Online cable sheath circulation monitoring system and method

Abstract

The invention discloses an online cable sheath circulating current monitoring system and method, and relates to the technical field of cable sheath circulating current monitoring, and the method comprises the following steps: installing a high-precision sensor outside a cable sheath for continuously monitoring a current signal of sheath circulating current, and carrying out the noise reduction and cleaning of collected data; key features reflecting abnormal tiny circulating current accumulation are extracted from the preprocessed current signals, the extracted features are further processed in a set monitoring window, and the processed features serve as an input feature set. Through a high-precision sensor, a deep convolutional neural network and a clustering algorithm, tiny circulation anomalies in a cable sheath are accurately identified and classified, the tiny circulation anomalies are subdivided into an accumulation appearance type, a distributed appearance type and an accidental appearance type, an alarm is given in time when risk accumulation occurs, false alarms are reduced, the monitoring efficiency and the safety of the cable sheath are improved, and the safety of the cable sheath is improved. Faults caused by heat accumulation are prevented, and stable operation and power supply reliability of a cable system are ensured.

Description

Technical Field

[0001] The invention relates to the technical field of cable sheath circulation monitoring, and in particular to an online cable sheath circulation monitoring system and method. Background Art

[0002] The online cable sheath circulation monitoring system is a real-time monitoring system designed to detect the circulation in the cable sheath to ensure the safe operation of the cable. The system uses sensors and data acquisition equipment to measure the sheath current and temperature, and processes and feedbacks the cable's operating status in real time through data analysis models. If the system detects abnormal circulation or temperature rise, it will automatically alarm and prompt the operation and maintenance personnel to take measures to avoid potential failure risks such as cable overheating and sheath damage caused by excessive circulation, thereby extending the cable life and ensuring the reliability of the power supply system.

[0003] The prior art has the following deficiencies:

[0004] During the monitoring of cable sheath circulation, the cumulative effect of weak circulation anomalies may occur, especially when the amplitude of the circulation anomaly is small. These subtle anomalies usually do not reach the alarm threshold and are easily ignored by the system. However, if these undetected small-amplitude circulation anomalies persist for a long time, they will often form heat accumulation in specific areas, causing local overheating of the cable sheath. As the temperature continues to rise, the sheath insulation material will gradually undergo thermal aging and deterioration, the insulation performance will decrease, and even physical damage such as cracking and dissolution will occur. Ultimately, this insulation damage may weaken the electrical isolation between the sheath and the conductor, greatly increasing the risk of sheath failure or cable short circuit. Long-term neglect of the cumulative impact of small circulation anomalies not only poses a threat to the safety and stability of the cable system, but may also cause large-scale failures and affect the normal operation of the power supply system.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide an online cable sheath circulation monitoring system and method, which uses high-precision sensors for continuous monitoring, and combines deep convolutional neural network analysis and clustering algorithms to identify and classify small circulation anomalies in the cable sheath, and accurately classify small circulation anomalies in the cable sheath that may cause heat accumulation. By subdividing the abnormal conditions into "accumulated type", "dispersed type" and "occasional type", timely alarms are given when significant cumulative risks occur, and operation and maintenance personnel are notified to take measures; and for intermittent and occasional anomalies, selective monitoring and recording are performed to avoid frequent false alarms, which significantly improves the safety and maintenance efficiency of the cable sheath, reduces the risk of major failures such as cable overheating, insulation degradation, and even short circuits caused by the accumulation of small anomalies that are not handled in time, thereby ensuring the stable operation of the cable system and the reliability of power supply, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: an online cable sheath circulating current monitoring method, comprising the following steps:

[0008] Install high-precision sensors on the outside of the cable sheath to continuously monitor the current signal of the sheath circulation and perform noise reduction and cleaning on the collected data;

[0009] Extract key features reflecting abnormal accumulation of tiny circulating currents from the preprocessed current signal, further process the extracted features under a set monitoring window, and use the processed features as input feature sets;

[0010] The features in the input feature set are input into a pre-trained deep convolutional neural network for analysis. The abnormal accumulation of small circulations in the cable sheath circulation is identified through the prediction results of the deep convolutional neural network.

[0011] When abnormal micro-circulation is identified in the cable sheath circulation, data is continuously collected under the set monitoring window, and multiple sets of circulation change information are recorded. The data are analyzed using a cluster analysis algorithm, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type.

[0012] For abnormal accumulation of small circulations identified as "accumulation patterns" in cluster analysis, an alarm will be issued immediately to notify relevant personnel to handle the situation;

[0013] For abnormal accumulation of small circulations identified as "dispersed appearance" in cluster analysis, do not issue an alarm immediately, but maintain continuous monitoring to observe the subsequent circulation change trend. If the dispersed anomaly evolves into an accumulation appearance in subsequent monitoring, trigger an alarm immediately to prevent the accumulation of abnormal small circulations from developing into a more serious problem.

[0014] For the accumulation of abnormal small circulations identified as "occasional occurrences" in cluster analysis, they will be automatically recorded but no warning will be triggered after being identified during the monitoring process.

[0015] Preferably, high-precision sensors installed on the outside of the cable sheath are set at key monitoring positions of the sheath to ensure that comprehensive and accurate current signals can be captured; the sensors are installed close to the sheath surface to reduce signal attenuation, and high-temperature resistant, waterproof and dustproof packaging materials are used to adapt to harsh environments; the sensors are arranged evenly at specified intervals, with emphasis on cable joints and bends.

[0016] Preferably, key features reflecting the accumulation of abnormal tiny circulating currents are extracted from the preprocessed current signal, wherein the extracted key features include the duration, frequency and growth rate of the change rate of the current in the current signal over time of the irregular fluctuations in the current signal. After extraction, the duration, frequency and growth rate of the change rate of the current in the current signal over time of the irregular fluctuations in the current signal are processed under a set monitoring window to generate an irregular persistence index and a circulating current slope growth index, respectively. The irregular persistence index is used to quantify the cumulative degree of the duration and frequency of the irregular fluctuations in the current signal, and the circulating current slope growth index is used to quantify the cumulative situation of the growth rate of the change rate of the current in the current signal over time.

[0017] Preferably, after obtaining the irregular persistence index and the circulation slope growth index generated by further processing the current signal, the irregular persistence index and the circulation slope growth index are used as the input feature set, and the irregular persistence index and the circulation slope growth index in the input feature set are input into a pre-trained deep convolutional neural network to generate a circulation accumulation index, and the abnormal small circulation accumulation phenomenon in the cable sheath circulation is predicted by the circulation accumulation index, and the abnormal small circulation accumulation situation existing in the cable sheath circulation is identified.

