, which controls the width of the Gaussian distribution. The advantage of choosing the SVR algorithm is that it can handle nonlinear problems well and has good robustness to outliers, which is especially important for complex nonlinear systems such as cable joint temperature.

[0045] In order to further improve the performance of the model, the genetic algorithm is used to optimize the hyperparameters of the support vector regression model to obtain the optimized support vector regression model. The genetic algorithm simulates the process of biological evolution and searches for the optimal solution through selection, crossover and mutation operations. In the SVR model, the hyperparameters that need to be optimized include the penalty coefficient C (controlling the balance between model complexity and degree of fit), the insensitive loss function ε (defining the size of the penalty-free area) and the kernel function parameters (such as γ of the RBF kernel). The implementation steps of the genetic algorithm include: 1) Initialize a set of random hyperparameter combinations; 2) Use these hyperparameters to train the SVR model and calculate the performance of the model (such as mean squared error); 3) Select outstanding individuals based on performance; 4) Perform crossover and mutation operations on the selected individuals to generate a new generation of hyperparameter combinations; 5) Repeat steps 2-4 until the preset number of iterations is reached or the performance is no longer significantly improved. For example, the initial population contains 100 individuals, each of which represents a set of hyperparameters [C, ε, γ]. After multiple iterations, the optimal hyperparameter combination [10, 0.1, 0.01] is obtained. The advantage of using genetic algorithms to optimize hyperparameters is that it can search for the optimal solution in a large range, avoiding falling into local optimality, thereby obtaining a model with better performance.

[0046] Finally, the optimized support vector regression model is used as the initial temperature relationship model. This model can accurately predict the internal temperature of the cable joint based on the input surface temperature, ambient temperature and other features. For example, given a new set of input features [0.7, 0.5, 0.9, 14, 200, 2], the model can output a predicted internal temperature value, such as 0.8 (corresponding to the actual temperature of 40°C). This initial temperature relationship model lays the foundation for subsequent real-time temperature monitoring and anomaly detection.

[0047] The advantage of this entire modeling process is that it uses a variety of machine learning techniques, from feature extraction to dimensionality reduction, to model training and optimization, and each step is carefully designed. Feature extraction ensures the comprehensiveness of input data, PCA dimensionality reduction improves computational efficiency, the SVR algorithm handles nonlinear relationships well, and the application of genetic algorithms further optimizes model performance. This multi-step, multi-technique approach not only improves the accuracy and generalization ability of the model, but also enhances the model's adaptability to complex environments, providing strong technical support for cable joint temperature abnormality warning.

[0048] Please continue reading Figure 1 , continuously collecting real-time surface temperature data and ambient temperature data of the cable joint, and calculating the real-time internal temperature and the degree of temperature anomaly using the initial temperature relationship model and the real-time collected temperature data; In one embodiment of the present invention, the real-time surface temperature data and ambient temperature data of the cable joint are continuously collected, and the real-time internal temperature and the degree of temperature anomaly are calculated using the initial temperature relationship model and the real-time collected temperature data, including: continuously collecting the real-time surface temperature data and ambient temperature data of the cable joint according to a preset sampling frequency to form a real-time temperature data stream; performing sliding window processing on the real-time temperature data stream to divide the continuous real-time temperature data into data windows of fixed duration; performing feature extraction and standardization processing on the real-time temperature data in each of the data windows to obtain a real-time feature vector consistent with the input format of the initial temperature relationship model; inputting the real-time feature vector into the initial temperature relationship model to calculate the predicted real-time internal temperature; calculating the deviation between the predicted real-time internal temperature and the preset normal temperature range, and using the deviation as a quantitative indicator of the degree of temperature anomaly.

[0049] Specifically, according to the preset sampling frequency, the real-time surface temperature data and ambient temperature data of the cable joint are continuously collected to form a real-time temperature data stream. The selection of the sampling frequency requires a balance between data accuracy and system load, and can usually be set to collect data once per second or once per minute. For example, for a high-voltage cable joint, data will be collected once per second to capture rapid temperature changes. This high-frequency data collection can reflect the temperature status of the cable joint in a timely manner, providing a real-time and accurate data basis for anomaly detection.

