Abnormal Detection Method, Device, Electronic Device and Storage Medium for Cable Grounding Circulating Current

Abstract

The present invention discloses an abnormal detection method, device, electronic device and storage medium for cable grounding loop current. The method includes: collecting in real time the grounding loop current data of any grounding node of the cable to be detected, and selecting the amplitudes of k consecutive four types of currents within the acquisition period T from the grounding loop current data of the grounding node, and obtaining k groups of labels and samples of the four types of currents through data processing; judging whether there is a detection model for the grounding node according to the grounding node model library; counting and calculating the abnormal proportion of the amplitude of each type of current among the k groups of amplitudes of the four types of currents of the grounding node, and judging whether the abnormal proportion of the amplitude of each type of current of the grounding node exceeds the alarm threshold. In addition, a detection device, an electronic device and a storage medium are also included. Therefore, the present invention can not only timely detect abnormal grounding nodes and fault causes, but also assist in diagnosing the health status of the cable, realizing the intelligentization and automation of cable operation and maintenance, and ensuring the safety of the cable and the electricity-using unit or individual.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-voltage cable fault diagnosis, and specifically, relates to a method, device, electronic device and storage medium for detecting anomalies in cable grounding circulating current. Background Art

[0002] Currently, electric power is generally transmitted and distributed by means of high-voltage power transmission. High-voltage cables are usually buried deep underground or in some special pipelines. To ensure the continuous and stable operation of the power grid, it is necessary to arrange personnel for manual inspection. However, this results in a large labor cost. Therefore, there is a technical solution of collecting data and then performing modeling analysis to determine whether there are problems with the line. This technical solution generally grounds the line and divides the line into multiple segments, each segment forming a loop. Then, some sampling devices are deployed in these loops, and these devices collect the current values in the grounded line at regular intervals. Finally, by detecting and judging information such as the amplitude and phase of the collected current, it is determined whether there is a fault in this line.

[0003] In the prior art, a Chinese patent document CN106814243A discloses an on-line monitoring device for the grounding circulating current of urban cable lines, which mainly solves the problem by judging whether the detected current amplitude exceeds a threshold. Since this method is too simple and crude, although it can detect obvious problems, it cannot detect potential anomalies. Another Chinese patent document CN112526286A discloses a method for detecting cable grounding circulating current faults based on statistics, which mainly collects a large amount of grounding circulating current data of directly grounded and cross-connected nodes of the cable, finds out the statistical laws of each node, and analyzes whether the grounding circulating current of each node is abnormal through the time distribution law of the circulating current data of a single node and the time distribution law of the circulating current data between multiple nodes, and then infers whether the cable has a fault. However, those of ordinary skill in the art know according to the law of large numbers that statistics requires a large amount of data collection. Otherwise, when the sample data is small, the data is not credible. Therefore, for a time span of a week or a month, a lot of data for many weeks or months needs to be statistically analyzed to achieve good results. For a too small time span, there will be other environmental variables (such as: line materials, humidity of the environment where it is located, etc.) such that different nodes and different loops have different statistical characteristic values, resulting in misjudgment when comparing or it is not easy to distinguish who has a real anomaly. Summary of the Invention

[0004] The present invention aims to provide a method, device and storage medium for detecting anomalies in cable grounding circulating current to solve the technical problem of insufficient accuracy of anomaly detection caused by too small time span and environmental factors.

[0005] To solve the above technical problems, the present invention provides a method for detecting abnormal cable grounding circulating current, which includes the following steps:

[0006] Step S1: Real-time collect the grounding circulating current data of any grounding node of the cable to be detected, and select the amplitudes of k consecutive A-phase, B-phase, C-phase, and total current in the grounding circulating current data of this grounding node within the acquisition period T. After data processing, obtain k groups of labels and samples of the four types of current, namely A-phase, B-phase, C-phase, and total current;

[0007] Step S2: According to the grounding node model library, judge whether there is a detection model for this grounding node. If not, start to construct the detection model of this grounding node. If so, determine whether there is an abnormality in the amplitudes of the k groups of A-phase, B-phase, C-phase, and total current of this grounding node according to the detection model;

[0008] Step S3: Statistically calculate the abnormal proportion in the amplitudes of the k groups of A-phase, B-phase, C-phase, and total current of this grounding node, and judge whether the abnormal proportion of this grounding node exceeds the alarm threshold. If so, issue an alarm and notify the operation and maintenance personnel that the current of this grounding node is abnormal. Otherwise, end this abnormal detection.

