Fault location method for transmission lines based on double-terminal traveling wave data

Abstract

The present invention relates to the technical field of power transmission line fault detection, and specifically to a power transmission line fault ranging method based on double-end traveling wave data; the method comprises: collecting historical traveling wave data, the historical traveling wave data includes traveling waves of multiple transmission lines; calculating the time-frequency relationship of each traveling wave to obtain a frequency-time series, and obtaining a companion matrix of the frequency-time series; constructing an importance matrix, the importance matrix is ​​composed of the classification importance of different frequency pairs; using the companion matrix of each traveling wave and the importance matrix as inputs of the constructed classification model, and the corresponding fault type as a label to train the classification model; using the trained classification model to identify the fault of the current transmission line to obtain the fault type; and calculating the fault position of the current transmission line according to the double-end traveling wave ranging method. That is, the scheme of the present invention can accurately identify the fault of the transmission line.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission line fault detection, and more specifically, to a power transmission line fault location method based on double-terminal traveling wave data. Background Art

[0002] Transmission line fault location technology is a crucial issue in power systems, especially in high-voltage and ultra-high-voltage power grids. With the continuous increase in power load and the expansion of power grid scale, the frequency and types of transmission line faults are becoming increasingly complex.

[0003] Traditional fault location methods face many challenges, such as low measurement accuracy, difficulty in fault type identification, slow system response speed, etc. Therefore, it is particularly important to study more accurate and efficient fault location technology. The accuracy of the current traveling wave data ranging method mainly depends on the recognition accuracy of the traveling wave, which performs frequency analysis through wavelet transform, but the current method does not take into account the impact of the number of occurrences of different frequencies after wavelet transform on fault identification, that is, there is a problem of error in fault identification.

[0004] Based on the above problems, in the related technology, such as the patent document with the authorization announcement number CN102129013B, a distribution network fault distance measurement method using natural frequency and artificial neural network is disclosed. It uses the principle that when the line fails at different points, the main natural frequency of the fault traveling wave of the line is different to perform fault distance measurement, and uses a hierarchical distributed artificial neural network for data processing. When a single-phase grounding fault occurs in the line, the sampling frequency is 1MHz, and a discrete zero-sequence transient signal with a sampling length of 2048 after the fault is selected, and an FFT transformation is performed. The transformation result is a 2048×2 matrix, and the absolute value of the matrix is ​​calculated. According to the model adopted by the present invention, 8 frequency values ​​with larger amplitudes are selected as sample attributes of the neural network, a suitable transfer function and learning rule are selected, and appropriate neural network parameters are set to construct a BP network model for fault distance measurement.

[0005] The above scheme selects and replicates larger frequency values ​​as the input of the network to train the network and finally obtain the fault location result. However, it should be noted that it selects some frequency values ​​and ignores the influence of other frequency values, and also ignores the influence of the number of occurrences of different frequencies. Therefore, the above scheme has the problem of inaccurate fault identification of the transmission line.

[0006] Therefore, how to improve the recognition accuracy of traveling wave data is particularly important. Summary of the invention

[0007] The purpose of the present invention is to propose a transmission line fault distance measurement method based on double-end traveling wave data, so as to solve the problem that the prior art cannot accurately identify the influence of different frequencies on faults, and thus cannot accurately identify faults; to this end, the present invention provides a solution in one of the following aspects.

[0008] The present invention provides a transmission line fault location method based on double-terminal traveling wave data, comprising:

[0009] Collecting historical traveling wave data, wherein the historical traveling wave data includes traveling waves of a plurality of transmission lines;

[0010] The time-frequency relationship of each traveling wave is obtained by wavelet transform, a frequency time series is obtained, and a companion matrix of the frequency time series is obtained; the element value of the companion matrix is ​​the number of occurrences of a frequency pair consisting of two adjacent frequencies in the frequency time series;

[0011] Constructing an importance matrix, the importance matrix consisting of classification importances of different frequency pairs; the classification importances characterize differences between the same frequency pairs in different fault types;

[0012] The companion matrix and importance matrix of each traveling wave are used as the input of the constructed classification model, and the corresponding fault type is used as the label to train the classification model;

[0013] The trained classification model is used to identify the fault of the current transmission line and obtain the fault type; and the fault location of the current transmission line is calculated based on the double-terminal traveling wave ranging method.

