Cross-connection error sheath ground current feature data augmentation method and system

Abstract

A method and system for augmenting the characteristic data of sheath grounding current in cross-connection errors, belonging to the field of cable fault detection, includes setting multiple current transformers at both ends of the cable under test to acquire actual current data at both ends of each phase core in the cable under test under different sheath connection states; the current data includes core current data and sheath current data; constructing an equivalent circuit model of the cable under test to simulate different sheath connection states to acquire virtual current data at both ends of each phase core in the cable under test; generating random noise based on the actual and virtual current data, and inputting the random noise into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain time-continuous extended current data. This application acquires current data under different sheath connection states by constructing an equivalent circuit model of the cable, and performs deep learning based on real and virtual data to extend discrete data into continuous data, increasing the required data volume.

Description

Technical Field

[0001] This application relates to the field of cable fault detection, specifically to a method and system for augmenting the characteristic data of sheath grounding current in cross-connection errors. Background Technology

[0002] Cross-connection errors are one of the most common sheath grounding defects in high-voltage cable systems. In the past, the collection of sheath grounding current characteristic data mainly relied on field inspections, and the construction of defect datasets was limited by the field inspection data. Furthermore, the cable's RLC parameters are unknown during field inspections, and the formulation of defect diagnosis criteria depends on limited field inspection samples and human experience, often leading to misjudgments.

[0003] Therefore, it is urgent to increase the amount of data in the defect dataset in order to improve the accuracy of locating sheath grounding defects. Summary of the Invention

[0004] This application provides a method for augmenting the characteristic data of the sheath grounding current in cross-connection errors, which can solve the technical problem of limited data collection in traditional field testing when the RLC parameters of the cable are unknown during field testing and the defect data is limited.

[0005] In a first aspect, embodiments of this application provide a method for augmenting the characteristic data of sheath grounding current in cross-interconnection errors, applicable to cables under test with cross-interconnection grounding of the sheath; the method includes:

[0006] Multiple current transformers are installed at both ends of the cable under test to obtain the actual current data at both ends of each phase conductor in the cable under test under different sheath connection states; the current data includes conductor current data and sheath current data.

[0007] Construct an equivalent circuit model of the cable under test to simulate different sheath connection states in order to obtain virtual current data at both ends of each phase core in the cable under test.

[0008] Random noise is generated based on actual and virtual current data. This random noise is then input into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain extended current data that is continuous in time.

[0009] In conjunction with the first aspect, in one embodiment, the current transformer is disposed between the cable to be tested and the direct grounding box.

[0010] In conjunction with the first aspect, in one embodiment, the different sheath connection states include a sheath connection correct state and multiple sheath connection error states corresponding to specified connection error types.

[0011] In conjunction with the first aspect, in one embodiment, the equivalent circuit model includes an inductively coupled circuit model and a capacitively coupled circuit model.

[0012] In conjunction with the first aspect, in one implementation, the equivalent circuit model is constructed based on the resistance, inductance, and capacitance of the cable under test.

[0013] In conjunction with the first aspect, in one implementation, generating random noise based on actual and virtual current data specifically includes the following steps:

[0014] Actual and virtual deduced current data are obtained by processing actual and virtual sheath current data respectively; the deduced current data includes the sheath current vector difference between the first and last ends of the same conductor, the amplitude ratio of sheath current at any same end of different conductors, and the amplitude and vector of sheath current at any end of the same conductor.

[0015] The actual and virtual core current data and the inferred current data are used as the data to be processed;

[0016] The random noise is generated based on the data to be processed.

[0017] In conjunction with the first aspect, in one embodiment, the method further includes:

[0018] The random noise is generated by combining the data to be processed and the attribute data of the cable to be tested; the attribute data includes cable length, cable type, cable laying method, and load current.