[0009] Preferably, it further includes: Step S4: Continuously statistically calculate the average abnormal proportion of the t times after the amplitude of this type of current corresponding to the alarm triggered by this grounding node, and judge whether the average abnormal proportion of the amplitude of this type of current exceeds the alarm threshold. If so, issue an alarm and notify the operation and maintenance personnel that the current of this grounding node in this direction is abnormal. If not, do not issue an alarm.

[0010] Preferably, the specific steps of "obtaining k groups of labels and samples of the four types of current, namely A-phase, B-phase, C-phase, and total current" in Step 1 are: Take the i-th current amplitude of each type in the obtained k consecutive A-phase, B-phase, C-phase, and total current amplitudes as the label, and then take the first m amplitudes and the last n amplitudes of the i-th current amplitude of this type as the sample, so as to obtain k groups of labels and samples of the four types of current, namely A-phase, B-phase, C-phase, and total current.

[0011] Further preferably, the specific steps of "if not, start to construct the detection model of this grounding node" in Step 2 are:

[0012] Form the training data set of this grounding node with the k groups of labels and samples of the four types of current, namely A-phase, B-phase, C-phase, and total current, and pour the training data set into the neural network for model training. After the model training is completed, the detection model of this grounding node is formed, and at the same time, the detection model of this grounding node is incorporated into the grounding node model library.

[0013] Further preferably, the specific steps of "if so, determine whether there are abnormalities in the k sets of current amplitudes of the A-phase, B-phase, C-phase, and total current of the grounding node according to the detection model" in step 2 are as follows:

[0014] Input the samples corresponding to the k sets of current amplitudes of the A-phase, B-phase, C-phase, and total current into the detection model of the grounding point to obtain the model outputs of the current amplitudes of the A-phase, B-phase, C-phase, and total current for each group;

[0015] Calculate the ratio of the number of XOR values between the labels and model outputs of the current amplitudes of the A-phase current, B-phase current, C-phase current, and total current for each group to the number of corresponding current amplitude samples, and determine whether this ratio is greater than the discrimination threshold. When this ratio is greater than the discrimination threshold, record the current amplitudes of the A-phase, B-phase, C-phase, and total current of this group as abnormal values; otherwise, record the current amplitudes of the A-phase, B-phase, C-phase, and total current of this group as normal values.

[0016] Even further preferably, the numerical ranges of the discrimination threshold, warning threshold, or alarm threshold are all between 0 and 1 respectively.

[0017] To solve the above technical problems, the present invention also provides a cable grounding loop current abnormality detection device, wherein the detection device includes:

[0018] A data acquisition unit, configured to collect the grounding loop current data of any grounding node of the cable to be detected in real time, and obtain the amplitudes of the continuous k A-phase, B-phase, C-phase, and total current of the four types of currents from the collected grounding loop current data within the acquisition period T of this grounding node, and obtain the labels and samples of the k sets of current amplitudes of the A-phase, B-phase, C-phase, and total current through data processing;

[0019] A model judgment unit, configured to judge whether there is a detection model for this grounding node according to the grounding node model library. If not, start to construct the detection model of this grounding node. If there is, determine whether there are abnormalities in the current amplitudes of the A-phase, B-phase, C-phase, and total current of each group of this grounding node according to the detection model;

[0020] A first warning unit, configured to count and calculate the abnormal proportion in the k sets of current amplitudes of the A-phase, B-phase, C-phase, and total current of this grounding node, judge whether the abnormal proportion of this grounding node exceeds the warning threshold. If so, issue a warning and notify the operation and maintenance personnel that the current of this grounding node is abnormal. Otherwise, end the current abnormality detection process.

