Double-sheath loop interphase short circuit defect detection method based on adjacent sheath current phasor difference

Abstract

The invention relates to a double-sheath loop interphase short circuit defect detection method based on adjacent sheath current phasor difference, and the method comprises the following specific steps: 1, constructing a circuit network model of a high-voltage cable, and recognizing different types of grounding fault mechanisms; 2, performing simulation to obtain sheath grounding current data, and performing feature extraction on the sheath grounding current data to obtain the amplitude and the phase of the current; step 3, training and testing the extracted current features by adopting a proper machine learning algorithm, and improving the accuracy of fault detection by constructing a classification model; and 4, classifying the newly collected current data by using the trained model, and judging the fault category, thereby taking corresponding maintenance measures in time, and ensuring the safe and stable operation of the cable. The accuracy of fault detection is improved by constructing the classification model, and the efficiency of sheath interphase short circuit maintenance at the present stage is improved.

Description

Technical Field

[0001] The present application relates to the field of high-voltage cable monitoring, and in particular to a method for detecting interphase short circuit defects in a double-sheath loop based on the phase difference of adjacent sheath currents. Background Art

[0002] High-voltage cables (110kV and above) are the backbone of urban transmission and distribution networks, and generally adopt a three-phase three-wire system. The sheath of high-voltage cables adopts a cross-interconnection method, which enhances the anti-interference and protection capabilities of the system. However, in practical applications, the aging or breakdown of the material at the joint is still an important factor leading to the short circuit of the double-sheath loop. This type of fault not only affects the reliability of the power system, but may also cause serious safety hazards. Therefore, the double-sheath loop phase-to-phase short-circuit defect detection method based on the phasor difference of adjacent sheath currents aims to timely identify and locate short-circuit defects by monitoring and analyzing the phasor difference of the sheath current, thereby improving the operation safety and maintenance efficiency of high-voltage cables, which has important theoretical significance and practical value for ensuring the stable operation of the power system. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a method for detecting phase-to-phase short circuit defects in a double-sheath loop based on the phase difference of adjacent sheath currents, thereby reducing short circuits and other power system accidents caused by sheath failures and improving the operating reliability of high-voltage cables.

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] The embodiment of the present application provides a method for detecting interphase short circuit defects in a double-sheath loop based on the phase difference of adjacent sheath currents, comprising the following specific steps:

[0006] Step 1: construct a circuit network model of the high-voltage cable to identify different types of ground fault mechanisms;

[0007] Step 2: obtain the sheath grounding current data through simulation, extract its features, and obtain the current amplitude and phase;

[0008] Step 3: Use appropriate machine learning algorithms to train and test the extracted current features, and improve the accuracy of fault detection by building a classification model;

[0009] Step 4: Use the trained model to classify the newly collected current data and determine the fault type, so that corresponding maintenance measures can be taken in time to ensure the safe and stable operation of the cable.

[0010] The circuit network model of the high-voltage cable is specifically constructed by establishing a three-phase nine-segment cross-interconnected high-voltage cable line model, wherein the nine metal sheath segments of the cable sheath of the high-voltage cable line model are represented by A1, A2, A3, B1, B2, B3, C1, C2, and C3, wherein the cross-interconnection mode is "n", and the metal sheath loops 1, 2, and 3 are represented by yellow, green, and red, respectively; is the core current of the three-phase cable; the direct grounding boxes on both sides are G1 and G2, and the cross-connected grounding boxes in the middle are J1 and J2. The sheath grounding current measured by the current transformer at the beginning and end of the installed cable line.

[0011] The mechanism for identifying different types of grounding faults is specifically that the sheath current of the cross-connected high-voltage cable is a superposition of the leakage current component and the inductive coupling current component. When the double-sheath circuit fails, the sheath grounding current of the two circuits will be abnormal, which may be caused by the breakdown of the epoxy preform and the epoxy board causing the double-sheath circuit phase short circuit. When there is no fault, the current flowing into or out of the sheath system should be zero. According to the KCL law:

[0012]

[0013] In the formula They are the sum of the leakage currents of loops 1, 2, and 3 respectively. Under normal circumstances, the phasor difference of the grounding current at both ends of the sheath is the phasor sum of the three leakage currents of the sheath loop. The relative change of the leakage current is ≤1%, which is used as a reference value for judging whether there is a grounding fault in the sheath.

[0014] For a double-sheath circuit, when a phase-to-phase short circuit occurs, the sum of the grounding current phasor differences of the two adjacent circuits is normal, but the grounding current is abnormal, so:

[0015]

[0016] Where I R is the short-circuit current, from which the phase-to-phase short-circuit fault of the double-sheath circuit can be distinguished.