[0014] In the above scheme, by analyzing the time-frequency relationship of each traveling wave in the collected historical traveling wave data, the frequency of the traveling wave at different times can be obtained, and then the frequency time series that can characterize the traveling wave can be obtained, and by analyzing the frequency time series of each traveling wave, the number of frequency pairs of different adjacent two frequency group leaders can be obtained to obtain the companion matrix that characterizes the traveling wave characteristics, and then by analyzing the differences between different frequency pairs, the importance matrix is ​​determined to form a training data set for training the classification model, and then the classification model is used to identify the fault of the current transmission line. That is, the scheme of the present invention can establish the relationship between the frequency of the traveling wave and the fault type through a comprehensive analysis of the traveling waves and fault types of different transmission lines, and then through the relationship between the fault type and the frequency, the analysis of the traveling wave of the current transmission line is realized to determine the corresponding fault type.

[0015] Optionally, the classification importance is:

[0016] ;

[0017] in, Represents the frequency pair in the companion matrix The importance of classification, represents normalization, Indicates Frequency pair of type fault The mean number of occurrences, Indicates Frequency pair of type fault The number of occurrences, is the total number of fault types.

[0018] In the above scheme, by obtaining the difference between the same frequency pair and different fault types, the importance of the frequency pair under different fault types can be determined.

[0019] Optionally, the classification importance is:

[0020] ;

[0021] in, Indicates frequency pair The importance of classification, represents normalization, Indicates The frequency of all traveling waves in this type of fault is The variance of the number of occurrences, Represents the frequency pairs of all traveling waves in the historical traveling wave data The variance of the number of occurrences, is the total number of fault types.

[0022] In the above scheme, by obtaining the fluctuation changes between the same frequency pairs in different fault types and the fluctuation characteristics between the same frequency pairs in all fault types, the importance of the frequency to the fault type can be determined.

[0023] Optionally, the average number of times is:

[0024] ;

[0025] in, Indicates Frequency pair of type fault The mean number of occurrences, Indicates that the label in the historical traveling wave data is The number of traveling waves of the type fault, Indicates Type of fault The frequency pair of the traveling wave Number of occurrences.

[0026] The calculation of the number mean in the above scheme can simplify the complexity.

[0027] Optionally, the method further includes: denoising each traveling wave in the acquired historical traveling wave data of the power transmission line to obtain a denoised traveling wave.

[0028] In the above scheme, the accuracy of historical traveling wave data is improved by removing noise from the data.

[0029] Optionally, the method further includes: sending the acquired fault type and fault location of the current power transmission line to a worker to remind the worker to perform maintenance.

[0030] Optionally, the specific process of training the classification model is:

[0031] The companion matrix and importance matrix of each historical traveling wave data are input into the classification model, and the output is the fault label corresponding to the historical traveling wave data. The cross entropy loss is calculated, and the model parameters are updated using the gradient descent algorithm. When the training reaches the maximum number of iterations or the model loss is less than the set threshold, the model training stops.

[0032] Optionally, the classification model is a convolutional neural network model.

[0033] The beneficial effects of the present invention are:

[0034] The scheme of the present invention analyzes the frequency of different types of fault traveling wave data to obtain the importance of different frequency pairs to fault classification, so as to train the classification model. It can deeply analyze the relationship between different frequency pairs and fault types, improve the accuracy of the trained classification model, and thus improve subsequent fault identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0036] Figure 1 A flowchart of the steps of the power transmission line fault location method based on double-terminal traveling wave data in this embodiment is schematically shown;

[0037] Figure 2 The flowchart of the steps of acquiring a training data set in the method for fault location of a power transmission line based on double-terminal traveling wave data in this embodiment is schematically shown. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0039] Taking a high-voltage transmission line of a certain substation as an example, the transmission line fault location method based on double-terminal traveling wave data of the present invention is introduced.