[0050] Next, the real-time temperature data stream is subjected to sliding window processing to divide the continuous real-time temperature data into data windows of fixed length. The size of the sliding window is usually determined according to the thermal characteristics of the cable joint and actual needs, and can be 5 minutes, 15 minutes, or longer. For example, if a window size of 15 minutes is selected and the sampling frequency is once per second, each window will contain 900 data points. The step size of the sliding window can be smaller than the window size to achieve window overlap and increase data utilization. For example, it can slide every 5 minutes, so that two adjacent windows will overlap for 10 minutes. The advantage of sliding window processing is that it can capture local trends in temperature changes while smoothing short-term fluctuations and improving the stability of anomaly detection.

[0051] The real-time temperature data in each data window is feature extracted and standardized to obtain a real-time feature vector that is consistent with the input format of the initial temperature relationship model. The feature extraction process includes calculating the average temperature, maximum temperature, minimum temperature, temperature change rate, etc. in the window. For example, for a 15-minute window, the following features can be extracted: average surface temperature , Maximum surface temperature , minimum surface temperature , surface temperature change rate , average ambient temperature The normalization process usually uses the same method as when training the initial temperature relationship model, such as minimum-maximum normalization. This feature extraction and normalization process ensures the consistency of the real-time data and the model input format, which is conducive to the accurate prediction of the model.

[0052] Then, the real-time feature vector is input into the initial temperature relationship model to calculate the predicted real-time internal temperature. Assuming that the initial temperature relationship model is a support vector regression (SVR) model, its mathematical expression is: ; in, is the predicted internal temperature, x is the input real-time feature vector, is the support vector in the training set, and is the weight coefficient of each support vector, is a kernel function (such as radial basis function RBF: , b is the bias term. For example, if the real-time feature vector x = [0.7, 0.8, 0.6, 0.05, 0.5] (corresponding to the standardized ), the model outputs a predicted internal temperature value, such as (The corresponding actual temperature is 45°C.) This temperature prediction method based on machine learning can capture complex nonlinear relationships and improve the accuracy of predictions.

[0053] Finally, the deviation between the predicted real-time internal temperature and the preset normal temperature range is calculated and used as a quantitative indicator of the degree of temperature anomaly. The normal temperature range can be determined according to the specifications of the cable connector and the working environment, for example, set to 35°C to 55°C. The degree of temperature anomaly can be calculated using the following formula: ; Among them, AD is the abnormal degree (Anomaly Degree), is the predicted internal temperature, and are the upper and lower limits of the normal temperature range respectively. The meaning of this formula is: if In the normal range, AD is 0; if exceeds the upper limit, AD is proportional to the degree of excess; if Below the lower limit, AD is proportional to the degree of below. For example, if , , when the predicted When AD=(60-55) / (55-35)=0.25, it means the temperature abnormality is 25%.

[0054] The advantages of this real-time monitoring and analysis method are: First, high-frequency data collection ensures timely grasp of the temperature status of the cable joint. Second, the sliding window processing not only captures short-term temperature changes, but also smoothes instantaneous fluctuations, improving the stability of the analysis. Third, feature extraction and standardization processing enable real-time data to be seamlessly connected with the pre-trained model, ensuring the accuracy of the prediction. Finally, the prediction results are compared with the normal range to obtain a quantitative abnormality index, which provides an intuitive and operational basis for subsequent early warning and decision-making.

[0055] Please continue reading Figure 1 , according to the abnormal temperature degree and its changing trend, the abnormal condition is divided into multiple levels, and temperature warning information including abnormal degree and location information is generated; In one embodiment of the present invention, the abnormal condition is divided into multiple levels according to the temperature abnormality degree and its changing trend, and temperature warning information containing abnormality degree and location information is generated, including: performing time series analysis on the temperature abnormality degree, calculating the rate of change and acceleration of the temperature abnormality degree, and obtaining temperature abnormality trend characteristics; evaluating the abnormal condition using a fuzzy logic algorithm according to the temperature abnormality degree and the temperature abnormality trend characteristics to obtain a fuzzy abnormality score; based on a preset multi-level threshold, dividing the fuzzy abnormality score into different abnormality levels, the abnormality levels including normal, slight abnormality, moderate abnormality and severe abnormality; obtaining the global location information of the cable joint and the local location information where the abnormality occurs, and associating the global location information and the local location information with the abnormality level; generating structured temperature warning information containing abnormality degree and location information according to the abnormality level, the global location information, the local location information and the temperature abnormality trend characteristics; and prioritizing the structured temperature warning information to determine the release order and notification scope of the warning information.