[0019] In conjunction with the first aspect, in one implementation, generating random noise based on the data to be processed specifically includes the following steps:

[0020] After cleaning the data to be processed, high-frequency noise is removed by wavelet transform to obtain intermediate data;

[0021] Multiple time-domain features are extracted from intermediate data; these features include maximum value, minimum value, mean, variance, standard deviation, peak value, root mean square value, peak factor, and waveform factor.

[0022] The random noise is obtained by normalizing the intermediate data.

[0023] Secondly, embodiments of this application provide a cross-connection error sheath grounding current characteristic data augmentation system, the system comprising:

[0024] The real data acquisition module is used to set multiple current transformers at both ends of the cable under test to acquire the actual current data at both ends of each phase core in the cable under test under different sheath connection states; the current data includes core current data and sheath current data.

[0025] The virtual data acquisition module is used to construct an equivalent circuit model of the cable under test, simulate different sheath connection states, and obtain virtual current data at both ends of each phase core in the cable under test.

[0026] An extended data acquisition module is used to generate random noise based on actual and virtual current data, and input the random noise into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain extended current data with continuous time.

[0027] In conjunction with the second aspect, in one implementation, when the extended data acquisition module generates random noise based on actual and virtual current data, it processes the actual and virtual sheath current data to obtain actual and virtual deduced current data respectively; the deduced current data includes the sheath current vector difference between the beginning and end of the same conductor, the amplitude ratio of the sheath current at any same end of different conductors, and the amplitude and vector of the sheath current at any end of the same conductor.

[0028] The actual and virtual core current data and the inferred current data are used as the data to be processed;

[0029] The random noise is generated based on the data to be processed.

[0030] The beneficial effects of the technical solutions provided in this application include:

[0031] By setting current transformers at both ends of the cable under test to obtain real current data, and by constructing an equivalent circuit model of the cable under test to obtain virtual current data, and given that both real and virtual current data are discrete and limited in quantity, deep learning is performed based on the real and virtual data to generate simulated current data as extended data. This not only increases the quantity of current data, but also the variety of current data, providing more comprehensive and accurate support for subsequent defect detection. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating an embodiment of the method for augmenting the characteristic data of the sheath grounding current in cross-connection errors according to this application;

[0033] Figure 2 A schematic diagram of the installation of a three-phase nine-segment high-voltage cable sensor;

[0034] Figure 3 A schematic diagram illustrating an incorrect connection of the sheath of a three-phase nine-segment high-voltage cable.

[0035] Figure 4 The inductive coupling circuit diagram of loop 1 when the high-voltage cable sheath is correctly connected;

[0036] Figure 5The capacitive coupling circuit diagram for loop 1 when the high-voltage cable sheath is correctly connected;

[0037] Figure 6 This is a schematic diagram of the functional modules of an embodiment of the cross-connection error sheath grounding current characteristic data augmentation system of this application. Detailed Implementation

[0038] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0039] First, some of the technical terms used in this application will be explained to help those skilled in the art understand this application.

[0040] Sheath: In high-voltage cables, it usually refers to the metal sheath (such as aluminum sheath, lead sheath or copper sheath), which is located on the outside of the cable insulation layer.

[0041] Cross-connection grounding: Long cables are segmented and then cross-connected to balance the induced voltage and reduce losses.

[0042] Three-phase nine-segment sheath: By dividing each phase sheath into 3 segments, for a total of 9 segments across the three phases, self-balancing of induced voltage is achieved. Each conductor has an independent metal sheath, but the sheaths are segmented and connected by a cross-connection box, rather than simply wrapped with multiple layers of sheath.

[0043] Convolutional Neural Networks (CNNs) are a type of feedforward neural network that incorporates convolutional computations and has a deep structure.

[0044] Generative Adversarial Networks (GANs) consist of two neural networks: a generator responsible for generating realistic fake data (such as images) and a discriminator responsible for determining whether the input data is real or faked by the generator. They continuously optimize through adversarial training; the generator attempts to "deceive" the discriminator, while the discriminator strives to improve its discriminative ability.