[0021] Preferably, it further includes a second warning unit, configured to count and calculate the abnormal proportion of the x current amplitudes of any type of current amplitude of this grounding node again, and judge whether the average abnormal proportion of this type of current amplitude exceeds the alarm threshold. If so, issue an alarm and notify the operation and maintenance personnel that the current of this phase of this grounding node is abnormal. Otherwise, no alarm is issued.

[0022] Preferably, the data acquisition unit includes a data processing module. The data processing module is configured to take the i-th current amplitude of each of the four types of current amplitudes of the A-phase, B-phase, C-phase, and total current in the continuously acquired k current amplitudes as labels, and then take the first m amplitudes and the last n amplitudes of the i-th current amplitude of this type as input values, so as to obtain the labels and samples of the four types of current of the A-phase, B-phase, C-phase, and total current in k groups.

[0023] More preferably, the model judgment unit includes a model construction module. The model construction module is configured to form a training data set of this grounding node with the labels and samples of the four types of current of the A-phase, B-phase, C-phase, and total current in k groups, and pour the training data set into a neural network for model training. After the model training is completed, a detection model of this grounding node is formed, and at the same time, the detection model of this grounding node is incorporated into the grounding node model library.

[0024] More preferably, the model judgment unit further includes an abnormality judgment module. The abnormality judgment module is configured to calculate the ratio of the number of XOR values between the label of each group of A-phase current amplitude, B-phase current amplitude, C-phase current amplitude, and total current amplitude and the model output to the number of corresponding current amplitude samples, and judge whether this ratio is greater than a discrimination threshold. When this ratio is greater than the discrimination threshold, the amplitudes of the four types of current of the A-phase, B-phase, C-phase, and total current in this group are recorded as abnormal values; otherwise, the amplitudes of the four types of current of the A-phase, B-phase, C-phase, and total current in this group are recorded as normal values.

[0025] To solve the above technical problems, the present invention further provides an electronic device, which includes:

[0026] A memory for storing a computer program;

[0027] A processor for implementing the cable grounding loop current abnormality detection method as described in any one of the above when executing the computer program.

[0028] To solve the above technical problems, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, the cable grounding loop current abnormality detection method as described in any one of the above is implemented.

[0029] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are:

[0030] 1) The present invention does not need to collect data of sufficient length (sufficient length of data specifically refers to units such as weeks and months), and individually models each grounding node to prevent the influence of other environmental variables on abnormality detection;

[0031] 2) The neural network model based on deep learning of the present invention models the current amplitude of the grounding node, enabling the model to remember or understand the timing of the current amplitude of the grounding node it is responsible for in the time domain;

[0032] 3) The detection model of the present invention can predict the intermediate target amplitude based on the amplitudes before and after, and then compare the predicted intermediate target amplitude with the actual real amplitude. If it is within a certain error range, it can be determined whether the real amplitude is abnormal;

[0033] 4) When the number of abnormal amplitudes in several directions of current amplitude detected at one time is large in the present invention, it can be determined which directions of current of the grounding node are abnormal. Description of the Drawings

[0034] Figure 1 It is a flowchart of the cable grounding loop current anomaly detection method according to Embodiment 1 of the present invention.

[0035] Figure 2 It is a flowchart of the specific method for constructing the detection model of the grounding node in step S2 in Embodiment 1 of the present invention.

[0036] Figure 3 It is a flowchart of the cable grounding loop current anomaly detection method according to Embodiment 2 of the present invention.

[0037] Figure 4 It is a framework diagram of the cable grounding loop current anomaly detection device according to Embodiment 3 of the present invention.

[0038] Figure 5 It is a framework diagram of the cable grounding loop current anomaly detection device according to Embodiment 4 of the present invention. Detailed Embodiments

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Therefore, all other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Embodiment 1

[0041] As Figure 1 shown, the embodiment of the present invention provides a cable grounding loop current anomaly detection method, which includes the following steps:

[0042] Step S1: Continuously collect the ground loop current data of any grounding node of the cable to be inspected in real time, and select k consecutive amplitude values of the four types of currents, namely phase A, phase B, phase C, and total current, within the acquisition period T (in each acquisition period, the number of current sampling values of each type is equal and greater than or equal to k) from the ground loop current data of this grounding node. After data processing, obtain k groups of labels and samples of the four types of currents, namely phase A, phase B, phase C, and total current.