[0040] Specifically, Figure 1 As shown, the transmission line fault location method based on double-terminal traveling wave data in this embodiment includes the following steps:

[0041] Step S1, obtaining historical traveling wave data to construct a training data set.

[0042] In this embodiment, the traveling waves of each transmission line are obtained, and multiple traveling waves constitute the historical traveling wave data. It should be noted that the above-mentioned transmission lines can be all high-voltage transmission lines of a substation, or part of the transmission lines in a substation. In this embodiment, the historical traveling waves on these transmission lines are obtained.

[0043] In one embodiment, each traveling wave in the acquired historical traveling wave data of the power transmission line is further denoised to obtain a denoised traveling wave.

[0044] like Figure 2 As shown, the process of acquiring the training data set in this embodiment is:

[0045] Step S11, using wavelet transform to transform each traveling wave in the historical traveling wave data, to obtain the frequency time series of each traveling wave, and to obtain the companion matrix of the frequency time series.

[0046] In this embodiment, the law of frequency variation over time in the traveling wave is firstly obtained by wavelet transformation to obtain a frequency time series.

[0047] Wavelet transform is a transform analysis method used to decompose a signal into components of different scales and provide a "time-frequency" window that changes with frequency. Therefore, since wavelet transform is an existing technology, its process will not be described in detail here.

[0048] Among them, the process of obtaining the companion matrix of the frequency time series is:

[0049] The number of occurrences of frequency pairs consisting of two adjacent frequencies in the frequency time series is counted. After counting the number of occurrences of each frequency pair, the companion matrix of the frequency time series is obtained, where the element value in the xth row and yth column of the companion matrix represents the number of occurrences of the frequency pair in the frequency time series. Therefore, the companion matrix is ​​composed of the number of occurrences of all frequency pairs.

[0050] The purpose of acquiring the frequency pair is that two adjacent frequencies can reflect the temporal sequence, that is, they can represent the temporal characteristics.

[0051] For example, when the frequency time series of a traveling wave is (3, 4, 5, 6, 4, 5), the corresponding data at different moments in the frequency time series are frequency values. When the frequency pair is (3, 4), the number of its occurrences is 1, when the frequency pair is (4, 5), the number of its occurrences is 2, and when the frequency pair is (5, 6), the number of its occurrences is 1. Therefore, the number of occurrences of the frequency pairs composed of each adjacent two frequencies can be obtained, and then the corresponding companion matrix is ​​constructed according to the different frequency numbers, specifically: ; When obtaining the companion matrix, the same frequencies must be merged first, and then the companion matrix is ​​generated.

[0052] At this point, the companion matrix of each traveling wave can be obtained.

[0053] In this embodiment, the effectiveness of determining the fault location of a transmission line by using historical traveling wave data mainly depends on the accuracy of the historical traveling wave data identification. The differences in traveling waves of different faults are mainly reflected in the waveform (frequency). By analyzing the relationship between the frequency of each traveling wave in the historical traveling wave data and time, the fault type can be well identified.

[0054] Step S12: construct an importance matrix. Specifically, the elements in the importance matrix are the classification importance of each frequency pair.

[0055] In this embodiment, after calculating the companion matrix of the frequency time series of each traveling wave in the historical traveling wave data, the classification importance of different frequency pairs can be obtained according to the difference between the companion matrices of traveling waves of different fault categories, and the importance matrix is ​​constructed based on the classification importance.

[0056] By calculating the differences between the traveling waves through the companion matrix of the frequency-time series, the differences between the traveling waves can be better extracted.

[0057] In one embodiment, the calculation formula for the classification importance of different frequency pairs is as follows:

[0058] ;

[0059] in, Represents the frequency pair in the companion matrix The importance of classification, represents normalization, Indicates Frequency pair of type fault The mean number of occurrences, Indicates Frequency pair of type fault The number of occurrences, is the total number of fault types.