[0018] Preferably, the circulation accumulation index generated when predicting the cable sheath circulation by the deep convolutional neural network is compared and analyzed with the preset circulation accumulation index reference threshold to identify the abnormal micro circulation accumulation in the cable sheath circulation. The specific identification steps are as follows:

[0019] If the circulation accumulation index is greater than or equal to the circulation accumulation index reference threshold, an abnormal micro-circulation accumulation signal is generated, indicating that abnormal micro-circulation accumulation exists in the cable sheath circulation;

[0020] If the circulating current accumulation index is less than the circulating current accumulation index reference threshold, a current normal operation signal is generated, indicating that the current fluctuation and change in the cable sheath are still within the normal range and no abnormal accumulation phenomenon is detected.

[0021] Preferably, the specific steps of processing the duration and frequency of the irregular fluctuations in the current signal within a set monitoring window to generate an irregular persistence index are as follows:

[0022] Under the set monitoring window, first calculate the duration of each irregular fluctuation in the current signal, and set the starting time point of the i-th irregular fluctuation in the current signal to be The end time of the i-th irregular fluctuation is Then the irregular fluctuation duration ΔT of the i-th irregular fluctuation is i The calculation expression is: By measuring the duration of irregular fluctuations ΔT i Perform accumulation and calculate the cumulative duration, the cumulative duration T total is the sum of the duration of all irregular fluctuations within the monitoring window, and the calculation expression is: Where n is the total number of irregular fluctuations detected within the monitoring window;

[0023] Calculate the frequency of irregular fluctuations in the current signal through the instantaneous frequency f irreg To represent the number of irregular fluctuations per unit time in the monitoring window, the calculation expression of instantaneous frequency is:

[0024]

[0025] , where T window is the total duration of the monitoring window. In order to consider the impact of frequency on the degree of accumulation, the weighted cumulative frequency F is introduced. weighted , use the attenuation factor α to reduce the weight when the number of fluctuations is small, and the calculation expression of the weighted cumulative frequency is:

[0026]

[0027] , where α is a constant less than 1, which controls the attenuation degree of irregular fluctuation frequency. is the instantaneous frequency of the i-th irregular fluctuation;

[0028] Use cumulative duration T total and weighted cumulative frequency F weighted Generate Irregularity Persistence Index I persist , to comprehensively quantify the duration and frequency of irregular fluctuations, the irregular persistence index I persist The generative expression for is: Among them, γ is the exponential parameter of the weighted cumulative frequency to increase the impact of high-frequency fluctuations on the irregular persistence index.

[0029] Preferably, the specific steps of generating the circulating current slope growth index after processing the growth amplitude of the rate of change of the current in the current signal over time under a set monitoring window are as follows:

[0030] After the current signal is preprocessed, the preprocessed current signal is represented by I(t) to extract the instantaneous rate of change of the current over time. In order to capture the severity of the current fluctuation, the growth rate of the current change rate within the monitoring time window is calculated. The calculation expression is:

[0031]

[0032] , where I(t) is the current value at time t, is the instantaneous rate of change of current over time, used to compare changes between time points, R(t) is the growth rate of the instantaneous rate of change, quantifying the relative change of current fluctuations;

[0033] In the monitoring window, the growth rate of the instantaneous change rate at each time point is calculated and accumulated to form a cumulative change rate function. The cumulative change rate function is defined as: Where C(t) is the cumulative change rate function, which represents the total accumulation of the change rate within the monitoring window, α j is the weight coefficient, reflecting the relative contribution of a time point to the accumulation, j represents the time point, N is the total number of sampling points in the monitoring time window, R(t j ) is t j The growth rate of the instantaneous rate of change at the moment;

[0034] The circulating current slope growth index is generated by using the cumulative change rate function C(t) to quantify the cumulative growth of the current change rate. The generation expression of the circulating current slope growth index is:

[0035]

[0036] , where Slope circu is the circulation slope growth index, which is used to quantify the accumulation of the growth amplitude of the current over time. β is the adjustment parameter, which controls the weight of the cumulative change rate function C(t) in the calculation. x The impact index indicates the degree of response to the maximum instantaneous change rate growth amplitude. max(R(t)) is the maximum value of the instantaneous change rate growth amplitude within the monitoring window, and ∈ is a small constant to avoid the denominator being zero.

[0037] Preferably, when there is an accumulation of abnormal micro-circulations in the cable sheath circulation, a cluster analysis algorithm is used to perform pattern analysis on the data, and the accumulation of abnormal micro-circulations is further divided. The specific steps are as follows:

[0038] Under the set monitoring window, the circulation accumulation index generated in several monitoring windows is continuously collected to establish a data set F, where F = {Cirl ac (t u )}={Cirl ac (t 1 ), Cirl ac (t 2 ), ..., Cirl ac (t y )}, the circulation accumulation index in the data set reflects the accumulation trend and change of the circulation in the cable sheath over a period of time. ac (t u ) is the circulation accumulation index generated in the u-th monitoring window, and y is the total number of monitoring windows collected;

[0039] Extract the feature vector F from the circulation accumulation index within each monitoring window x , used for cluster analysis, eigenvector F x Defined as:

[0040]

[0041] , where ΔCirl ac (t u ) is the increment of the cumulative index of circulation between two adjacent monitoring windows, indicating the magnitude of the change. It represents the cumulative value of the circulation accumulation index in two adjacent monitoring windows, which is used to evaluate the overall trend. It represents the instantaneous rate of change of the circulation accumulation index, indicating the rate of change;

[0042] The generated feature vector F x Input into the clustering analysis algorithm, and define the expression of clustering objective function J as:

[0043]

[0044] , where K is the number of categories after clustering, C k represents the kth cluster, μ k is the cluster center of the kth cluster, which is used to measure the distance between the feature vector and the center point.

[0045] Preferably, each cluster center μ k and the preset cluster center reference threshold T agg Compare to determine the type of abnormal micro-circulation accumulation in the cable sheath circulation. The specific classification rules are as follows:

[0046] If μ k ≥T agg , then the cluster is divided into the accumulation type;

[0047] If 0.5T agg ≤μ k <T agg , then the cluster is classified as a dispersed appearance type;

[0048] If μ k <0.5T agg , then the cluster is classified as accidental occurrence type.