[0017] The simulation obtains the sheath grounding current data, and extracts its features to obtain the current amplitude and phase, specifically,

[0018] Considering the changing law of grounding current under different faults, load status and grounding resistance change factors, two types of features are extracted: the grounding current at six grounding points and the phasor difference of adjacent sheath grounding currents. The phasor difference of adjacent sheath grounding currents can find out the sheath phase-to-phase faults. Since the current characteristics have no directionality, the amplitude and phase angle of the grounding current phasor are introduced as a common judgment basis.

[0019] Aiming at the typical faults of high-voltage cable sheath, a simulation model of cross-connection of high-voltage cable sheath is built, and fault data is generated by simulation. The prepared data set consists of multiple operation matrices with a depth equal to the number of input data channels, and the data is randomly shuffled. The data is divided into training set and test set, and data preprocessing is performed. The characteristic parameters are normalized by using the premnmx normalization function, and its calculation expression is as follows:

[0020]

[0021] In the formula, X is the normalized value, x is the original data, min is the minimum value of the characteristic parameter, and max is the maximum value. Four types of phase difference characteristic parameters are extracted from the signal of the defective cable. In addition, there is also the mean square error. The calculation expression and explanation are as follows:

[0022]

[0023] The mean square error represents the arithmetic square root of the variance of the temperature data. Its physical meaning reflects the degree of discreteness of the data. After training, the model is evaluated by making predictions on the test set. Common evaluation indicators include accuracy, error curve, feature importance, and confusion matrix.

[0024] Appropriate machine learning algorithms are used to train and test the extracted current features, and the accuracy of fault detection is improved by building a classification model. Specifically, the specific steps of cable defect recognition and classification based on SVM are as follows: first, the sheath current is extracted through the simulation model, and then 4 types of grounding current phasor difference data are extracted and normalized. The grounding current phasor difference data are divided into a test set and a training set in a ratio of 3:1, and the defect classification is completed through the selection of classifiers, kernel functions and penalty factors. Among the six types of samples, each type of sample constitutes a binary classification problem with other types of samples. A total of 15 SVMs are constructed, and the test samples classify and judge all constructed SVMs. The trained model can be used to classify and distinguish new data, and the category of unknown samples is predicted through the voting mechanism to achieve efficient and accurate classification and discrimination of sheath faults.

[0025] Compared with the prior art, the beneficial effects of the present application are: improving the accuracy of fault detection by constructing a classification model, thereby improving the efficiency of current sheath interphase short-circuit maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 A flow chart of the method of this application;

[0028] Figure 2 This is a diagram of the sheath structure of a three-phase nine-section cross-interconnected high-voltage cable of the present application;

[0029] Figure 3 This is the sheath topology diagram of the three-phase nine-section cross-interconnected high-voltage cable of this application;

[0030] Figure 4 This is a bar graph of the correlation analysis of this application;

[0031] Figure 5 A line graph showing the accuracy of the phase-to-phase short circuit fault classification of the present application;

[0032] Figure 6 Visualization of the t-distributed random neighborhood embedding for this application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0034] The terms "comprises," "comprising," or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0035] The terms "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and should not be understood as indicating or implying relative importance, nor should they be understood as requiring or implying any such actual relationship or order between these entities or operations.

[0036] The present invention proposes to use two groups of six current sensors to be installed on the grounding boxes at both ends of the high-voltage cable, and to detect the sheath currents of the six cross-connected grounding points at the same time. The sheath currents at different detection points are synchronized online. This scheme aims to detect and classify the interphase short circuit defects of the double sheath loop through a systematic method to improve the operating safety and reliability of the high-voltage cable.

[0037] like Figure 1As shown, a method for detecting interphase short circuit defects in a double-sheath circuit based on the phase difference of adjacent sheath currents comprises the following specific steps:

[0038] Step 1: construct a circuit network model of the high-voltage cable to identify different types of ground fault mechanisms;

[0039] Step 2: obtain the sheath grounding current data through simulation, extract its features, and obtain the current amplitude and phase;

[0040] Step 3: Use appropriate machine learning algorithms to train and test the extracted current features, and improve the accuracy of fault detection by building a classification model;

[0041] Step 4: Use the trained model to classify the newly collected current data and determine the fault type, so that corresponding maintenance measures can be taken in time to ensure the safe and stable operation of the cable.