[0060] It should be noted that the greater the difference in the number of occurrences of frequency pairs between different types of faults, the more different the performance of the frequency pairs on different types of faults, and it is easier to identify the fault through the difference of the frequency pairs under different fault types.

[0061] The calculation formula of the frequency mean is as follows:

[0062] ;

[0063] in, Indicates Frequency pair of type fault The mean number of occurrences, Indicates that the label in the historical traveling wave data is The number of traveling waves of the type fault, Indicates Type of fault The frequency pair of the traveling wave The number of occurrences, i.e. The element value of the xth row and yth column in the companion matrix of the traveling wave, where i and j correspond to frequency values ​​at different times, and x and y are the corresponding row and column numbers.

[0064] By calculating the above-mentioned companion matrix, the number of occurrences of each frequency pair in the historical traveling wave data can be obtained, and then the classification importance of each frequency pair can be obtained.

[0065] In another embodiment, the greater the classification importance of a frequency pair, the smaller the difference in the number of occurrences of the frequency pair in the traveling waves of the same type of fault, and the greater the difference in the number of occurrences of the frequency pair in the traveling waves of different types of faults. Therefore, the classification importance of each frequency pair can also be calculated based on the variance of the frequency pair between different fault types and the variance of the frequency pair in the same type of fault. The specific calculation formula is as follows: ;

[0066] in, Indicates frequency pair The importance of classification, represents normalization, Indicates The frequency of all traveling waves in this type of fault is The variance of the number of occurrences, Represents the frequency pairs of all traveling waves in the historical traveling wave data The variance of the number of occurrences, is the total number of fault types.

[0067] in, Characterize the frequency pairs of all traveling waves corresponding to all types of faults The variance of the number of occurrences is equivalent to the between-class variance, Characterization The frequency pairs of all traveling waves corresponding to this type of fault The variance of the number of occurrences is equivalent to the intra-class variance; then the larger the inter-class variance, the larger the intra-class variance. The smaller the value, the greater the difference in the number of occurrences of the frequency pair under different faults, and the greater the classification importance of the frequency pair.

[0068] Step S13, taking the matrix, importance matrix and label corresponding to each traveling wave in the historical traveling wave data as a training data set.

[0069] Specifically, the historical traveling wave data includes multiple traveling waves, each traveling wave corresponds to a companion matrix and an importance matrix; then the historical traveling wave data corresponds to a combination of multiple companion matrices and importance matrices.

[0070] The label in this embodiment is the fault type corresponding to the historical traveling wave data. The fault type is no fault or a specific fault type. The specific fault types are short circuit fault, disconnection fault, transient fault, permanent fault, other fault types, etc. Of course, short circuit faults can also be divided into more specific categories, such as single-phase grounding fault, two-phase short circuit fault, two-phase grounding fault and three-phase short circuit fault.

[0071] Step S2, training the constructed classification model.

[0072] In one embodiment, the classification model is a convolutional neural network model.

[0073] In this embodiment, the associated matrix and the importance matrix corresponding to each traveling wave in the historical traveling wave data are used as inputs of the constructed classification model, and the corresponding fault type is used as a label to train the classification model.

[0074] For example, taking the convolutional neural network model as an example, the input of the model is the companion matrix and importance matrix of each wave, and the output is the fault label corresponding to the wave. The loss function of the model is the cross entropy loss. During training, the gradient descent algorithm is used to update the model parameters. When the training reaches the maximum number of iterations or the model loss is less than the set threshold, the model stops training.

[0075] Step S3, using the trained classification model to identify the fault type of the current transmission line; and calculating the fault location of the current transmission line according to the double-terminal traveling wave ranging method.

[0076] In this embodiment, a trained classification model is used to identify whether the traveling wave on the current transmission line is caused by a fault. If it is caused by a fault, the specific fault type is output, and a two-terminal traveling wave ranging method is used to calculate the fault location of the current transmission line.