[0049] An online cable sheath circulation monitoring system, comprising a data acquisition and preprocessing module, a feature extraction and feature set construction module, a deep learning analysis and anomaly recognition module, a cluster analysis and anomaly classification module, an accumulation type anomaly alarm module, a decentralized type anomaly monitoring module and an occasional type anomaly recording module;

[0050] The data acquisition and preprocessing module installs high-precision sensors on the outside of the cable sheath to continuously monitor the current signal of the sheath circulation and perform noise reduction and cleaning on the collected data;

[0051] The feature extraction and feature set construction module extracts key features reflecting abnormal accumulation of small circulating currents from the preprocessed current signal, further processes the extracted features under the set monitoring window, and uses the processed features as the input feature set;

[0052] The deep learning analysis and anomaly recognition module inputs the features in the input feature set into a pre-trained deep convolutional neural network for analysis. Through the prediction results of the deep convolutional neural network, it identifies the abnormal accumulation of small circulations in the cable sheath circulation.

[0053] The cluster analysis and abnormal classification module, when identifying the presence of abnormal micro-circulation in the cable sheath circulation, continuously collects data under the set monitoring window and records multiple sets of circulation change information. The cluster analysis algorithm is used to perform pattern analysis on the data, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type.

[0054] The accumulation type abnormal alarm module immediately issues an alarm for abnormal micro-circulation accumulation identified as "accumulation type" in cluster analysis, and notifies relevant personnel to handle it;

[0055] The dispersed anomaly monitoring module does not immediately issue an alarm for abnormal micro-circulation accumulation identified as "dispersed appearance type" in cluster analysis, but maintains continuous monitoring to observe the subsequent circulation change trend. If the dispersed anomaly evolves into an accumulation type in subsequent monitoring, the alarm is immediately triggered to prevent the abnormal micro-circulation accumulation from developing into a more serious problem;

[0056] The accidental anomaly recording module automatically records the accumulation of abnormal micro-circulations identified as "occasional" in cluster analysis during the monitoring process, but does not trigger an early warning.

[0057] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0058] The present invention utilizes high-precision sensors for continuous monitoring, and combines deep convolutional neural network analysis and clustering algorithms to identify and classify tiny circulation anomalies in the cable sheath, and accurately classify tiny circulation anomalies in the cable sheath that may cause heat accumulation. By subdividing the abnormal conditions into "accumulated type", "dispersed type" and "occasional type", timely alarms are issued when significant cumulative risks occur, and operation and maintenance personnel are notified to take measures; and for intermittent and occasional anomalies, selective monitoring and recording are carried out to avoid frequent false alarms, which significantly improves the safety and maintenance efficiency of the cable sheath, and reduces the risks of major failures such as cable overheating, insulation degradation, and even short circuits caused by the accumulation of tiny anomalies that are not handled in time, thereby ensuring the stable operation of the cable system and the reliability of power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0060] Figure 1 The present invention is a method flow chart of an online cable sheath circulating current monitoring method.

[0061] Figure 2 The present invention is a module schematic diagram of an online cable sheath circulating current monitoring system. DETAILED DESCRIPTION

[0062] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0063] The present invention provides Figure 1 An online cable sheath circulating current monitoring method is shown, comprising the following steps:

[0064] Install high-precision sensors outside the cable sheath to continuously monitor the current signal of the sheath circulation;

[0065] High-precision sensors installed on the outside of the cable sheath should be set at key monitoring positions of the sheath to ensure that the most comprehensive and accurate current signals can be captured. The sensor should be installed close to the surface of the sheath to reduce signal attenuation, and use high-temperature resistant, waterproof, and dustproof packaging materials to adapt to harsh environments such as outdoor or underground. To ensure the continuity and stability of the data, the layout of the sensors should be evenly distributed at a certain distance, especially at the cable joints and bends. When installing the sensor, mechanical vibration interference should be avoided, and it should be fixed with high-strength adhesives or special clamps, while considering the convenience of maintenance and replacement.

[0066] High-precision sensors installed on the outside of the cable sheath should be set at key monitoring locations of the sheath, such as cable joints, connection points, and bends, which are more susceptible to mechanical stress and heat accumulation. In addition, sensors should be placed in areas of the cable path where high ambient temperatures or external electromagnetic interference may exist to capture potential abnormal circulation phenomena. It is important to place sensors near the transition section of the cable and the junction box, because these locations are areas where current distribution and sheath stress changes are more concentrated, which can provide more sensitive and accurate monitoring data and help to detect potential fault hazards early.

[0067] The sensor should have a high sampling rate and high sensitivity to ensure that it can capture extremely small changes in circulation. In order to ensure the real-time and accuracy of the data, the data acquisition system should adopt an anti-interference design to reduce the impact of the external environment on the monitoring results. At the same time, select a sensor bracket suitable for the installation location to ensure the stability and long-term working ability of the monitoring equipment in harsh environments.

[0068] The collected data may be affected by electromagnetic interference, noise and sensor errors, so it is necessary to reduce noise and clean the collected data;

[0069] A low-pass filter can be used to remove high-frequency noise, while wavelet transform technology can be used for multi-resolution analysis to retain signal details and eliminate unnecessary interference. This step can effectively improve signal quality and make subsequent analysis more reliable. In addition, in order to deal with sudden interference, an adaptive filtering algorithm can be introduced to dynamically adjust the noise reduction parameters.

[0070] Extract key features reflecting abnormal accumulation of tiny circulating currents from the preprocessed current signal, further process the extracted features under a set monitoring window, and use the processed features as input feature sets;

[0071] The features in the input feature set are input into a pre-trained deep convolutional neural network for analysis. The abnormal accumulation of small circulations in the cable sheath circulation is identified through the prediction results of the deep convolutional neural network.

[0072] The key features reflecting the accumulation of abnormal tiny circulating currents are extracted from the preprocessed current signal, wherein the extracted key features include the duration, frequency and the growth rate of the change rate of the current in the current signal over time of the irregular fluctuations. After extraction, the duration, frequency and the growth rate of the change rate of the current in the current signal over time of the irregular fluctuations are processed under the set monitoring window to generate an irregular persistence index and a circulating current slope growth index, respectively. The irregular persistence index is used to quantify the cumulative degree of the duration and frequency of the irregular fluctuations in the current signal, and the circulating current slope growth index is used to quantify the cumulative situation of the growth rate of the change rate of the current in the current signal over time.

[0073] After obtaining the irregular persistence index and circulating current slope growth index generated by further processing the current signal, the irregular persistence index and circulating current slope growth index are used as input feature sets, and the irregular persistence index and circulating current slope growth index in the input feature set are input into a pre-trained deep convolutional neural network to generate a circulating current accumulation index. The circulating current accumulation index is used to predict the abnormal small circulating current accumulation phenomenon in the cable sheath circulating current, and the abnormal small circulating current accumulation in the cable sheath circulating current is identified.