[0056] Specifically, first, it is necessary to perform a time series analysis on the temperature anomaly degree, calculate the rate of change and acceleration of the temperature anomaly degree, and thus obtain the temperature anomaly trend characteristics. Specifically, assuming that there are corresponding temperature anomaly degrees AD1, AD2, ..., ADn in the time series t1, t2, ..., tn, the rate of change and acceleration can be calculated by the following formula: Rate of change: R=(ADn-AD(n-1)) / (tn-t(n-1)).

[0057] Acceleration: A=(Rn-R(n-1)) / (tn-t(n-1)).

[0058] Among them, R represents the rate of change, A represents acceleration, AD represents the degree of temperature anomaly, and t represents time. This calculation method can reflect the changing trend of the degree of temperature anomaly and provide an important basis for subsequent anomaly assessment. For example, if it is found that the degree of temperature anomaly is not only increasing (R>0), but also the rate of increase is accelerating (A>0), this indicates a rapidly deteriorating abnormal situation.

[0059] Next, according to the degree of temperature anomaly and the characteristics of temperature anomaly trend, the fuzzy logic algorithm is used to evaluate the abnormal condition and obtain the fuzzy anomaly score. The advantage of the fuzzy logic algorithm is that it can handle uncertainty and ambiguity, which is particularly important in the evaluation of temperature anomaly. In specific implementation, fuzzy sets and fuzzy rules can be defined. For example, the degree of temperature anomaly AD can be defined as three fuzzy sets: low (L), medium (M), and high (H), the rate of change R can be defined as negative (N), zero (Z), and positive (P), and the acceleration A can also be defined as negative (N), zero (Z), and positive (P). Then formulate fuzzy rules, such as: If AD=H, and R=P, and A=P, then the anomaly score is very high.

[0060] If AD=M, and R=Z, and A=N, then the anomaly score is medium.

[0061] If AD=L, and R=N, and A=N, then the anomaly score is low.

[0062] Through these rules and fuzzy reasoning, a fuzzy anomaly score between 0 and 1 can be obtained. This score comprehensively considers the degree of temperature anomaly and its changing trend, providing a more comprehensive assessment of abnormal conditions.

[0063] Based on the preset multi-level thresholds, the fuzzy anomaly scores are divided into different anomaly levels, including normal, slight anomaly, moderate anomaly and severe anomaly. For example, the following thresholds can be set: 0≤score<0.3: normal.

[0064] 0.3≤Score<0.5: Slightly abnormal.

[0065] 0.5≤Score<0.8: Moderately abnormal.

[0066] 0.8≤Score≤1.0: Severe abnormality.

[0067] This multi-level division can more accurately reflect the severity of the abnormal situation and facilitate the adoption of corresponding treatment measures.

[0068] After determining the abnormality level, it is necessary to obtain the global location information of the cable joint and the local location information where the abnormality occurred, and associate these location information with the abnormality level. The global location information can be the geographical coordinates or number of the cable joint in the entire power grid, such as "Substation A-3 Line-15 kilometers"; the local location information can be the specific location of the cable joint where the abnormality occurred, such as "north side of the joint shell". This association of location information is crucial for quickly locating and handling abnormalities.

[0069] According to the abnormal level, global location information, local location information and temperature abnormal trend characteristics, structured temperature warning information containing abnormal degree and location information is generated. This structured information can be in JSON format, for example: { "Abnormality level": "Moderate abnormality", "Global Location": "Substation A-Line 3-15km", "Local Position": "North side of connector shell", "Temperature Anomaly Degree": 0.65, "Rate of change": 0.02, "acceleration": 0.001, "Timestamp": "2024-10-01 14:30:00" } This structured warning information facilitates automatic system processing and quick human understanding, improving the efficiency and accuracy of information transmission.