[0045] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0046] In a first aspect, embodiments of this application provide a method for augmenting the characteristic data of the sheath grounding current in cross-connection errors.

[0047] In one embodiment, reference is made to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the method for augmenting the characteristic data of sheath grounding current in cross-connection errors according to this application. Figure 1 As shown, the methods for augmenting the characteristic data of the sheath grounding current in cross-connection errors include:

[0048] Step S1: Install multiple current transformers at both ends of the cable under test to obtain the actual current data at both ends of each phase conductor in the cable under test under different sheath connection conditions. The current data includes conductor current data and sheath current data.

[0049] Step S2: Construct an equivalent circuit model of the cable under test to simulate different sheath connection states in order to obtain virtual current data at both ends of each phase core in the cable under test.

[0050] Step S3: Generate random noise based on actual and virtual current data, and input the random noise into a CNN-based GAN to obtain time-continuous extended current data.

[0051] In this embodiment, real current data is obtained by setting current transformers at both ends of the cable under test, and virtual current data is obtained by constructing an equivalent circuit model of the cable under test. When both real and virtual current data are discrete data and the amount of data is limited, deep learning is performed based on real and virtual data to generate simulated current data as extended data. This not only increases the amount of current data, but also increases the types of current data, providing more comprehensive and accurate support for subsequent defect detection.

[0052] Furthermore, in one embodiment, the aforementioned current transformer is disposed between the cable to be tested and the direct grounding box.

[0053] In this embodiment, current transformers are installed between the high-voltage cable body and the direct grounding box to accurately detect the current data of the conductor and sheath.

[0054] Furthermore, in one embodiment, the different sheath connection states include a sheath connection correct state and multiple sheath connection error states corresponding to specified connection error types.

[0055] In this embodiment, grounding current data of high-voltage cable sheaths under different operating conditions are collected, including normal conditions and typical cross-interconnection error conditions.

[0056] Furthermore, in one embodiment, the above equivalent circuit model includes an inductively coupled circuit model and a capacitively coupled circuit model.

[0057] In this embodiment, the sheath current of the cross-connected high-voltage cable is the superposition of leakage current component and inductive coupling current component. Therefore, it is necessary to establish equivalent circuit models for the leakage component and the inductive coupling component respectively.

[0058] In one specific embodiment, refer to Figure 2 and Figure 3 As shown, Figure 2 This is a schematic diagram of the installation of a three-phase nine-segment high-voltage cable sensor. Figure 3 This is a schematic diagram of an incorrect connection of the sheath of a three-phase nine-segment high-voltage cable. The diagram shows that the nine metal sheath segments of the cable are connected by… , , , , , , , , This indicates that the metal sheath circuits 1, 2, and 3 are respectively , , . — This is a sheath protector installed in a cross-connection grounding box. , , , , , To install current transformers at the beginning and end of the cable line, the sheath current at both ends of the cable sheath is measured. The current transformer clipped onto the cable body measures the superposition of the core current and the sheath current of the cable body.

[0059] Reference Figure 4 As shown, Figure 4 The diagram shows the inductive coupling circuit of loop 1 when the high-voltage cable sheath is correctly connected. , The induced electromotive force of the short section of the sheath is calculated using the following formula (1):

[0060] (1)

[0061] in, , , They are respectively Induced electromotive force in the cable sheath. , , They are respectively , Section cable sheath impedance, , , They are respectively The current phasors of phase A, phase B, and phase C of a section of cable core under a purely inductive load.

[0062] The node voltage equation for the inductively coupled circuit of loop 1 is as follows: ,in, This is the topology matrix of the inductively coupled circuit of loop 1. Let be the admittance matrix of node 1 of loop. The column vector of nodal voltages. These are the phasors of the voltage source. The following formula (2) is used to calculate the result. The following formula (3) is used to calculate the result. The following formula (4) is used to calculate the result. The following formula (5) is used to calculate:

[0063] (2)

[0064] (3)

[0065] (4)

[0066] (5)

[0067] in, and The grounding impedance is on both the left and right sides. and The equivalent impedance of the sheath protector. and These are the voltages of the ground branch node where the sheath protector is located relative to the ground.