[0074] A pre-trained deep convolutional neural network (CNN) is a neural network that has been trained and optimized using a large amount of historical data and manually annotated training data sets before the model is actually applied. It learns complex patterns and relationships of input features through multiple layers of convolution operations, nonlinear activation functions, and pooling layers. During training, the network uses feedforward and backpropagation algorithms to adjust its weights and biases to minimize the loss function, thereby improving the accuracy of the model in identifying and predicting abnormal patterns in input data. Pre-trained deep convolutional neural networks have the ability to recognize specific types of data (such as current signal characteristics) before use, and can provide fast and efficient predictions when new data is input. This pre-training process ensures that the network can adapt to a variety of input features and has strong generalization capabilities, enabling it to handle and identify complex and difficult-to-detect anomalies in real-time monitoring.

[0075] In the cable sheath circulation monitoring application, the pre-trained deep convolutional neural network is fully trained and optimized to recognize and process multi-dimensional features in the input feature set, such as the irregular persistence index and the circulation slope growth index. During the training process, the deep convolutional neural network learns the characteristic patterns in different time series data, such as the subtle changes in the accumulation of abnormal small circulations, and can identify the association between these changes and potential faults. Through multiple layers of convolution and pooling operations, the deep convolutional neural network extracts the local pattern of features and complicates them layer by layer, from simple edge and curve features to higher-order time series correlations. The output layer combines the weight relationship of the input features to generate the circulation accumulation index, which is used to predict the accumulation effect of small circulations that may occur in the cable sheath. The hierarchical structure and deep learning ability of the deep convolutional neural network enable it to process complex time series data and identify subtle and inconspicuous signal features. As a result, the system can timely identify and predict potential dangerous accumulation phenomena in the cable sheath, provide early warnings, and help operation and maintenance personnel take necessary precautions to prevent serious consequences such as cable overheating, sheath damage, and system failure.

[0076] Irregular fluctuations in the current signal of the cable sheath circulating current last for a long time and have a high frequency, which usually indicates the potential accumulation of abnormal micro-circulations. When these features in the current signal remain at a high level for a long time, it may indicate that the local area of ​​the cable sheath is experiencing continuous micro-circulation fluctuations, resulting in heat accumulation and chronic degradation of the insulation material. Although these fluctuations may not be enough to cause an alarm in a short period of time, their cumulative effect will gradually weaken the insulation performance of the sheath, increase the risk of sheath overheating and damage, and may eventually cause sheath failure or cable short circuit. Therefore, monitoring and analyzing these irregular fluctuation characteristics is a key step in early detection of potential problems and prevention of more serious failures.

[0077] The specific steps of processing the duration and frequency of irregular fluctuations in the current signal under the set monitoring window to generate the irregular persistence index are as follows:

[0078] Under the set monitoring window, first calculate the duration of each irregular fluctuation in the current signal, and set the starting time point of the i-th irregular fluctuation in the current signal to be The end time of the i-th irregular fluctuation is Then the irregular fluctuation duration ΔT of the i-th irregular fluctuation is i The calculation expression is: By measuring the duration of irregular fluctuations ΔT i Perform accumulation and calculate the cumulative duration, the cumulative duration T total is the sum of the duration of all irregular fluctuations within the monitoring window, and the calculation expression is: Where n is the total number of irregular fluctuations detected within the monitoring window;

[0079] To identify each irregular fluctuation in the current signal, a variety of existing methods can be used, such as wavelet transform, short-time Fourier transform (STFT) and edge detection algorithms. Wavelet transform can decompose the signal into different frequency components and capture subtle changes in the signal in the time and frequency domains, which is very suitable for detecting local instantaneous changes and irregular fluctuations. Short-time Fourier transform can identify instantaneous abnormal fluctuations in the current signal by dividing the signal into short time windows to analyze its frequency components. In addition, edge detection algorithms based on mathematical derivatives, such as the Sobel operator or the Canny operator, can be used to identify the edges of signal fluctuations and help quickly locate the starting and ending points of the fluctuations. These methods are used in combination to effectively identify and capture each irregular fluctuation in the current signal, ensuring early detection and analysis of abnormal conditions;

[0080] This cumulative duration represents the cumulative degree of irregular fluctuations within the current monitoring window. A higher cumulative duration indicates frequent and continuous irregular fluctuations, which is the basic parameter of the irregular persistence index.

[0081] Calculate the frequency of irregular fluctuations in the current signal through the instantaneous frequency f irreg To represent the number of irregular fluctuations per unit time in the monitoring window, the calculation expression of instantaneous frequency is:

[0082]

[0083] , where T window is the total duration of the monitoring window. In order to consider the impact of frequency on the degree of accumulation, the weighted cumulative frequency F is introduced. weighted , use an attenuation factor α to reduce the weight when the number of fluctuations is small, and the calculation expression of the weighted cumulative frequency is:

[0084]

[0085] , where α is a constant less than 1, controlling the attenuation of the irregular fluctuation frequency, f irregi is the instantaneous frequency of the i-th irregular fluctuation;

[0086] Higher weighted cumulative frequency F weighted It indicates the frequent accumulation of fluctuations within the monitoring window, providing a weighted frequency parameter for the subsequent generation of the irregular persistence index.

[0087] Use cumulative duration T total and weighted cumulative frequency F weighted Generate Irregularity Persistence Index I persist, to comprehensively quantify the duration and frequency of irregular fluctuations, the irregular persistence index I persist The generative expression for is: Wherein, γ is the exponential parameter of the weighted cumulative frequency, which is usually set to a constant greater than 1, preferably 3, to increase the impact of high-frequency fluctuations on the irregular persistence index;

[0088] The irregular persistence index I perdist The higher it is, the more frequent and longer the irregular fluctuations of the current signal in the monitoring window are, indicating that there is a potential risk of accumulation of abnormal tiny circulating currents in the cable sheath.

[0089] It can be seen from the irregular persistence index that the greater the performance value of the irregular persistence index generated after processing the duration and frequency of irregular fluctuations in the current signal under the set monitoring window, the greater the risk of abnormal small circulating current accumulation in the cable sheath circulating current. This is because the irregular persistence index comprehensively quantifies the duration and frequency accumulation of irregular fluctuations in the current signal. A higher irregular persistence index means that the current signal frequently and for a long time shows irregular fluctuations within the monitoring window, indicating that there may be continuous small circulating current anomalies, which in turn increases the risk of heat accumulation and sheath insulation degradation. On the contrary, if the irregular persistence index is small, it means that the irregular fluctuations in the current signal are less or of shorter duration, indicating that the risk of abnormal small circulating current accumulation in the cable sheath circulating current is low and the overall state is relatively stable.