[0042] Through the above steps, the technical solution can realize accurate detection and classification of double-layer sheath loop grounding defects, and provide a scientific basis for fault warning and maintenance of cable systems.

[0043] 1) Circuit network model

[0044] Figure 2 The figure is a schematic diagram of a three-phase nine-section cross-interconnected high-voltage cable line. In the figure, the nine metal sheath sections of the cable sheath are represented by A1, A2, A3, B1, B2, B3, C1, C2, and C3, and the cross-interconnection method is "n", n, and p". The metal sheath loops 1, 2, and 3 are represented by yellow, green, and red, respectively. is the core current of the three-phase cable; the direct grounding boxes on both sides are G1 and G2, and the cross-connected grounding boxes in the middle are J1 and J2. The sheath grounding current measured by the current transformer at the beginning and end of the installed cable line.

[0045] In order to quantitatively analyze the influence of sheath interphase short circuit fault on the sheath current phasor difference of the direct grounding point of the same sheath loop, an equivalent circuit model of the cross-connected high-voltage cable under sheath fault state is established, and then the sinusoidal steady-state solution of the capacitive and inductive coupling components of the sheath current is derived based on the conduction coupling theory, such as Figure 3 shown.

[0046] In the picture: Z iA1 , Z iB2 , Z iC3 The insulation resistance of the cables in sections A1, B2 and C3, Ω; I LA1 ,I LB2 ,I LC3 Respectively represent the capacitive coupling of A1, B2, and C3 segments, A; Z SA1 , ZSB2 , Z SC3 A1, B2, C3 are the cable sheath impedances, Ω. For a network with N+1 nodes and B branches, since the voltage and current of each branch are unknown, there are 2B unknown quantities. KCL can list N independent equations, and KVL can list BN independent equations. The vector form of these equations is

[0047] A m I bm =0 (1)

[0048] B fm U bm =0 (2)

[0049] Where Am is the circuit association matrix, B fm is the basic loop matrix, I bm is the branch current vector, U bm is the branch voltage vector. For a network without a controlled source, the branch current matrix equation of the network can be written as:

[0050] U bm =Z bm I bm (s)+Z bm I sm -U sm (3)

[0051] Where Zbm is the passive element impedance matrix, Ism is the current source phasor matrix, and the voltage source phasor matrix Usm includes the induced voltage and the line core voltage. The induced voltage is solved according to the existing literature method.

[0052] I sm =0 (4)

[0053] Combining 123, we can solve for the grounding current on both sides of the sheath:

[0054]

[0055] 2) Differentiate the failure mechanism

[0056] The sheath current of the cross-connected high-voltage cable is the superposition of the leakage current component and the inductive coupling current component. When the double sheath circuit fails, the sheath grounding current of the two circuits will be abnormal, which may be caused by the breakdown of the epoxy preform and the epoxy board causing the double sheath circuit phase short circuit. When there is no fault, the current flowing into or out of the sheath system should be zero. According to the KCL law:

[0057]

[0058] In the formula They are the sum of the leakage currents of loops 1, 2, and 3. Under normal circumstances, the phasor difference of the grounding current at both ends of the sheath is the phasor sum of the leakage currents of the three segments of the sheath loop. Although the lengths of the three small segments of cable are different, the lengths of the three-phase cable must be consistent, so the phasor sum of the leakage current of the sheath system is zero. When there is no obvious aging of the main insulation of the cable, the relative change of the leakage current is ≤1%, so it can be selected as a reference value for judging whether there is a grounding fault in the sheath.

[0059] Similarly, for a double-sheath circuit, when a phase-to-phase short circuit occurs, the sum of the grounding current phasor differences of the two adjacent circuits is normal, but the grounding current is abnormal. It can be obtained that:

[0060]

[0061] Where I R is the short-circuit current, from which the phase-to-phase short-circuit fault of the double-sheath circuit can be distinguished.

[0062] 3) Current feature extraction

[0063] Based on the analysis of the fault mechanism of power cables, a cross-connected high-voltage cable fault simulation model was established using Matlab, and the ground current signal characteristics under short-circuit faults with different fault impedances and different load currents were studied. This provides important data support for the subsequent introduction of machine learning for fault detection.