[0068] Inductive current components at both ends of the sheath of circuit 1 , The following formula (6) is used to calculate the result.

[0069] (6)

[0070] Reference Figure 5 As shown, Figure 5 This is the capacitive coupling circuit diagram for loop 1 when the high-voltage cable sheath is correctly connected. , , They are respectively , , The insulation impedance of the cable segment is given in Ω, and the node voltages relative to ground at each node are given as follows: , , , , , , , They represent , , Leakage current (i.e., capacitive current) of the cable sheath. Assuming ground is the reference node, list the node voltage equations for loop 1. Calculate the phase voltages respectively. , , Under individual action, the nodal voltage equations of the five nodes relative to the ground are as follows: The circuit topology matrix of the capacitive coupling current in loop 1. The nodal admittance matrix is ​​calculated using the following formula (7), and the nodal voltage column vector is calculated using the following formula (8). The following formula (9) is used to calculate:

[0071] (7)

[0072] (8)

[0073] (9)

[0074] x can be A, B, or C. This represents the node voltage generated by phase x at nodes 1 to 5. The column vector of voltage sources generated by phase x is calculated using the following formula (10):

[0075] (10)

[0076] Sheath current at both ends a1 and c2 in circuit 1 The following formula (11) is used to calculate:

[0077] (11)

[0078] and The current sensor collects the actual grounding current of the sheath, which includes inductive current and capacitive current (i.e., leakage current).

[0079] Furthermore, in one embodiment, the above equivalent circuit model is constructed based on the resistance, inductance, and capacitance of the cable under test.

[0080] In this embodiment, when the actual RLC (resistance R, inductance L, capacitance C) parameters of the cable are unknown and the defect data is limited, an efficient circuit model is constructed based on the theoretical RLC. Then, deep learning methods are used to extend new simulation data based on real and virtual current data, overcoming the limitations of traditional field detection data collection.

[0081] Furthermore, in one embodiment, the above-mentioned generation of random noise based on actual and virtual current data specifically includes the following steps:

[0082] Actual and virtual deduced current data are obtained by processing actual and virtual sheath current data respectively. The deduced current data includes the vector difference of sheath current at the beginning and end of the same conductor, the ratio of the amplitude of sheath current at any identical end of different conductors, and the amplitude and vector of the sheath current at any end of the same conductor.

[0083] The actual and virtual core current data and the inferred current data are used as the data to be processed.

[0084] Based on the data to be processed, the above-mentioned random noise is generated.

[0085] In this embodiment, the core current, the sheath current at the beginning and end of the same circuit, the phasor difference of the sheath current at the beginning and end, the ratio of the current amplitude at the beginning (or end) of the sheath in different circuits, and the current amplitude and phase at the beginning (or end) of the sheath are collected as characteristic quantities.

[0086] Furthermore, in one embodiment, the above method further includes:

[0087] The random noise is generated by combining the data to be processed and the attribute data of the cable to be tested. The attribute data includes cable length, cable type, cable laying method, and load current.

[0088] In this embodiment, when collecting data, basic parameters such as cable length, cable type, and laying method, as well as operating conditions such as load current, are recorded to increase the dimensionality of the dataset and facilitate the later location of sheath connection errors for different types of cables under different load current conditions.

[0089] Furthermore, in one embodiment, the above-mentioned generation of random noise based on the data to be processed specifically includes the following steps:

[0090] After cleaning the data to be processed, high-frequency noise is removed by wavelet transform to obtain intermediate data.

[0091] Multiple time-domain features are extracted from the intermediate data. These features include maximum value, minimum value, mean, variance, standard deviation, peak value, root mean square value, peak factor, and waveform factor.

[0092] After normalizing the intermediate data, the above random noise is obtained.