[0070] Finally, the structured temperature warning information is prioritized to determine the order of release and notification scope of the warning information. The sorting can be based on multiple factors, such as abnormality level, temperature abnormality degree, change rate, acceleration, etc. For example, a weighted summation method can be used: Priority score = w1×abnormality level + w2×temperature abnormality degree + w3×change rate + w4×acceleration.

[0071] Among them, w1, w2, w3, and w4 are weight coefficients, which can be adjusted according to actual conditions. The higher the priority score, the higher the priority of the warning information. According to the priority, the scope and method of notification can be determined. For example, high-priority warnings need to be notified to on-site staff and management immediately, while low-priority warnings only need to be recorded in the system log.

[0072] The advantages of this whole process are: it not only considers the static degree of temperature anomaly, but also analyzes its dynamic change trend, providing a more comprehensive anomaly assessment; the use of fuzzy logic algorithm can better handle the uncertainty in temperature anomaly assessment; multi-level anomaly division and structured warning information generation make the warning more accurate and easy to understand; the priority sorting mechanism ensures that the most urgent situations can get the fastest response. This method is of great significance in the preventive maintenance of power systems, which can effectively reduce the risk of equipment failure and improve the reliability and safety of the system.

[0073] Please continue reading Figure 1When the temperature warning information appears, the previously established historical temperature anomaly case library is retrieved to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status.

[0074] In one embodiment of the present invention, when the temperature warning information appears, a pre-established historical temperature anomaly case library is retrieved to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status, including: when the temperature warning information appears, feature extraction is performed on the temperature warning information to construct a multidimensional feature vector including the degree of anomaly, the location of anomaly, the development trend of anomaly and the environmental conditions; the multidimensional feature vector is used as a query condition to retrieve the pre-established historical temperature anomaly case library to preliminarily screen out potential similar cases; the multidimensional feature vector is calculated and compared with the The similarity between the feature vectors of the potential similar cases is calculated by using a weighted Euclidean distance method to obtain a similarity score; based on the similarity score, the top N historical cases with the highest similarity are selected to form a candidate case set; the current power grid operation status, including the load level, power grid topology and available resources, is obtained, and compared with the historical cases in the candidate case set; based on the comparison results of the processing methods in the candidate case set and the current power grid operation status, a preliminary temperature anomaly processing suggestion is generated; a feasibility assessment is performed on the preliminary temperature anomaly processing suggestion, and a final targeted temperature anomaly processing suggestion is formed by considering the implementation difficulty, required resources and potential risks.

[0075] Specifically, when temperature warning information appears, it is necessary to first extract features from this information and construct a multidimensional feature vector containing the degree of abnormality, abnormal location, abnormal development trend and environmental conditions. This process can be achieved by converting the key parameters in the warning information into numerical form. For example, the degree of abnormality can be represented by a numerical value between 0 and 1, the abnormal location can be converted into a coordinate value relative to a reference point, the abnormal development trend can be represented by the temperature change rate and acceleration, and the environmental conditions can include parameters such as the current temperature and humidity. Suppose an 8-dimensional feature vector is constructed: [0.7, 100, 50, 0.02, 0.001, 35, 60, 1], where 0.7 represents the degree of abnormality, 100 and 50 represent the x and y coordinates of the abnormal location, 0.02 and 0.001 represent the temperature change rate and acceleration respectively, 35 represents the ambient temperature, 60 represents humidity, and 1 represents weather conditions (such as sunny days). The construction of this multidimensional feature vector provides a basis for subsequent case retrieval and similarity calculation.

[0076] Next, using this multidimensional feature vector as a query condition, the pre-established historical temperature anomaly case library is searched to preliminarily screen out potential similar cases. This step can be achieved by setting a preliminary similarity threshold, and only cases whose similarity exceeds this threshold will be selected as potential similar cases. For example, cosine similarity can be used as a preliminary screening criterion. The calculation formula for cosine similarity is: cos(θ)=(A·B) / (||A||||B||), where A and B represent the query vector and the historical case vector, respectively, · represents the dot product of the vector, and ||A|| represents the modulus of vector A. Assuming that the threshold is set to 0.8, only historical cases whose cosine similarity with the query vector is greater than 0.8 will be selected as potential similar cases. This preliminary screening can greatly reduce the number of cases that require detailed comparison later and improve retrieval efficiency.