[0064] Table 1 Matlab parameter settings

[0065] Parameter / Unit Numeric Parameter / Unit Numeric AB phase distance / m 0.27 Working voltage / kV 110 BC phase spacing / m 0.27 <![CDATA[Resistance per unit length of sheath / Ω·m -1 > <![CDATA[4.26×10 -5 ]]> AC phase spacing / m 0.54 Insulation layer volume resistivity / Ω·m <![CDATA[10 15 ]]> Ground resistance / Ω 0.50 Length of the first section / m 425 Metal conductor outer diameter / mm 35.90 Length of the second section / m 477 Insulation layer thickness / mm 17.30 Length of the third section / m 536 Metal sheath thickness / mm 2 Free space dielectric constant <![CDATA[8.85×10 -12 ]]> Frequency / Hz 50 Relative dielectric constant 2.30 Load current 300-500 Fault Impedance 1-100

[0066] Considering the changing law of grounding current under different faults, load status, grounding resistance change and other factors, two types of features are extracted: the grounding current at six grounding points and the phasor difference of adjacent sheath grounding currents. The phasor difference of adjacent sheath grounding currents can find out the sheath phase-to-phase faults. Since the current characteristics have no directionality, the amplitude and phase angle of the grounding current phasor are introduced as a common judgment basis to improve accuracy.

[0067] 4) Machine Learning Methods

[0068] Due to the large number of actual interference factors, the use of machine learning methods has strong adaptability and can be continuously optimized with the input of new data to maintain efficient fault identification capabilities. And combined with real-time monitoring, machine learning can provide timely warnings to help managers take quick measures to reduce maintenance costs and improve the reliability and safety of the power system. This patent transforms rule knowledge into feature variables and uses it as a supplement to the feature space of the original training data set, and then retrains the model using the augmented samples, that is, introducing the phase difference of adjacent sheath currents, which significantly improves the accuracy compared to using only the sheath grounding current.

[0069] This patent uses support vector machine (SVM) for fault classification. Simulation shows that when looking for the specific fault location, the regularity of the grounding current and its phasor difference is very weak. Therefore, it is necessary to use the concept of nonlinear SVM to determine the classification hyperplane in high-dimensional space.

[0070] like Figure 4 Here, the new feature (phase difference of adjacent sheath current) is compared with the traditional criterion (sheath grounding current). First, the correlation between the feature variables and the results needs to be evaluated. In the table, 1-4 represents the grounding current of the two circuits, 5-6 represents the phase difference of adjacent sheath current, the white bar graph is based only on the correlation result of the grounding current, and the black curve represents the addition of features 5-6 to the above basis. The total score of the correlation score is 10, and the results show that the new feature plays a decisive role.

[0071] 5) Determine the fault type

[0072] like Figure 5 As shown in the figure, the fault type, fault time, fault location, cable parameters, etc. can be set through the Matlab simulation model. A large number of sheath current signal samples of different fault types are generated based on the simulation model. Based on the characteristic of the phasor difference of adjacent sheath currents, the fault classification is carried out in combination with the support vector machine (SVM). The average classification accuracy of the diagnosis of six types of power cable phase-to-phase short-circuit faults can reach 99.5%, and the model prediction accuracy is improved by 7.5% compared with the model trained only with the original data.

[0073] like Figure 6 As shown, the t-SNE (t-distributed stochastic neighbor embedding) algorithm is applied to map the original data features and the output of the SVM algorithm to a two-dimensional plane, as shown in Figure 6 As shown, it can be found that except for extremely similar fault conditions, all fault conditions can be effectively distinguished, which illustrates the feasibility of the double-sheath circuit phase-to-phase short-circuit defect detection method based on the adjacent sheath current phasor difference.

[0074] In the present invention, the fault diagnosis process is as follows:

[0075] (1) If the grounding currents at the four grounding points of the cable sheath are abnormal, while the sum of the phase differences of the currents in adjacent sheaths is normal, then there is a phase-to-phase short circuit fault in the cable sheath; otherwise, it is another situation.

[0076] (2) Use the trained model to classify the newly collected current data and determine the fault location.

[0077] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting short-circuit defects between phases of a double-sheath circuit based on the phase difference of adjacent sheath currents, characterized in that: The specific steps include: Step 1: construct a circuit network model of the high-voltage cable to identify different types of ground fault mechanisms; Step 2: obtain the sheath grounding current data through simulation, extract its features, and obtain the current amplitude and phase; Step 3: Use appropriate machine learning algorithms to train and test the extracted current features, and improve the accuracy of fault detection by building a classification model; Step 4: Use the trained model to classify the newly collected current data and determine the fault type, so that corresponding maintenance measures can be taken in time to ensure the safe and stable operation of the cable.