[0093] In this embodiment, outliers are removed by moving median filtering. The specific steps are as follows:

[0094] a) For the original signal data sequence Define a sliding window of length k, where k is odd and 3≤k≤n / 10.

[0095] b) Slide the window sequentially from the beginning of the sequence, and calculate the median Mi for the k data points within the window.

[0096] c) If the absolute value of the difference between the current center point xi and the median Mi exceeds the preset threshold. If the value is not found in the original value, it is considered an outlier, and the median Mi is used to replace the original value.

[0097] If |xi - Mi|> Then xi = Mi.

[0098] d) Repeat steps b to c until the entire data sequence has been processed.

[0099] Furthermore, high-frequency noise is removed using wavelet transform, with the following specific steps:

[0100] a) For the signal data sequence after moving median filtering, select an appropriate wavelet basis function to perform wavelet decomposition and obtain wavelet coefficients at each scale.

[0101] b) Set the threshold based on the frequency characteristics of the signal. Thresholding is performed on the high-frequency wavelet coefficients:

[0102] If |dj,k| < If , then dj,k = 0.

[0103] If |dj,k| ≥ , then dj,k = sgn(dj,k)·(|dj,k | - ).

[0104] Where dj,k is the k-th wavelet coefficient at scale j.

[0105] c) Use the processed wavelet coefficients to perform wavelet reconstruction to obtain the signal data sequence after removing high-frequency noise.

[0106] Furthermore, the signal data sequence after data cleaning processing... Extract the following time-domain features:

[0107] Maximum value: .

[0108] Minimum value: .

[0109] Mean: .

[0110] variance: .

[0111] Standard deviation: .

[0112] Peak value: .

[0113] Root mean square value: .

[0114] Peak factor: CF = Peak / RMS.

[0115] Waveform factor: .

[0116] In one specific embodiment, the Z-score normalization method is used to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1, thereby eliminating scale differences under different measurement conditions.

[0117] By extracting data features, a cross-connection misconnection identification algorithm is implemented, fully describing the characteristics of sheath current changes over time and capturing the unique time-varying characteristics of different defect types. Specific cable attribute parameters are shown in Table 1 below:

[0118] Table 1. List of Cable Attribute Parameters

[0119]

[0120] Based on Table 1 and all current parameters, the feature mean vector and covariance matrix are calculated for each defect type. Principal Component Analysis (PCA) is used to extract the main feature patterns for each type of sheath connection error defect.

[0121] For defect types with sufficient sample size, cluster analysis is performed to capture potential sub-category features.

[0122] Multiple classifiers, including Random Forest, SVM, and Gradient Boosting Tree, are trained, and their predictions are integrated using a confidence-weighted ensemble method. A deep learning model is constructed where static features are processed through a fully connected network, and temporal features are processed through a CNN-LSTM hybrid network. These two types of features are fused at a deep layer to jointly predict the defect type and location. The similarity between the input features and the defect template is calculated, and Mahalanobis distance is used to consider the correlation between features, thereby improving recognition accuracy.

[0123] In another embodiment, the generator network structure uses a one-dimensional convolutional transpose layer (Conv1DTranspose) to generate temporal features, generates simulated sheath current feature data from random noise, and outputs a generation channel.

[0124] The discriminator network structure uses a one-dimensional convolutional network to extract temporal features, and a multi-layer convolutional structure to progressively extract higher-level feature representations, ultimately outputting a binary classification result to determine whether a sample is real or generated.

[0125] The GAN training process involves alternating training of the discriminator and the generator. The discriminator learns to distinguish between real and generated sheath current features, while the generator learns to generate more realistic sheath current features to "deceive" the discriminator.

[0126] The fault diagnostic tool uses a trained GAN model to generate additional samples to enhance the training data. It is based on a CNN classifier structure and is specifically designed for fault diagnosis based on sheath current characteristics. It outputs the fault classification results and their probabilities.