[0090] If the rate of change of the current over time in the current signal of the cable sheath circulating current is high, this indicates that there may be potential accumulation of abnormal micro-circuits in the cable sheath circulating current. A high rate of change means that the fluctuation amplitude of the current is significantly enhanced in a short period of time, reflecting the gradual accumulation of abnormal behavior in the circulating current. This cumulative effect may cause an increase in heat in local areas of the cable sheath and cause aging and deterioration of the insulation material. If not identified and intervened in time, this phenomenon may further develop into serious overheating and damage of the cable sheath, and even lead to major problems such as short circuits that endanger the stability and safety of the cable system. Therefore, monitoring and analyzing the rate of increase of the current over time is particularly important for early warning and maintaining the safe operation of the cable sheath.

[0091] The specific steps of processing the growth rate of the current change over time in the current signal under the set monitoring window to generate the circulating current slope growth index are as follows:

[0092] After the current signal is preprocessed, the preprocessed current signal is represented by I(t) to extract the instantaneous rate of change of the current over time. In order to capture the severity of the current fluctuation, the growth rate of the current change rate within the monitoring time window is calculated. The calculation expression is:

[0093]

[0094] , where I(t) is the current value at time t, is the instantaneous rate of change of current with time (T window is the total duration of the monitoring window), used to compare changes between time points, R(t) is the growth rate of the instantaneous change rate, and quantifies the relative change of current fluctuations;

[0095] This step provides basic data for the subsequent generation of accumulation and circulating current slope growth index by extracting the growth amplitude of the current change rate, ensuring that potential signs of abnormal current accumulation can be identified.

[0096] In the monitoring window, the growth rate of the instantaneous change rate at each time point is calculated and accumulated to form a cumulative change rate function. The cumulative change rate function is defined as: Where C(t) is the cumulative change rate function, which represents the total accumulation of the change rate within the monitoring window, α j is the weight coefficient, which reflects the relative contribution of a time point to the accumulation. It can be a linear or exponential function. j represents the time point, N is the total number of sampling points in the monitoring time window, and R(t j ) is t j The growth rate of the instantaneous rate of change at the moment;

[0097] This step provides quantification of the overall current fluctuation trend by accumulating the growth rate of the instantaneous change rate at each time point.

[0098] The circulating current slope growth index is generated by using the cumulative change rate function C(t) to quantify the cumulative growth of the current change rate. The generation expression of the circulating current slope growth index is:

[0099]

[0100] , where Slope circu is the circulation slope growth index, which is used to quantify the accumulation of the growth amplitude of the current over time. β is the adjustment parameter, which controls the weight of the cumulative change rate function C(t) in the calculation. x The impact index indicates the degree of response to the maximum instantaneous rate of change growth. max(R(t)) is the maximum value of the instantaneous rate of change growth within the monitoring window. ∈ is a small constant to avoid a denominator of zero, usually taking a minimum value such as 0.0001.

[0101] This step converts the cumulative change rate function into a circulating current slope growth index to quantify and assess the cumulative current fluctuations, helping to identify potential abnormal accumulation risks in the cable sheath.

[0102] It can be seen from the irregular persistence index that the larger the value of the circulating current slope growth index generated by processing the growth rate of the current change rate over time in the current signal under the set monitoring window, the higher the risk of abnormal small circulating current accumulation in the cable sheath. This is because a higher circulating current slope growth index indicates that the rate of change of the current has increased significantly in a short period of time, indicating a continuous current fluctuation and cumulative effect, which may lead to local heat accumulation and ultimately damage the insulation layer, increasing the risk of cable sheath failure. On the contrary, if the value of the circulating current slope growth index is low, it indicates that the rate of change of the current over time is stable and the cumulative effect is small, which means that the risk of abnormal small circulating current accumulation in the cable sheath is low, so that the system operation is safer and more stable.

[0103] The above-mentioned deep convolutional neural network is not specifically limited here, and can realize the irregular persistence index I persist and the circulation slope growth index Slope circu Comprehensive analysis is performed to generate the circulation accumulation index Cirl a In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; the circulation accumulation index Circul accumu The resulting calculation formula is:

[0104]

[0105] , where l 1 , l 2 Irregularity persistence index I persist and circulation slope growth index Slope circu The preset proportionality factor, and l 1 , l 2 Both are greater than 0.

[0106] It can be seen from the circulating current accumulation index that the greater the performance value of the irregular persistence index generated after processing the duration and frequency of irregular fluctuations in the current signal under the set monitoring window, the greater the performance value of the circulating current slope growth index generated after processing the growth rate of the change rate of the current in the current signal over time under the set monitoring window. In other words, the greater the performance value of the circulating current accumulation index generated when the cable sheath circulating current is predicted by the deep convolutional neural network, the higher the risk of abnormal small circulating current accumulation in the cable sheath, and vice versa.

[0107] The circulation accumulation index generated when predicting the cable sheath circulation through the deep convolutional neural network is compared and analyzed with the preset circulation accumulation index reference threshold to identify the abnormal micro-circulation accumulation in the cable sheath circulation. The specific identification steps are as follows:

[0108] If the circulating current accumulation index is greater than or equal to the circulating current accumulation index reference threshold, an abnormal small circulating current accumulation signal is generated, indicating that there is abnormal small circulating current accumulation in the cable sheath circulating current, indicating that significant circulating current accumulation has been detected in the cable sheath, indicating that current fluctuations may have been gradually accumulating and reached a state that requires attention;

[0109] If the circulating current accumulation index is less than the circulating current accumulation index reference threshold, a current normal operation signal is generated, indicating that the current fluctuations and changes in the cable sheath are still within the normal range, no significant abnormal accumulation phenomenon is detected, and the cable operation status is considered to be stable and safe.

[0110] When abnormal micro-circulation is identified in the cable sheath circulation, data is continuously collected under the set monitoring window, and multiple sets of circulation change information are recorded. Cluster analysis algorithms (such as K-means or hierarchical clustering) are used to perform pattern analysis on the data, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type.

[0111] When there is abnormal accumulation of small circulations in the cable sheath circulation, a cluster analysis algorithm is used to analyze the data pattern and further divide the abnormal accumulation of small circulations. The specific steps are as follows:

[0112] Under the set monitoring window, the circulation accumulation index generated in several monitoring windows is continuously collected to establish a data set F, where F = {Cirl ac (t u )}={Cirl ac (t 1 ), Cirl ac (t 2 ), ..., Cirl ac (t y )}, the circulation accumulation index in the data set reflects the accumulation trend and change of the circulation in the cable sheath over a period of time. ac (t u ) is the circulation accumulation index generated in the u-th monitoring window, and y is the total number of monitoring windows collected;

[0113] This step builds a data set by collecting circulation accumulation indices within multiple monitoring windows for subsequent pattern recognition and analysis.