[0043] Step S2: Determine whether there is a detection model for this grounding node according to the grounding node model library. If not, start constructing the detection model for this grounding node. If so, determine whether there are abnormalities in the k groups of amplitude values of the four types of currents, namely phase A, phase B, phase C, and total current, of this grounding node according to the detection model.

[0044] Step S3: Statistically calculate the abnormal proportion of the amplitude value of each type of current among the k groups of amplitude values of the four types of currents, namely phase A, phase B, phase C, and total current, of this grounding node, and determine whether the abnormal proportion of the amplitude value of each type of current of this grounding node exceeds the alarm threshold. If so, trigger an alarm and notify the operation and maintenance personnel that there is an abnormality in the current of this grounding node. Otherwise, no alarm is issued.

[0045] In the present invention, the grounding node generally collects the ground loop current data through a deployed ground loop current detection system, that is, the amplitude values of the four types of currents, namely phase A, phase B, phase C, and total current, are derived from this ground loop current detection system. The ground loop current detection system is a third-party system, mainly used to detect the cable voltage and current at the cable grounding point; since this system is not the content protected by the present invention and is generally known to those skilled in the art, it will not be elaborated here. The abnormal proportion of the grounding node is the ratio of the number of XOR values between the labels of the amplitude values of phase A current, phase B current, phase C current, and total current and the model output to the number of sampling values in the corresponding current amplitude sample. The grounding node model library is a collection in the system that records the detection models trained for each grounding node, and each grounding node corresponds to a corresponding detection model one by one.

[0046] In the embodiment of the present invention, the specific steps of "obtaining k groups of labels and samples of the four types of currents, namely phase A, phase B, phase C, and total current, through data processing" in step 1 are as follows: Take the i-th current amplitude value of each type as the label value from the obtained k consecutive amplitude values of the four types of currents, namely phase A, phase B, phase C, and total current, and then take the first m amplitude values and the last n amplitude values of the i-th current amplitude value of the class as the input values, so as to obtain k groups of labels and samples of the four types of currents, namely phase A, phase B, phase C, and total current.

[0047] For example: within the acquisition period T, it includes V 1 、V 2 、V 3 、……V kSampling values. Assume that within the sampling period T, the amplitude values of four types of currents, namely phase A, phase B, phase C, and total current, form a matrix M 4x100 , that is, each type of current amplitude has 100 sampling values. Assume the amplitude of phase A current is (P A1 , P A2 , P A3 , ……, P A100 ), the amplitude of phase B current is (P B1 , P B2 , P B3 , ……, P B100 ), the amplitude of phase C current is (P C1 , P C2 , P C3 , ……, P C100 ), and the amplitude of total current is (P N1 , P N2 , P N3 , ……, P N100 ).

[0048] At the i-th sampling point, the tag value of the amplitude of phase A current is P Ai , then the sample is the sum of the first m amplitudes and the last n amplitudes of the i-th amplitude of phase A current (P A(i-m) , P A(i-(m-1)) , ……, P A(i-1) , P A(i+1) , ……P A(i+(n-1)) , P A(i+n) ); and so on to obtain the samples of the amplitudes of the three types of currents, namely phase B, phase C, and total current, at the sampling point i:

[0049] (P B(i-m) , P B(i-(m-1)) , ……, P B(i-1) , P B(i+1) , ……P B(i+(n-1)) , P B(i+n) );

[0050] (P C(i-m) , P C(i-(m-1)) , ……, P C(i-1) , P C(i+1) , ……P C(i+(n-1)) , P C(i+n) );

[0051] (P N(i-m) , P N(i-(m-1)) , ……, P N(i-1) , P N(i+1) , ……P N(i+(n-1)) , P N(i+n) );

[0052] After data processing, the following tag and sample inputs are obtained:

[0053] The label matrix P of the label i is (P Ai , P Bi , P Ci , P Ni )

[0054] The sample input matrix Q of the sample i is

[0055]