[0077] Then, the similarity between the multidimensional feature vector and the feature vector of the potential similar case is calculated, and the similarity score is obtained using the weighted Euclidean distance method. The setting of weights reflects the importance of different features in the similarity calculation. For example, higher weights will be given to the degree of abnormality and the trend of abnormal development, because these factors are more critical to the selection of processing methods. The similarity score can be obtained by converting the distance to similarity, such as s=1 / (1+d), so that the smaller the distance, the higher the similarity score. This weighted Euclidean distance method takes into account the differences in the importance of each feature and can more accurately reflect the similarity between cases.

[0078] According to the similarity score, the top N historical cases with the highest similarity are selected to form a candidate case set. The selection of N requires a trade-off between accuracy and efficiency, and usually 5 to 10 cases can be selected. These candidate cases will provide a reference for the subsequent generation of processing suggestions. At the same time, the current grid operation status, including load level, grid topology, and available resources, is obtained and compared with the historical cases in the candidate case set. The purpose of this step is to take into account the specific conditions of the current grid, because the same temperature anomaly requires different processing measures under different grid operation conditions. For example, when the grid load is light, a more aggressive processing method can be adopted, while when the load is heavy, more caution is required.

[0079] Based on the comparison results of the treatment methods in the candidate case set and the current grid operation status, a preliminary temperature anomaly treatment suggestion is generated. This process can adopt a weighted voting method, that is, a weight is given according to the similarity score of each candidate case and the applicability of its treatment method (considering the current grid status), and then the weighted votes of various treatment methods are counted, and the method with the highest votes will be used as a preliminary suggestion. For example, if there are 3 highly similar cases that all recommend reducing the load, and the current grid load level is high, then "reducing the load" will become a high-priority preliminary suggestion.

[0080] Finally, conduct a feasibility assessment on the preliminary temperature anomaly handling suggestions, consider the difficulty of implementation, required resources and potential risks, and form the final targeted temperature anomaly handling suggestions. This step can be achieved by establishing a scoring matrix, where the rows of the matrix represent different handling suggestions, the columns represent evaluation factors (such as implementation difficulty, required resources, potential risks, expected effects, etc.), and the value of each cell represents the score of the suggestion on that factor. By calculating the comprehensive score of each suggestion, the final handling suggestion can be obtained. For example, if "reducing the load" has the highest comprehensive score, then the final handling suggestion is "reducing the load of the line by 20% in the next 2 hours while closely monitoring temperature changes."

[0081] This method of generating treatment suggestions based on historical cases and current conditions has several significant advantages: first, it makes full use of historical experience and can quickly find effective treatment methods through retrieval and analysis of similar cases; second, it takes into account the actual operating conditions of the current power grid to ensure that the generated suggestions are feasible; third, through multi-step screening and evaluation, it can generate treatment suggestions that are both targeted and feasible; finally, this method has strong adaptability and scalability, and as the historical case library continues to be enriched, the quality of treatment suggestions will continue to improve.

[0082] The above describes the early warning method of the cable joint in the embodiment of the present invention. The following describes the early warning device of the cable joint in the embodiment of the present invention. Figure 2 , an embodiment of the early warning device of the cable joint in the embodiment of the present invention includes: The data collection unit 201 is used to collect the internal temperature data, surface temperature data and ambient temperature data of the cable connector to obtain a valid temperature data set; A data preprocessing unit 202 is used to preprocess the effective temperature data set to obtain a standardized historical temperature data set; A model building unit 203 is used to build an initial temperature relationship model that characterizes the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint according to the standardized historical temperature data set; The real-time monitoring unit 204 is used to continuously collect the real-time surface temperature data and the ambient temperature data of the cable joint, and calculate the real-time internal temperature and the degree of temperature anomaly by using the initial temperature relationship model and the real-time collected temperature data; The warning generation unit 205 is used to divide the abnormal condition into multiple levels according to the abnormal temperature degree and its change trend, and generate temperature warning information including abnormal degree and location information; The decision support unit 206 is used to retrieve a pre-established historical temperature anomaly case library when the temperature warning information appears, find historical cases similar to the current temperature anomaly condition, and generate targeted temperature anomaly processing suggestions based on the retrieved similar historical cases and the current power grid operation status.