[0114] Extract the feature vector F from the circulation accumulation index within each monitoring window x , used for cluster analysis, eigenvector F x Defined as:

[0115]

[0116] , where ΔCirl ac (t u ) is the increment of the cumulative index of circulation between two adjacent monitoring windows, indicating the magnitude of the change. It represents the cumulative value of the circulation accumulation index in two adjacent monitoring windows, which is used to evaluate the overall trend. It represents the instantaneous rate of change of the circulation accumulation index, indicating the rate of change;

[0117] The generated feature vector F x Input into the clustering analysis algorithm, such as K-means or density-based clustering algorithm (DBSCAN), the expression defining the clustering objective function J is:

[0118]

[0119] , where K is the number of categories after clustering, C k represents the kth cluster, μ k is the cluster center of the kth cluster, which is used to measure the distance between the feature vector and the center point;

[0120] Cluster analysis algorithms are used to identify and classify the patterns of circulation accumulation indices into different accumulation types, such as accumulated occurrence type, dispersed occurrence type, and occasional occurrence type.

[0121] Each cluster center μ k and the preset cluster center reference threshold T agg Compare and determine the type of abnormal micro-circulation accumulation in the cable sheath circulation. The specific classification rules are as follows:

[0122] If μ k ≥T agg , then the cluster is classified as accumulation type, indicating that the circulation accumulation index in this category is continuously high, indicating the risk of long-term accumulation;

[0123] If 0.5T agg ≤μ j <T agg , then the cluster is classified as a dispersed appearance type, indicating that circulation fluctuations exist but are not significant enough;

[0124] If μ k <0.5T agg, then the cluster is classified as an occasional occurrence type, indicating that the circulation accumulation index is low and the anomaly is short-lived and not sustained.

[0125] Function: This step clearly divides the types of circulation anomalies into different categories by comparing the cluster center with the threshold, which facilitates further risk assessment and response strategies.

[0126] Classification helps to better understand the nature of the anomaly and its possible development trend, thereby making more precise response decisions.

[0127] For abnormal accumulation of small circulations identified as "accumulation patterns" in cluster analysis, an alarm will be issued immediately to notify relevant personnel to handle the situation;

[0128] In cluster analysis, if a certain type of abnormal micro-circulation accumulation is identified as an "accumulation type", it means that this type of abnormal accumulation has the characteristics of persistence and gradual increase, which may cause heat accumulation in the cable sheath and damage to the insulation material, leading to serious cable failure. In order to prevent potential risks, the system will immediately issue an alarm to notify relevant operation and maintenance personnel to take necessary measures, such as checking or repairing the cable, to prevent the fault from further deteriorating and ensure the safe and stable operation of the cable system.

[0129] For abnormal accumulation of small circulations identified as "dispersed occurrence type" in cluster analysis, which means that the abnormal circulation is intermittent and irregularly distributed, no alarm will be issued immediately, but continuous monitoring will be maintained to observe the subsequent circulation change trend. If the dispersed anomaly evolves into an accumulation type in subsequent monitoring, an alarm will be triggered immediately to prevent the accumulation of abnormal small circulations from developing into a more serious problem.

[0130] In cluster analysis, if a certain type of abnormal micro-circulation accumulation is identified as a "dispersed appearance type", this means that the circulation anomaly is intermittent and irregularly distributed, and will not immediately cause serious problems. Therefore, the system will not issue an alarm immediately, but will continue to monitor the anomaly and closely observe its subsequent change trends. If subsequent monitoring finds that the dispersed anomaly gradually evolves into a persistent, cumulative "accumulation type" anomaly, the system will immediately trigger an alarm and notify relevant personnel to take action to prevent the accumulation of circulation anomalies from further deteriorating and developing into serious cable failures.

[0131] For the accumulation of abnormal micro-circulations identified as "occasionally occurring" in cluster analysis, the micro-circulation anomalies that occur occasionally are usually short-term and instantaneous, and will not cause long-term impact on the cable sheath. After such anomalies are identified during the monitoring process, the system will automatically record them but will not trigger an early warning;

[0132] In cluster analysis, if a certain type of abnormal micro-circulation accumulation is identified as "occasional type", it means that this abnormality is short-term and instantaneous, and will not have a long-term or serious impact on the cable sheath. Therefore, although the system will identify this type of abnormality during the monitoring process and record it for subsequent analysis, it will not trigger an alarm or notify the operation and maintenance personnel. This processing method helps to avoid unnecessary alarms due to short-term and harmless abnormalities, reduce the burden of operation and maintenance, and maintain monitoring and recording of the overall operating status.

[0133] The present invention utilizes high-precision sensors for continuous monitoring, and combines deep convolutional neural network analysis and clustering algorithms to identify and classify tiny circulation anomalies in the cable sheath, and accurately classify tiny circulation anomalies in the cable sheath that may cause heat accumulation. By subdividing the abnormal conditions into "accumulated type", "dispersed type" and "occasional type", timely alarms are issued when significant cumulative risks occur, and operation and maintenance personnel are notified to take measures; and for intermittent and occasional anomalies, selective monitoring and recording are carried out to avoid frequent false alarms, which significantly improves the safety and maintenance efficiency of the cable sheath, and reduces the risks of major failures such as cable overheating, insulation degradation, and even short circuits caused by the accumulation of tiny anomalies that are not handled in time, thereby ensuring the stable operation of the cable system and the reliability of power supply.

[0134] The present invention provides Figure 2 An online cable sheath circulating current monitoring system shown includes a data acquisition and preprocessing module, a feature extraction and feature set construction module, a deep learning analysis and anomaly recognition module, a cluster analysis and anomaly classification module, an accumulation type anomaly alarm module, a decentralized type anomaly monitoring module and an occasional type anomaly recording module;

[0135] The data acquisition and preprocessing module installs high-precision sensors on the outside of the cable sheath to continuously monitor the current signal of the sheath circulation and perform noise reduction and cleaning on the collected data;

[0136] The feature extraction and feature set construction module extracts key features reflecting abnormal accumulation of small circulating currents from the preprocessed current signal, further processes the extracted features under the set monitoring window, and uses the processed features as the input feature set;

[0137] The deep learning analysis and anomaly recognition module inputs the features in the input feature set into a pre-trained deep convolutional neural network for analysis. Through the prediction results of the deep convolutional neural network, it identifies the abnormal accumulation of small circulations in the cable sheath circulation.