[0056] Therefore, within the sampling period T, after data processing, the labels and samples of the amplitudes of the four types of currents, namely the A-phase, B-phase, C-phase, and total current, are obtained as follows:

[0057] The label matrix P of the label values

[0058] The sample input matrix Q of the sample is

[0059]

[0060] As Figure 2 shown, in the embodiment of the present invention, the specific steps of "if not, start constructing the detection model of this grounding node" in step 2 are as follows:

[0061] Construct the training data set of this grounding node with the labels and samples of the four types of currents of the k groups of A-phase, B-phase, C-phase, and total current, and pour the training data set into the neural network for model training. After the model training is completed, the detection model of this grounding node is formed, and at the same time, the detection model of this grounding node is incorporated into the grounding node model library.

[0062] In the embodiment of the present invention, by separately modeling each grounding node, the influence of other environmental variables on anomaly detection is prevented. At the same time, in the embodiment of the present invention, the neural network model based on deep learning is used to model the current amplitude of the grounding node, so that the model can remember or understand the temporal sequence of the current amplitude of the grounding node it is responsible for in the time domain.

[0063] In the embodiment of the present invention, the specific steps of "if so, determine whether there is an anomaly in the amplitudes of the four types of currents of the k groups of A-phase, B-phase, C-phase, and total current of this grounding node according to the detection model" in step 2 are as follows:

[0064] Input the samples corresponding to the amplitudes of the four types of currents of the k groups of A-phase, B-phase, C-phase, and total current into the detection model of this grounding point to obtain the model outputs of each group of the four types of currents of A-phase, B-phase, C-phase, and total current;

[0065] Calculate the ratio of the number of XOR values ​​between the label and model output of each group of A-phase current amplitude, B-phase current amplitude, C-phase current amplitude, and total current amplitude to the number of corresponding current amplitude samples, and determine whether the ratio is greater than the discrimination threshold. When the ratio is greater than the discrimination threshold, the amplitudes of the four currents of the group A-phase, B-phase, C-phase, and total current are recorded as abnormal values; otherwise, the amplitudes of the four currents of the group A-phase, B-phase, C-phase, and total current are recorded as normal values.

[0066] Therefore, the detection model described in the embodiment of the present invention can predict the intermediate target amplitude based on the previous and next amplitudes, and then compare the predicted intermediate target amplitude with the actual true amplitude. Within a certain error range, it can be determined whether the true amplitude is abnormal.

[0067] For example: receive four types of input current, convert them into corresponding samples and labels, input the samples into the detection model, and the model will output the corresponding output. For example, within the sampling period T, after data processing, k groups of labels and samples of four types of currents, namely phase A, phase B, phase C, and total current, are obtained. After conversion, the original data of each type of current (such as 100 current amplitudes) will be converted into a sample matrix Q with k rows and (m+n) columns. A , Q B , Q C or Q N , and a label matrix P with k rows and 1 column A , P B , P C or P N .

[0068] For example: According to the above method, the A phase current sample input matrix Q is obtained A for

[0069]

[0070] According to the above method, the label matrix P of phase A current is obtained A for

[0071]

[0072] Input the samples of phase A current into matrix Q A Input the detection model, and the detection model will output a matrix P with k rows and 1 column. A ', the array P A ' and the previously formed label matrix P A For comparison, when P A Any current amplitude and label matrix P in ' AIf the ratio of the current amplitude at the corresponding position is less than the discrimination threshold (or greater than the reciprocal of the discrimination threshold), then the current amplitude at the sampling point corresponding to the grounding node is considered an outlier. In the embodiments of the present invention, the discrimination threshold is an empirical value or the optimal value determined through repeated tests, but the numerical range is between 0 and 1. Therefore, the embodiments of the present invention can promptly detect abnormal grounding nodes, assist in diagnosing the health status of cables, realize the intelligentization and automation of cable operation and maintenance, and thus ensure the safety of cables and power-consuming units or individuals.