[0127] In summary, a CNN-based method for augmenting the sheath grounding current characteristic data of high-voltage cable cross-connection errors is proposed. This method effectively overcomes the limitations faced by traditional field inspection data collection when cable RLC parameters are unknown and defect data is limited. This method can intelligently augment cable characteristic data to generate diverse simulation data, providing more comprehensive and accurate support for defect detection.

[0128] Secondly, embodiments of this application also provide a system for augmenting the characteristic data of sheath grounding current in cross-connection errors.

[0129] In one embodiment, reference is made to Figure 6 , Figure 6 This is a functional module diagram of an embodiment of the cross-connection error sheath grounding current characteristic data augmentation system of this application. Figure 6 As shown, the cross-connection error sheath grounding current characteristic data augmentation system includes:

[0130] The real data acquisition module 1 is used to install multiple current transformers at both ends of the cable under test to acquire the actual current data at both ends of each phase conductor in the cable under test under different sheath connection states. The aforementioned current data includes conductor current data and sheath current data.

[0131] Virtual data acquisition module 2 is used to construct an equivalent circuit model of the cable under test, simulate different sheath connection states, and obtain virtual current data at both ends of each phase core in the cable under test.

[0132] Extended data acquisition module 3 is used to generate random noise based on actual and virtual current data, and input the random noise into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain extended current data with continuous time.

[0133] In this embodiment, real current data is obtained by setting current transformers at both ends of the cable under test, and virtual current data is obtained by constructing an equivalent circuit model of the cable under test. When both real and virtual current data are discrete data and the amount of data is limited, deep learning is performed based on real and virtual data to generate simulated current data as extended data. This not only increases the amount of current data, but also increases the types of current data, providing more comprehensive and accurate support for subsequent defect detection.

[0134] Furthermore, in one embodiment, the aforementioned current transformer is disposed between the cable to be tested and the direct grounding box.

[0135] In this embodiment, current transformers are installed between the high-voltage cable body and the direct grounding box to accurately detect the current data of the conductor and sheath.

[0136] Furthermore, in one embodiment, the different sheath connection states include a sheath connection correct state and multiple sheath connection error states corresponding to specified connection error types.

[0137] In this embodiment, grounding current data of high-voltage cable sheaths under different operating conditions are collected, including normal conditions and typical cross-interconnection error conditions.

[0138] Furthermore, in one embodiment, the above equivalent circuit model includes an inductively coupled circuit model and a capacitively coupled circuit model.

[0139] In this embodiment, the sheath current of the cross-connected high-voltage cable is the superposition of leakage current component and inductive coupling current component. Therefore, it is necessary to establish equivalent circuit models for the leakage component and the inductive coupling component respectively.

[0140] The functions of each module in the aforementioned cross-interconnection error sheath grounding current characteristic data augmentation device correspond to the steps in the aforementioned cross-interconnection error sheath grounding current characteristic data augmentation method embodiment, and their functions and implementation processes will not be described in detail here.

[0141] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0142] The terms "comprising" and "having," and any variations thereof, in the specification, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. The terms "first," "second," and "third," etc., are used to distinguish different objects, etc., and do not indicate a sequence, nor do they limit "first," "second," and "third" to different types.

[0143] In the description of the embodiments of this application, terms such as "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a concrete manner.

[0144] In the description of the embodiments of this application, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "multiple" means two or more.

[0145] In some processes described in the embodiments of this application, multiple operations or steps are included in a specific order. However, it should be understood that these operations or steps may not be executed in the order they appear in the embodiments of this application, or they may be executed in parallel. The sequence number of the operation is only used to distinguish different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed sequentially or in parallel, and these operations or steps may be combined.