[0083] above Figure 2 The early warning device for the cable joint in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The early warning device for the cable joint in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0084] Figure 3 1 is a schematic diagram of the structure of an early warning device for a cable joint provided by an embodiment of the present invention. The early warning device 300 for the cable joint may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 may be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the early warning device 300 for the cable joint. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the early warning device 300 for the cable joint to implement the steps of the early warning method for the above-mentioned cable joint.

[0085] The cable joint warning device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The structure of the early warning device for the cable joint shown does not constitute a limitation on the early warning device for the cable joint provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0086] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the cable connector early warning method.

[0087] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0088] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0089] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A cable joint early warning method, It is characterized in that include: Collect the internal temperature data, surface temperature data and ambient temperature data of the cable joint to obtain a valid temperature data set; Preprocessing the effective temperature data set to obtain a standardized historical temperature data set; According to the standardized historical temperature data set, an initial temperature relationship model is constructed to characterize the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint; Continuously collecting real-time surface temperature data and ambient temperature data of the cable joint, and calculating the real-time internal temperature and the degree of temperature anomaly using the initial temperature relationship model and the real-time collected temperature data; According to the abnormal temperature degree and its changing trend, the abnormal condition is divided into multiple levels, and temperature warning information including abnormal degree and location information is generated; When the temperature warning information appears, the pre-established historical temperature anomaly case library is searched to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status.

2. The cable joint early warning method according to claim 1, It is characterized in that The collecting of the internal temperature data, the surface temperature data and the ambient temperature data of the cable joint to obtain the effective temperature data set includes: An internal temperature sensor is implanted at the core position of the cable joint, multiple surface temperature sensors are evenly arranged along the circumference of the cable joint surface, and an ambient temperature sensor is set in the surrounding environment of the cable joint; Performing time synchronization setting on the internal temperature sensor, the surface temperature sensor and the ambient temperature sensor to obtain a unified data acquisition time series; Setting a temperature data collection frequency, and synchronously collecting the internal temperature data of the internal temperature sensor, the surface temperature data of the surface temperature sensor, and the ambient temperature data of the ambient temperature sensor according to the unified data collection time sequence; The collected temperature data were preliminarily verified, and obvious outliers were removed to obtain a valid temperature data set.

3. The cable joint early warning method according to claim 1, It is characterized in that The preprocessing of the effective temperature data set to obtain a standardized historical temperature data set includes: Performing time series segmentation on the effective temperature data set, dividing the continuous temperature data into data segments of fixed length; Performing statistical analysis on the temperature data in each of the data segments, calculating the mean, standard deviation, maximum and minimum values ​​of the temperature, and obtaining temperature statistical characteristics; The temperature data in the data segment is smoothed by using a moving average filtering method to eliminate short-term fluctuations and obtain a smoothed temperature curve; Normalizing the smoothed temperature curve to map temperature data of different dimensions into a unified interval to obtain standardized temperature data; The temperature statistical features are combined with the standardized temperature data to form a standardized historical temperature data set containing statistical information and time series information.

4. The cable joint early warning method according to claim 1, It is characterized in that The initial temperature relationship model characterizing the relationship between the internal temperature, surface temperature and ambient temperature of the cable joint is constructed according to the standardized historical temperature data set, including: Extracting features from the standardized historical temperature data set, selecting feature parameters that are closely related to temperature, and constructing a feature vector set; Performing dimensionality reduction processing on the feature vector set by using a principal component analysis method to obtain a feature data set after dimensionality reduction; A support vector regression algorithm is adopted, with the feature data set after dimension reduction as input and the internal temperature as output, to train a support vector regression model; Using a genetic algorithm to optimize the hyperparameters of the support vector regression model to obtain an optimized support vector regression model; The optimized support vector regression model is used as the initial temperature relationship model.

5. The cable joint early warning method according to claim 1, It is characterized in that The continuously collecting the real-time surface temperature data and the ambient temperature data of the cable joint, and calculating the real-time internal temperature and the degree of temperature anomaly by using the initial temperature relationship model and the real-time collected temperature data, comprises: According to a preset sampling frequency, the real-time surface temperature data and the ambient temperature data of the cable joint are continuously collected to form a real-time temperature data stream; Performing sliding window processing on the real-time temperature data stream to divide the continuous real-time temperature data into data windows of fixed length; Performing feature extraction and standardization processing on the real-time temperature data in each of the data windows to obtain a real-time feature vector consistent with the input format of the initial temperature relationship model; Inputting the real-time feature vector into the initial temperature relationship model to calculate and obtain the predicted real-time internal temperature; The deviation between the predicted real-time internal temperature and a preset normal temperature range is calculated, and the deviation is used as a quantitative indicator of the degree of temperature abnormality.