[0073] Assume that the sample matrix formed by inputting four types of currents is M 4x100 , that is, each type of current amplitude has 100 sampling values; and the current amplitudes considered abnormal by the detected model are [0, 12, 0, 88], indicating that there are no abnormal current amplitudes in the A-phase current, there are 12 abnormal current amplitudes in the B-phase current, there are no abnormal current amplitudes in the C-phase current, and there are 88 abnormal current amplitudes in the total current. Then the corresponding calculation method is: the abnormal proportion of the A-phase current is 0 / 100, the abnormal proportion of the B-phase current is 12 / 100, the abnormal proportion of the C-phase current is 0 / 100, and the abnormal proportion of the total current is 88 / 100. In the embodiments of the present invention, both m and n are respectively less than or equal to 100. In the embodiments of the present invention, the alarm threshold is an empirical value or the optimal value determined through repeated tests, but the numerical range is between 0 and 1.

[0074] Embodiment 2

[0075] As Figure 3 shown, the main difference between the cable grounding loop current abnormal detection method provided by the embodiments of the present invention and Embodiment 1 of the present invention is: further including: Step S5, continuously statistically calculate the average abnormal proportion of the current amplitude of this type corresponding to the grounding node being triggered for alarm in the subsequent t times, and determine whether the average abnormal proportion of the current amplitude of this type exceeds the alarm threshold. If so, issue an alarm and notify the operation and maintenance personnel that there is an abnormality in the current of this phase of the grounding node; otherwise, do not issue an alarm. In the embodiments of the present invention, the alarm threshold is an empirical value or the optimal value determined through repeated tests, but the numerical range is between 0 and 1. Therefore, the embodiments of the present invention can solve the problem that when the number of abnormal amplitude values in the current amplitudes of a certain number of phases is relatively large during a single detection, it can be determined that there is an abnormality in the current of a certain number of phases of this grounding node, thereby promptly locating the cause of the fault, assisting in diagnosing the health status of the cable, realizing the intelligentization and automation of cable operation and maintenance, and thus ensuring the safety of cables and power-consuming units or individuals.

[0076] In the above-mentioned first and second embodiments of the present invention, it is mainly based on a neural network model of deep learning, and uses the technology without the prerequisite of complex feature engineering. Therefore, through the autonomous learning of the neural network in the above-mentioned embodiments of the present invention, it is no longer limited to simply judging whether the current amplitude exceeds the threshold, but the model learns the order of these time-series data by itself through the collected data, and then when encountering time-series data that has not been seen before, it is considered that there may be an anomaly; at the same time, the time span of the data required during the training process does not need to be very long, and a corresponding neural network model is trained for each grounding node, so as to eliminate the influence of other environmental factors on the discrimination result.

[0077] In summary, the detection method of the present invention can not only timely detect abnormal grounding nodes and fault causes, but also assist in diagnosing the health status of cables, realizing the intelligentization and automation of cable operation and maintenance, and ensuring the safety of cables and electricity-using units or individuals.

[0078] Embodiment 3

[0079] As Figure 4 shown, the embodiment of the present invention further provides a cable grounding loop current anomaly detection device, wherein the detection device includes:

[0080] A data acquisition unit 100, configured to collect in real time the grounding loop current data of any grounding node of the cable to be detected, and select the amplitudes of k consecutive A-phase, B-phase, C-phase, and total current in the grounding loop current data of the grounding node within the acquisition period T, and obtain k groups of labels and samples of the A-phase, B-phase, C-phase, and total current through data processing;

[0081] A model judgment unit 200, configured to judge whether there is a detection model for the grounding node according to the grounding node model library. If not, start to establish a detection model for the grounding node. If so, jump to the next step S3;

[0082] A first warning unit 300, configured to count and calculate the abnormal proportion in the amplitudes of the k groups of A-phase, B-phase, C-phase, and total current of the grounding node, and judge whether the abnormal proportion of the grounding node exceeds the warning threshold. If so, issue a warning and notify the operation and maintenance personnel that the current of the grounding node is abnormal. Otherwise, end the current anomaly detection process.

[0083] In the embodiment of the present invention, the data acquisition unit 100 includes a data processing module 101, and the data processing module 101 is configured to take the i-th current amplitude of each type as the label value among the obtained k consecutive A-phase, B-phase, C-phase, and total current amplitudes, and then take the first m amplitudes and the last n amplitudes of the i-th current amplitude as input values, so as to obtain k groups of labels and samples of the A-phase, B-phase, C-phase, and total current.