[0138] The cluster analysis and abnormal classification module, when identifying the presence of abnormal micro-circulation in the cable sheath circulation, continuously collects data under the set monitoring window and records multiple sets of circulation change information. The cluster analysis algorithm is used to perform pattern analysis on the data, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type.

[0139] The accumulation type abnormal alarm module immediately issues an alarm for abnormal micro-circulation accumulation identified as "accumulation type" in cluster analysis, and notifies relevant personnel to handle it;

[0140] The dispersed anomaly monitoring module does not immediately issue an alarm for abnormal micro-circulation accumulation identified as "dispersed appearance type" in cluster analysis, but maintains continuous monitoring to observe the subsequent circulation change trend. If the dispersed anomaly evolves into an accumulation type in subsequent monitoring, the alarm is immediately triggered to prevent the abnormal micro-circulation accumulation from developing into a more serious problem;

[0141] The accidental anomaly recording module automatically records the accumulation of abnormal micro-circulations identified as "occasional" in cluster analysis during the monitoring process, but does not trigger an early warning.

[0142] An online cable sheath circulating current monitoring method provided by an embodiment of the present invention is implemented by the above-mentioned online cable sheath circulating current monitoring system. The specific method and process of an online cable sheath circulating current monitoring system are detailed in the embodiment of the above-mentioned online cable sheath circulating current monitoring method, which will not be repeated here.

[0143] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0144] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0145] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An online cable sheath circulating current monitoring method, characterized in that: The following steps are involved: Install high-precision sensors on the outside of the cable sheath to continuously monitor the current signal of the sheath circulation and perform noise reduction and cleaning on the collected data; Extract key features reflecting abnormal accumulation of tiny circulating currents from the preprocessed current signal, further process the extracted features under a set monitoring window, and use the processed features as input feature sets; The features in the input feature set are input into a pre-trained deep convolutional neural network for analysis. The abnormal accumulation of small circulations in the cable sheath circulation is identified through the prediction results of the deep convolutional neural network. When abnormal micro-circulation is identified in the cable sheath circulation, data is continuously collected under the set monitoring window, and multiple sets of circulation change information are recorded. The data are analyzed using a cluster analysis algorithm, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type. For abnormal accumulation of small circulations identified as "accumulation pattern" in cluster analysis, an alarm will be issued immediately to notify relevant personnel to handle the situation; For abnormal accumulation of small circulations identified as "dispersed appearance" in cluster analysis, do not immediately issue an alarm, but maintain continuous monitoring to observe the subsequent circulation change trend. If the dispersed anomaly evolves into an accumulation appearance in subsequent monitoring, immediately trigger an alarm to prevent the accumulation of abnormal small circulations from developing into a more serious problem; For the abnormal accumulation of small circulations identified as "occasional type" in cluster analysis, they are automatically recorded but no warning is triggered after being identified during the monitoring process.

2. The method for monitoring the circulating current of the online cable sheath according to claim 1, characterized in that: The high-precision sensors installed on the outside of the cable sheath are set at the key monitoring positions of the sheath to ensure that comprehensive and accurate current signals can be captured; the sensors are installed close to the sheath surface to reduce signal attenuation, and use high-temperature resistant, waterproof and dust-proof packaging materials to adapt to harsh environments; the sensors are arranged evenly at specified intervals, with emphasis on cable joints and bends.

3. The method for online cable sheath circulation monitoring according to claim 1, characterized in that: The key features reflecting the accumulation of abnormal tiny circulating currents are extracted from the preprocessed current signal, wherein the extracted key features include the duration, frequency and the growth rate of the change rate of the current in the current signal over time of the irregular fluctuations. After extraction, the duration, frequency and the growth rate of the change rate of the current in the current signal over time of the irregular fluctuations are processed under the set monitoring window to generate an irregular persistence index and a circulating current slope growth index, respectively. The irregular persistence index is used to quantify the cumulative degree of the duration and frequency of the irregular fluctuations in the current signal, and the circulating current slope growth index is used to quantify the cumulative situation of the growth rate of the change rate of the current in the current signal over time.

4. The method for monitoring the circulating current of the online cable sheath according to claim 3 is characterized in that: After obtaining the irregular persistence index and circulating current slope growth index generated by further processing the current signal, the irregular persistence index and circulating current slope growth index are used as input feature sets, and the irregular persistence index and circulating current slope growth index in the input feature set are input into a pre-trained deep convolutional neural network to generate a circulating current accumulation index. The circulating current accumulation index is used to predict the abnormal small circulating current accumulation phenomenon in the cable sheath circulating current, and the abnormal small circulating current accumulation in the cable sheath circulating current is identified.

5. The method for monitoring the online cable sheath circulation according to claim 4, characterized in that: The circulation accumulation index generated when predicting the cable sheath circulation through the deep convolutional neural network is compared and analyzed with the preset circulation accumulation index reference threshold to identify the abnormal micro-circulation accumulation in the cable sheath circulation. The specific identification steps are as follows: If the circulation accumulation index is greater than or equal to the circulation accumulation index reference threshold, an abnormal micro-circulation accumulation signal is generated, indicating that abnormal micro-circulation accumulation exists in the cable sheath circulation; If the circulating current accumulation index is less than the circulating current accumulation index reference threshold, a current normal operation signal is generated, indicating that the current fluctuation and change in the cable sheath are still within the normal range and no abnormal accumulation phenomenon is detected.

6. The method for monitoring the online cable sheath circulation according to claim 3, characterized in that: The specific steps of processing the duration and frequency of irregular fluctuations in the current signal under the set monitoring window to generate the irregular persistence index are as follows: Under the set monitoring window, first calculate the duration of each irregular fluctuation in the current signal, and set the starting time point of the i-th irregular fluctuation in the current signal to be The end time of the i-th irregular fluctuation is Then the irregular fluctuation duration ΔT of the i-th irregular fluctuation is i The calculation expression is: By measuring the duration of irregular fluctuations ΔT i Perform accumulation and calculate the cumulative duration, the cumulative duration T total is the sum of the duration of all irregular fluctuations within the monitoring window, and the calculation expression is: Where n is the total number of irregular fluctuations detected within the monitoring window; Calculate the frequency of irregular fluctuations in the current signal through the instantaneous frequency f irreg To represent the number of irregular fluctuations per unit time in the monitoring window, the calculation expression of instantaneous frequency is: ， Among them, T window is the total duration of the monitoring window. In order to consider the impact of frequency on the degree of accumulation, the weighted cumulative frequency F is introduced. weighted , use the attenuation factor α to reduce the weight when the number of fluctuations is small, and the calculation expression of the weighted cumulative frequency is: ， Among them, α is a constant less than 1, which controls the attenuation degree of irregular fluctuation frequency. is the instantaneous frequency of the i-th irregular fluctuation; Use cumulative duration T total and weighted cumulative frequency F weighted Generate Irregularity Persistence Index I persist , to comprehensively quantify the duration and frequency of irregular fluctuations, the irregular persistence index I persist The generative expression for is: Among them, γ is the exponential parameter of the weighted cumulative frequency to increase the impact of high-frequency fluctuations on the irregular persistence index.