6. The cable joint early warning method according to claim 1, It is characterized in that The abnormal condition is divided into multiple levels according to the abnormal temperature degree and its changing trend, and temperature warning information including abnormal degree and location information is generated, including: Performing a time series analysis on the temperature anomaly degree, calculating the rate of change and acceleration of the temperature anomaly degree, and obtaining the temperature anomaly trend characteristics; According to the temperature anomaly degree and the temperature anomaly trend characteristics, the abnormal condition is evaluated using a fuzzy logic algorithm to obtain a fuzzy anomaly score; Based on a preset multi-level threshold, the fuzzy anomaly score is divided into different anomaly levels, the anomaly levels including normal, slightly abnormal, moderately abnormal and severely abnormal; Acquire the global position information of the cable joint and the local position information where the abnormality occurs, and associate the global position information and the local position information with the abnormality level; Generate structured temperature warning information including abnormality degree and location information according to the abnormality level, the global location information, the local location information and the temperature abnormality trend characteristics; The structured temperature warning information is prioritized to determine the release order and notification scope of the warning information.

7. The cable joint early warning method according to claim 1, It is characterized in that When the temperature warning information appears, a pre-established historical temperature anomaly case library is retrieved to find historical cases similar to the current temperature anomaly condition, and targeted temperature anomaly handling suggestions are generated based on the retrieved similar historical cases and the current power grid operation status, including: When the temperature warning information appears, feature extraction is performed on the temperature warning information to construct a multi-dimensional feature vector including abnormal degree, abnormal location, abnormal development trend and environmental conditions; Using the multi-dimensional feature vector as a query condition, searching a pre-established historical temperature anomaly case library, and preliminarily screening out potential similar cases; Calculating the similarity between the multidimensional feature vector and the feature vector of the potential similar case, using a weighted Euclidean distance method to obtain a similarity score; According to the similarity score, select the top N historical cases with the highest similarity to form a candidate case set; Obtaining current grid operation status, including load level, grid topology, and available resources, and comparing it with historical cases in the candidate case set; Based on the comparison results of the processing methods in the candidate case set and the current power grid operation status, generate preliminary temperature anomaly processing suggestions; Conduct a feasibility assessment on the preliminary temperature anomaly handling suggestions, consider the difficulty of implementation, required resources and potential risks, and form final targeted temperature anomaly handling suggestions.

8. An early warning device for a cable joint, It is characterized in that The early warning device of the cable joint comprises: A data acquisition unit is used to collect the internal temperature data, surface temperature data and ambient temperature data of the cable connector to obtain a valid temperature data set; A data preprocessing unit, used for preprocessing the effective temperature data set to obtain a standardized historical temperature data set; A model building unit, used to build an initial temperature relationship model that characterizes the relationship between the internal temperature, surface temperature and ambient temperature of the cable connector according to the standardized historical temperature data set; A real-time monitoring unit, used to continuously collect real-time surface temperature data and ambient temperature data of the cable joint, and calculate the real-time internal temperature and the degree of temperature anomaly by using the initial temperature relationship model and the real-time collected temperature data; An early warning generating unit, configured to divide the abnormal condition into multiple levels according to the abnormal temperature degree and its changing trend, and generate temperature early warning information including abnormal degree and location information; The decision support unit is used to retrieve a pre-established historical temperature anomaly case library when the temperature warning information appears, find historical cases similar to the current temperature anomaly condition, and generate targeted temperature anomaly handling suggestions based on the retrieved similar historical cases and the current power grid operation status.

9. An early warning device for cable joints, It is characterized in that The cable joint warning device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the cable joint early warning device to perform the steps of the cable joint early warning method according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, It is characterized in that When the instructions are executed by the processor, the steps of the early warning method for the cable joint as described in any one of claims 1-7 are implemented.

Images