[0084] In an embodiment of the present invention, the model judgment unit 200 includes a model construction module 201, which is used to form a training data set for the grounding node with k groups of labels and samples of four types of current, namely, phase A, phase B, phase C, and total current, and to feed the training data set into a neural network for model training. After the model training is completed, a detection model of the grounding node is formed, and the detection model of the grounding node is incorporated into the grounding node model library.

[0085] In the embodiment of the present invention, the model judgment unit 200 further includes a sample output module 202 and an abnormality judgment module 203, wherein the sample output module 202 is used to input the samples corresponding to the amplitudes of the four types of currents of k groups, namely, phase A, phase B, phase C, and total current, into the detection model of the grounding point, and obtain the model output of each group of the four types of currents, namely, phase A, phase B, phase C, and total current;

[0086] The abnormality judgment module 203 is used to calculate the ratio of the number of XOR values ​​between the label and model output of each group of A-phase current amplitude, B-phase current amplitude, C-phase current amplitude, and total current amplitude to the number of corresponding current amplitude samples, and judge whether the ratio is greater than the discrimination threshold. When the ratio is greater than the discrimination threshold, the amplitudes of the four types of currents of the group A-phase, B-phase, C-phase, and total current are recorded as abnormal values; otherwise, the amplitudes of the four types of currents of the group A-phase, B-phase, C-phase, and total current are recorded as normal values.

[0087] Embodiment 4

[0088] like Figure 5 As shown, the embodiment of the present invention provides a cable grounding loop current anomaly detection device, which is mainly different from the third embodiment of the present invention in that it also includes a second early warning unit 400, which is used to count and calculate the abnormal proportion of x current amplitudes of any type of current amplitude of the grounding node, and determine whether the average abnormal proportion of the current amplitude of this type exceeds the alarm threshold. If so, an alarm is issued and the operation and maintenance personnel are notified that the current in this direction of the grounding node is abnormal, otherwise no alarm is issued. In the embodiment of the present invention, the numerical range of the alarm threshold is between 0 and 1.

[0089] In summary, the detection device described in the present invention can not only detect abnormal grounding nodes and fault causes in a timely manner, but also assist in diagnosing the health status of the cable, realize the intelligent and automated operation and maintenance of the cable, and ensure the safety of the cable and electricity users or individuals.

[0090] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses, and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0091] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0092] Accordingly, the present invention further provides a computer-readable storage medium storing computer-executable instructions, which, when loaded and executed by a processor, implement the cable grounding loop current anomaly detection method as described in any one of the above.

[0093] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting abnormal cable grounding circulating current, characterized in that, it includes the following steps: Step S1: Continuously collect the grounding loop current data of any grounding node of the cable to be inspected in real time, and select the amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current, within the acquisition period T from the grounding loop current data of this grounding node. After data processing, obtain k sets of labels and samples for the four types of currents of phase A, phase B, phase C, and total current; the specific steps of "obtaining k sets of labels and samples for the four types of currents of phase A, phase B, phase C, and total current" in the above step S1 are as follows: Among the obtained continuous k amplitudes of the four types of currents of phase A, phase B, phase C, and total current, take the k th current amplitude of each type as the label value, and then take the first i amplitudes and the last i amplitudes of the m th current amplitude of this type as the input values, so as to obtain n sets of labels and samples for the four types of currents of phase A, phase B, phase C, and total current; k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k [[ Step S2: Determine whether there is a detection model for this grounding node according to the grounding node model library. If not, start constructing the detection model for this grounding node. If so, determine whether there are abnormalities in the amplitudes of the four types of currents in k Group A phase, B phase, C phase, and total current; Step S3: Count and calculate the k abnormal proportion of the amplitude of each type of current among the amplitudes of the four types of current, namely, the A-phase current, B-phase current, C-phase current, and total current, of the grounding node, and determine whether the abnormal proportion of the amplitude of each type of current of the grounding node exceeds the alarm threshold. If so, trigger an alarm and notify the operation and maintenance personnel that there is an abnormality in the current of the grounding node; otherwise, no alarm is issued.

