On-load tap changer status determination method, device, terminal and storage medium

Abstract

This invention provides a method, apparatus, terminal, and storage medium for determining the state of an on-load tap changer. The method includes: acquiring the main vibration waveform corresponding to the switching action of the on-load tap changer; extracting the time-frequency features of the main vibration waveform; and inputting the time-frequency features into a trained random forest classifier to obtain the state determination result of the on-load tap changer. This invention uses the time-frequency features of the vibration signal as the input value of the random forest classifier, which can accurately determine the switching state of the on-load tap changer.

Description

Technical Field

[0001] This invention relates to the field of on-load tap changer technology, and in particular to a method, apparatus, terminal and storage medium for determining the state of an on-load tap changer. Background Technology

[0002] On-load tap changers are widely used in power transformers, playing a crucial role in on-load voltage regulation. As the only operational component of a transformer, the on-load tap changer possesses complex mechanical and electrical characteristics. Faults in on-load tap changers are primarily mechanical, although electrical faults are often caused by mechanical issues. Therefore, to avoid serious consequences from on-load tap changer failures, it is essential to promptly detect mechanical faults in the on-load tap changer.

[0003] On-load tap changers are typically housed within transformer tanks and immersed in insulating oil. Non-invasive monitoring methods, such as vibration and sound signal analysis, are the most common monitoring techniques for on-load tap changers. Currently, monitoring of on-load tap changers usually involves analyzing the vibration signals generated during the switching action. Since the switching actions of on-load tap changers have a fixed timing sequence, combining the vibration signals generated during the switching action with the switching state of the on-load tap changer allows for more effective monitoring of mechanical faults. Vibration signals can reflect the switching state of the on-load tap changer to some extent. However, as a typical non-stationary signal, traditional time-domain and frequency-domain analysis methods struggle to fully extract the characteristics of vibration signals, making it difficult to accurately determine the switching state of the on-load tap changer. Summary of the Invention

[0004] This invention provides a method, apparatus, terminal, and storage medium for determining the state of an on-load tap changer, in order to solve the problem of accurately determining the switching state of an on-load tap changer.

[0005] In a first aspect, the present invention provides a method for determining the state of an on-load tap changer, comprising:

[0006] Obtain the main vibration waveform corresponding to the switching action of the on-load tap changer;

[0007] Extract the time-frequency characteristics of the main vibration waveform;

[0008] The time-frequency features are input into a trained random forest classifier to obtain the state judgment result of the on-load tap changer.

[0009] In one possible implementation, the method further includes, before inputting the time-frequency features into the trained random forest classifier:

[0010] Calculate the lag autocorrelation coefficients for each time-domain feature;

[0011] Based on the magnitude of each lagged autocorrelation coefficient, a subset of lagged autocorrelation coefficients are selected from each lagged autocorrelation coefficient.

[0012] Inputting time-frequency features into a trained random forest classifier includes:

[0013] The time-frequency features corresponding to each selected lag autocorrelation coefficient are input into the trained random forest classifier.

[0014] In one possible implementation, the main vibration waveform corresponding to the switching action of the on-load tap changer includes:

[0015] Acquire the current and vibration signals of the on-load tap changer;

[0016] The vibration signal is initially extracted based on the start and end times of the current signal;

[0017] Using the peak position of the initially intercepted vibration signal as the midpoint of the interception, a signal of a preset length is intercepted from the initially intercepted vibration signal to obtain the main vibration waveform corresponding to the switching action of the on-load tap changer.

[0018] In one possible implementation, the method further includes, before inputting the time-frequency features into the trained random forest classifier:

[0019] Establish an initial random forest classifier;

[0020] The initial random forest classifier is trained using a sample set to obtain a trained random forest classifier. The sample set includes multiple training samples, each of which is the main vibration waveform corresponding to the switching action of an on-load tap changer, and the label of each training sample is the switching state of the on-load tap changer corresponding to that training sample.