7. The method for monitoring the circulating current of the online cable sheath according to claim 6, characterized in that: The specific steps of processing the growth rate of the current change over time in the current signal under the set monitoring window to generate the circulating current slope growth index are as follows: After the current signal is preprocessed, the preprocessed current signal is represented by I(t) to extract the instantaneous rate of change of the current over time. In order to capture the severity of the current fluctuation, the growth rate of the current change rate within the monitoring time window is calculated. The calculation expression is: ， Where I(t) is the current value at time t, is the instantaneous rate of change of current over time, used to compare changes between time points, R(t) is the growth rate of the instantaneous rate of change, quantifying the relative change of current fluctuations; In the monitoring window, the growth rate of the instantaneous change rate at each time point is calculated and accumulated to form a cumulative change rate function. The cumulative change rate function is defined as: Where C(t) is the cumulative change rate function, which represents the total accumulation of the change rate within the monitoring window, α j is the weight coefficient, reflecting the contribution of a time point to the accumulation, j represents the time point, N is the total number of sampling points in the monitoring time window, R(t j ) is t j The growth rate of the instantaneous rate of change at the moment; The circulating current slope growth index is generated by using the cumulative change rate function C(t) to quantify the cumulative growth of the current change rate. The generation expression of the circulating current slope growth index is: ， Where Slope circu is the circulation slope growth index, which is used to quantify the accumulation of the growth amplitude of the current over time. β is the adjustment parameter, which controls the weight of the cumulative change rate function C(t) in the calculation. x The impact index indicates the degree of response to the maximum instantaneous change rate growth amplitude. max(R(t)) is the maximum value of the instantaneous change rate growth amplitude within the monitoring window, and ∈ is a small constant to avoid the denominator being zero.

8. The method for monitoring the circulating current of the online cable sheath according to claim 4, characterized in that: When there is abnormal accumulation of small circulations in the cable sheath circulation, a cluster analysis algorithm is used to analyze the data pattern and further divide the abnormal accumulation of small circulations. The specific steps are as follows: Under the set monitoring window, the circulation accumulation index generated in several monitoring windows is continuously collected to establish a data set F, where F = {Cirl ac (t u )}={Cirl ac (t1), Cirl ac (t2), ..., Cirl ac (t y )}, the circulation accumulation index in the data set reflects the accumulation trend and change of the circulation in the cable sheath over a period of time. ac (t u ) is the circulation accumulation index generated in the u-th monitoring window, and y is the total number of monitoring windows collected; Extract the feature vector F from the circulation accumulation index within each monitoring window x , used for cluster analysis, eigenvector F x Defined as: ， Where ΔCirl ac (t u ) is the increment of the cumulative index of circulation between two adjacent monitoring windows, indicating the magnitude of the change. It represents the cumulative value of the circulation accumulation index in two adjacent monitoring windows, which is used to evaluate the overall trend. It represents the instantaneous rate of change of the circulation accumulation index, indicating the rate of change; The generated feature vector F x Input into the clustering analysis algorithm, and define the expression of clustering objective function J as: ， In the formula, K is the number of categories after clustering, C k represents the kth cluster, μ k is the cluster center of the kth cluster, which is used to measure the distance between the feature vector and the center point.

9. The method for monitoring the circulating current of the online cable sheath according to claim 8, characterized in that: Each cluster center μ k and the preset cluster center reference threshold T agg Compare to determine the type of abnormal micro-circulation accumulation in the cable sheath circulation. The specific classification rules are as follows: If μ k ≥T agg , then the cluster is divided into the accumulation type; If 0.5T agg ≤μ k <T agg , then the cluster is classified as a dispersed appearance type; If μ k <0.5T agg , then the cluster is classified as accidental occurrence type.

10. An online cable sheath circulation monitoring system, used to implement the online cable sheath circulation monitoring method according to any one of claims 1 to 9, characterized in that: It includes data collection and preprocessing module, feature extraction and feature set construction module, deep learning analysis and anomaly recognition module, cluster analysis and anomaly classification module, accumulation anomaly alarm module, decentralized anomaly monitoring module and accidental anomaly recording module; The data acquisition and preprocessing module installs high-precision sensors on the outside of the cable sheath to continuously monitor the current signal of the sheath circulation and perform noise reduction and cleaning on the collected data; The feature extraction and feature set construction module extracts key features reflecting abnormal accumulation of small circulating currents from the preprocessed current signal, further processes the extracted features under the set monitoring window, and uses the processed features as the input feature set; The deep learning analysis and anomaly recognition module inputs the features in the input feature set into a pre-trained deep convolutional neural network for analysis. Through the prediction results of the deep convolutional neural network, it identifies the abnormal accumulation of small circulations in the cable sheath circulation. The cluster analysis and abnormal classification module, when identifying the presence of abnormal micro-circulation in the cable sheath circulation, continuously collects data under the set monitoring window and records multiple sets of circulation change information. The cluster analysis algorithm is used to perform pattern analysis on the data, and the accumulation of abnormal micro-circulation is further divided into three categories: accumulation type, dispersed type and occasional type. The accumulation type abnormal alarm module immediately issues an alarm for abnormal micro-circulation accumulation identified as "accumulation type" in cluster analysis, and notifies relevant personnel to handle it; The distributed anomaly monitoring module does not immediately issue an alarm for abnormal micro-circulation accumulation identified as "dispersed appearance type" in cluster analysis, but maintains continuous monitoring to observe the subsequent circulation change trend. If the distributed anomaly evolves into an accumulation type in subsequent monitoring, the alarm is immediately triggered to prevent the abnormal micro-circulation accumulation from developing into a more serious problem; The accidental anomaly recording module automatically records the accumulation of abnormal micro-circulations identified as "occasional" in cluster analysis during the monitoring process, but does not trigger an early warning.

Images