2. The method for detecting abnormal cable grounding circulating current according to claim 1, characterized in that, It further includes: Step S4, after continuously counting and calculating the average abnormal ratio of the current amplitude of this type corresponding to the grounding node being triggered for alarm t for the number of times, and determining whether the average abnormal ratio of the current amplitude of this type exceeds the alarm threshold. If so, an alarm is issued and the operation and maintenance personnel are notified that there is an abnormality in the current of this type of the grounding node; otherwise, no alarm is issued.

3. The method for detecting abnormal cable grounding circulating current according to claim 1, characterized in that, the specific steps of "if not, start constructing the detection model of this grounding node" in step S2 are: The k labels and samples of the four types of current, namely, the group A-phase, B-phase, C-phase, and total current, form the training data set of this grounding node, and the training data set is poured into the neural network for model training. After the model training is completed, the detection model of this grounding node is formed, and at the same time, the detection model of this grounding node is incorporated into the grounding node model library.

4. The method for detecting abnormal cable grounding circulating current according to claim 1, characterized in that, In the step S2, "if so, determine the k specific steps for whether the amplitudes of the four types of currents of group A-phase, B-phase, C-phase, and total current are abnormal" are as follows: Input k the samples corresponding to the amplitudes of the four types of currents, namely, the phase A, phase B, phase C, and total current of the group, into the detection model of the grounding node to obtain the model outputs of the four types of currents, namely, the phase A, phase B, phase C, and total current of each group; Calculate the ratio of the number of XOR values between the label of the amplitude of each phase A current, phase B current, phase C current, and total current amplitude and the model output to the corresponding current amplitude sample quantity, and respectively judge whether this ratio is greater than the discrimination threshold. When this ratio is greater than the discrimination threshold, record the amplitude of this type of current among the amplitudes of the four types of currents of phase A, phase B, phase C, and total current in this group as an abnormal value; otherwise, record the amplitude of this type of current among the amplitudes of the four types of currents of phase A, phase B, phase C, and total current in this group as a normal value.

5. The method for detecting abnormal cable grounding circulating current according to claim 1 or 2 or 4, characterized in that, the numerical ranges of the discrimination threshold, warning threshold, and alarm threshold are all between 0 and 1 respectively.

6. A device for detecting abnormal cable grounding circulating current, characterized in that, this detection device includes: A data acquisition unit, configured to collect in real time the ground loop current data of any grounding node of a cable to be detected, and obtain, from the ground loop current data collected within the acquisition period T of the selected grounding node, the amplitudes of the continuous k amplitudes of four types of currents, namely, phase A, phase B, phase C, and total current, and obtain, through data processing, k groups of labels and samples of the amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current; the specific content of "obtain, through data processing, k groups of labels and samples of the amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current" is: from the obtained continuous k amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current, respectively take the i th current amplitude of each type as the label value, and then take the first i current amplitudes of each type, the first m amplitudes and the last n amplitudes as input values, so as to obtain k groups of labels and samples of the amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current; A model judgment unit, used to judge whether there is a detection model for this grounding node according to the grounding node model library. If not, start constructing the detection model of this grounding node. If there is, determine whether there is an abnormality in the amplitude of each phase A, phase B, phase C, and total current of this grounding node according to the detection model; The first warning unit is used to count and calculate the abnormal proportion of the grounding node in k the abnormal proportion of the amplitudes of the four types of currents, namely, phase A, phase B, phase C, and total current, of the group, and determine whether the abnormal proportion of the grounding node exceeds the warning threshold. If so, an alarm is issued and the operation and maintenance personnel are notified that there is an abnormality in the current of the grounding node; otherwise, the current abnormal detection process ends.

7. An electronic device, characterized in that, it includes: A memory, used to store computer programs; A processor, used to implement the method for detecting abnormal cable grounding circulating current according to any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium, characterized in that, the computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are loaded and executed by a processor, the method for detecting abnormal cable grounding circulating current according to any one of claims 1 to 5 is implemented.

Images