2. According to claim 1, a double-sheath circuit interphase short circuit defect detection method based on adjacent sheath current phasor difference is characterized in that: The circuit network model of the high-voltage cable is specifically constructed by establishing a three-phase nine-segment cross-interconnected high-voltage cable line model, wherein the nine metal sheath segments of the cable sheath of the high-voltage cable line model are represented by A1, A2, A3, B1, B2, B3, C1, C2, and C3, wherein the cross-interconnection mode is "n", and the metal sheath loops 1, 2, and 3 are represented by yellow, green, and red, respectively; is the core current of the three-phase cable; the direct grounding boxes on both sides are G1 and G2, and the cross-connected grounding boxes in the middle are J1 and J2. The sheath grounding current measured by the current transformer at the beginning and end of the installed cable line.

3. The method for detecting interphase short circuit defects in a double-sheath circuit based on the phase difference of adjacent sheath currents according to claim 2 is characterized in that: The mechanism for identifying different types of grounding faults is specifically that the sheath current of the cross-connected high-voltage cable is a superposition of the leakage current component and the inductive coupling current component. When the double-sheath circuit fails, the sheath grounding current of the two circuits will be abnormal, which may be caused by the breakdown of the epoxy preform and the epoxy board causing the double-sheath circuit phase short circuit. When there is no fault, the current flowing into or out of the sheath system should be zero. According to the KCL law: In the formula They are the sum of the leakage currents of loops 1, 2, and 3 respectively. Under normal circumstances, the phasor difference of the grounding current at both ends of the sheath is the phasor sum of the three leakage currents of the sheath loop. The relative change of the leakage current is ≤1%, which is used as a reference value for judging whether there is a grounding fault in the sheath. For a double-sheath circuit, when a phase-to-phase short circuit occurs, the sum of the grounding current phasor differences of the two adjacent circuits is normal, but the grounding current is abnormal, so: Where I R is the short-circuit current, from which the phase-to-phase short-circuit fault of the double-sheath circuit can be distinguished.

4. The method for detecting interphase short circuit defects in a double sheath circuit based on the phase difference of adjacent sheath currents according to claim 1 is characterized in that: The simulation obtains the sheath grounding current data, and extracts its features to obtain the current amplitude and phase, specifically, Considering the changing law of grounding current under different faults, load status and grounding resistance change factors, two types of features are extracted: the grounding current at six grounding points and the phasor difference of adjacent sheath grounding currents. The phasor difference of adjacent sheath grounding currents can find out the sheath phase-to-phase faults. Since the current characteristics have no directionality, the amplitude and phase angle of the grounding current phasor are introduced as a common judgment basis.

5. The method for detecting interphase short circuit defects in a double sheath circuit based on the phase difference of adjacent sheath currents according to claim 1 is characterized in that: Aiming at the typical faults of high-voltage cable sheath, a simulation model of cross-connection of high-voltage cable sheath is built, and fault data is generated by simulation. The prepared data set consists of multiple operation matrices with a depth equal to the number of input data channels, and the data is randomly shuffled. The data is divided into training set and test set, and data preprocessing is performed. The characteristic parameters are normalized by using the premnmx normalization function, and its calculation expression is as follows: In the formula, X is the normalized value, x is the original data, min is the minimum value of the characteristic parameter, and max is the maximum value. Four types of phase difference characteristic parameters are extracted from the signal of the defective cable. In addition, there is also the mean square error. The calculation expression and explanation are as follows: The mean square error represents the arithmetic square root of the variance of the temperature data. Its physical meaning reflects the degree of discreteness of the data. After training, the model is evaluated by making predictions on the test set. Common evaluation indicators include accuracy, error curve, feature importance, and confusion matrix.

6. The method for detecting interphase short circuit defects in a double sheath circuit based on the phase difference of adjacent sheath currents according to claim 1 is characterized in that: Appropriate machine learning algorithms are used to train and test the extracted current features, and the accuracy of fault detection is improved by building a classification model. Specifically, the specific steps of cable defect recognition and classification based on SVM are as follows: first, the sheath current is extracted through the simulation model, and then 4 types of grounding current phasor difference data are extracted and normalized. The grounding current phasor difference data are divided into a test set and a training set in a ratio of 3:1, and the defect classification is completed through the selection of classifiers, kernel functions and penalty factors. Among the six types of samples, each type of sample constitutes a binary classification problem with other types of samples. A total of 15 SVMs are constructed, and the test samples classify and judge all constructed SVMs. The trained model can be used to classify and distinguish new data, and the category of unknown samples is predicted through the voting mechanism to achieve efficient and accurate classification and discrimination of sheath faults.

Images