[0146] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0147] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for augmenting the characteristic data of sheath grounding current in cross-connection errors, characterized in that, Applicable to cables under test with cross-connected grounding of sheaths; the method includes: Multiple current transformers are installed at both ends of the cable under test to obtain the actual current data at both ends of each phase conductor in the cable under test under different sheath connection states; the current data includes conductor current data and sheath current data. An equivalent circuit model of the cable under test is constructed to simulate different sheath connection states in order to obtain virtual current data at both ends of each phase core in the cable under test; the equivalent circuit model includes an inductive coupling circuit model and a capacitive coupling circuit model. Random noise is generated based on actual and virtual current data, and then input into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain temporally continuous extended current data. The specific steps for generating random noise based on actual and virtual current data include the following: Actual and virtual deduced current data are obtained by processing actual and virtual sheath current data respectively; the deduced current data includes the sheath current vector difference between the first and last ends of the same conductor, the amplitude ratio of sheath current at any same end of different conductors, and the amplitude and vector of sheath current at any end of the same conductor. The actual and virtual core current data and the inferred current data are used as the data to be processed; The random noise is generated based on the data to be processed.

2. The method for augmenting the characteristic data of the sheath grounding current in cross-connection errors as described in claim 1, characterized in that, The current transformer is installed between the cable to be tested and the direct grounding box.

3. The method for augmenting the characteristic data of the sheath grounding current in cross-connection errors as described in claim 1, characterized in that, The different sheath connection states include a correct sheath connection state and multiple sheath connection error states corresponding to specified connection error types.

4. The method for augmenting the characteristic data of the sheath grounding current in cross-connection errors as described in claim 1, characterized in that, The equivalent circuit model is constructed based on the resistance, inductance, and capacitance of the cable under test.

5. The method for augmenting the characteristic data of the sheath grounding current in cross-connection errors as described in claim 1, characterized in that, The method further includes: The random noise is generated by combining the data to be processed and the attribute data of the cable to be tested; the attribute data includes cable length, cable type, cable laying method, and load current.

6. The method for augmenting the characteristic data of the sheath grounding current in cross-connection errors as described in claim 1, characterized in that, The random noise is generated based on the data to be processed, specifically including the following steps: After cleaning the data to be processed, high-frequency noise is removed by wavelet transform to obtain intermediate data; Multiple time-domain features are extracted from intermediate data; these features include maximum value, minimum value, mean, variance, standard deviation, peak value, root mean square value, peak factor, and waveform factor. The random noise is obtained by normalizing the intermediate data.

7. A system for augmenting the characteristic data of sheath grounding current in cross-connection errors, characterized in that, The system includes: The real data acquisition module is used to set multiple current transformers at both ends of the cable under test to acquire the actual current data at both ends of each phase core in the cable under test under different sheath connection states; the current data includes core current data and sheath current data. The virtual data acquisition module is used to construct an equivalent circuit model of the cable under test, simulate different sheath connection states, and obtain virtual current data at both ends of each phase core in the cable under test; the equivalent circuit model includes an inductive coupling circuit model and a capacitive coupling circuit model. An extended data acquisition module is used to generate random noise based on actual and virtual current data, and input the random noise into a generative adversarial network (GAN) based on a convolutional neural network (CNN) to obtain extended current data with continuous time. The generation of random noise based on actual and virtual current data specifically includes the following steps: Actual and virtual deduced current data are obtained by processing actual and virtual sheath current data respectively; the deduced current data includes the sheath current vector difference between the first and last ends of the same conductor, the amplitude ratio of sheath current at any same end of different conductors, and the amplitude and vector of sheath current at any end of the same conductor. The actual and virtual core current data and the inferred current data are used as the data to be processed; The random noise is generated based on the data to be processed.

8. The cross-connection error sheath grounding current characteristic data augmentation system as described in claim 7, characterized in that, When the extended data acquisition module generates random noise based on actual and virtual current data, it processes the actual and virtual sheath current data to obtain actual and virtual inferred current data respectively. The deduced current data includes the vector difference of sheath current at the beginning and end of the same conductor, the ratio of the amplitude of sheath current at any identical end of different conductors, and the amplitude and vector of sheath current at any end of the same conductor. The actual and virtual core current data and the inferred current data are used as the data to be processed; The random noise is generated based on the data to be processed.

Images