[0077] In the above, when making a fault judgment on the current traveling wave of the current transmission line, the number of occurrences of each frequency pair in the current traveling wave is first calculated to determine the companion matrix of the current traveling wave and the classification importance of each frequency pair in the companion matrix, and then the importance matrix is ​​determined. The companion matrix and the importance matrix of the current traveling wave are used as inputs of the trained classification model to identify the fault type of the traveling wave.

[0078] The importance matrix determined above is based on the frequency pairs in the matrix of the current traveling wave, and searches for the classification importance of the same frequency pairs in the historical traveling wave data, and composes the importance matrix of the current traveling wave with the found classification importances.

[0079] At the same time, the double-terminal traveling wave ranging method is also used to determine the fault location of the current transmission line; specifically, the fault location is calculated based on the time difference between the traveling wave reaching the two ends of the transmission line.

[0080] Since the double-terminal traveling wave ranging method is an existing technology, the specific process will not be described here.

[0081] In this embodiment, the fault type and fault location of the power transmission line are sent to relevant personnel to remind them to carry out timely maintenance of equipment related to the fault.

[0082] The scheme of the present invention analyzes the frequency of different types of fault traveling wave data to obtain the importance of different frequency pairs to fault classification, trains a classification model, takes the number of occurrences of frequency pairs in each traveling wave and the importance of the frequency pairs as input of the classification model, outputs the fault type of the traveling wave data, and then uses the double-end traveling wave ranging method to calculate the fault location, so as to accurately identify faults of transmission lines.

[0083] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0084] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A transmission line fault location method based on double-terminal traveling wave data, characterized in that: include: Collecting historical traveling wave data, wherein the historical traveling wave data includes traveling waves of a plurality of transmission lines; The time-frequency relationship of each traveling wave is obtained by wavelet transform, a frequency time series is obtained, and a companion matrix of the frequency time series is obtained; the element value of the companion matrix is ​​the number of occurrences of a frequency pair consisting of two adjacent frequencies in the frequency time series; Constructing an importance matrix, the importance matrix consisting of classification importances of different frequency pairs; the classification importances characterize differences between the same frequency pairs in different fault types; The companion matrix and importance matrix of each traveling wave are used as the input of the constructed classification model, and the corresponding fault type is used as the label to train the classification model; Use the trained classification model to identify the faults of the current transmission line and obtain the fault type; And the fault location of the current transmission line is calculated based on the double-terminal traveling wave ranging method.

2. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 1, characterized in that: The classification importance is: ； in, Represents the frequency pair in the companion matrix The importance of classification, represents normalization, Indicates Frequency pair of type fault The mean number of occurrences, Indicates Frequency pair of type fault The number of occurrences, is the total number of fault types.

3. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 1, characterized in that: The classification importance is: ； in, Indicates frequency pair The importance of classification, represents normalization, Indicates The frequency of all traveling waves in this type of fault is The variance of the number of occurrences, Represents the frequency pairs of all traveling waves in the historical traveling wave data The variance of the number of occurrences, is the total number of fault types.

4. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 2, characterized in that: The mean of the times is: ； in, Indicates Frequency pair of type fault The mean number of occurrences, Indicates that the label in the historical traveling wave data is The number of traveling waves of the type fault, Indicates Type of fault The frequency pair of the traveling wave Number of occurrences.

5. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 1, characterized in that: Also includes: Each traveling wave in the acquired historical traveling wave data of the transmission line is denoised to obtain a denoised traveling wave.

6. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 1, characterized in that: Also includes: The acquired fault type and fault location of the current transmission line are sent to the staff to remind them to carry out maintenance.

7. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 1, characterized in that: The specific process of training the classification model is as follows: The companion matrix and importance matrix of each historical traveling wave data are input into the classification model, and the output is the fault label corresponding to the historical traveling wave data. The cross entropy loss is calculated, and the model parameters are updated using the gradient descent algorithm. When the training reaches the maximum number of iterations or the model loss is less than the set threshold, the model training stops.

8. The method for fault location of a power transmission line based on double-terminal traveling wave data according to claim 7, characterized in that: The classification model is a convolutional neural network model.

Images