[0021] In one possible implementation, calculating the lag autocorrelation coefficients of each time-domain feature includes:

[0022] For each time-domain feature, the lag autocorrelation coefficient of that time-domain feature is calculated based on a preset formula; the formula for calculating the lag autocorrelation coefficient is as follows:

[0023]

[0024] Where n represents the length of the main vibration waveform, l represents the preset hysteresis value, and σ 2 X represents the variance of the amplitude of the principal vibration waveform, μ represents the mean of the amplitude of the principal vibration waveform, and X represents the variance of the amplitude of the principal vibration waveform. t X represents the value of this time-domain feature at time t. t-l This represents the value of the time-domain feature at time tl.

[0025] In one possible implementation, selecting a subset of lagged autocorrelation coefficients based on their magnitudes includes:

[0026] Sort the lagged autocorrelation coefficients by size;

[0027] Select a preset number of lag autocorrelation coefficients from largest to smallest.

[0028] In one possible implementation, the base classifier for the random forest classifier is a C4.5 decision tree.

[0029] Secondly, the present invention provides a state determination device for an on-load tap changer, comprising:

[0030] The acquisition module is used to acquire the main vibration waveform corresponding to the switching action of the on-load tap changer;

[0031] The extraction module is used to extract the time-frequency characteristics of the main vibration waveform;

[0032] The judgment module is used to input the time-frequency features into a trained random forest classifier to obtain the state judgment result of the on-load tap changer.

[0033] Thirdly, the present invention provides a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method shown in the first aspect above or any possible implementation of the first aspect.

[0034] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method shown in the first aspect or any possible implementation thereof.

[0035] This invention provides a method, apparatus, terminal, and storage medium for determining the state of an on-load tap changer. The method includes: acquiring the main vibration waveform corresponding to the switching action of the on-load tap changer; extracting the time-frequency features of the main vibration waveform; and inputting the time-frequency features into a trained random forest classifier to obtain the state determination result of the on-load tap changer. This invention uses the time-frequency features of the vibration signal as the input value of the random forest classifier, which can accurately determine the switching state of the on-load tap changer. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating the implementation of the on-load tap changer status determination method provided in this embodiment of the invention.

[0038] Figure 2 This is a schematic diagram of the structure of the on-load tap changer status determination device provided in an embodiment of the present invention;

[0039] Figure 3 This is a waveform diagram of the vibration signal provided in an embodiment of the present invention;

[0040] Figure 4 This is a waveform diagram of the vibration signal after being extracted, provided in an embodiment of the present invention;

[0041] Figure 5 This is the AUC diagram of the classification results of the random forest classifier provided in this embodiment of the invention;

[0042] Figure 6 This is a schematic diagram of the terminal provided in an embodiment of the present invention. Detailed Implementation

[0043] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0044] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0045] See Figure 1 The flowchart illustrating the implementation of the on-load tap changer state determination method provided in this embodiment of the invention is described in detail below:

[0046] Step 101: Obtain the main vibration waveform corresponding to the switching action of the on-load tap changer.

[0047] In this embodiment, the terminal can acquire the vibration signal generated by the switching action of the on-load tap changer collected by the accelerometer, and then process the vibration signal to obtain the main vibration waveform corresponding to the switching action. The main vibration waveform contains the vibration characteristics of the on-load tap changer switching action.

[0048] Step 102: Extract the time-frequency characteristics of the main vibration waveform.

[0049] In this embodiment, the time-frequency characteristics of the main vibration waveform include both time-domain and frequency-domain characteristics. The time-domain characteristics may include the sum, mean, standard deviation, and variance of the amplitudes, while the frequency-domain characteristics may include the real part, imaginary part, absolute value, angle, energy, and correlation coefficient after wavelet transform of the Fourier coefficients under different coefficients.

[0050] Step 103: Input the time-frequency features into the trained random forest classifier to obtain the state judgment result of the on-load tap changer.

[0051] In this embodiment, the status judgment result of the on-load tap changer includes switching from an odd-numbered range to an even-numbered range and switching from an even-numbered range to an odd-numbered range.

[0052] In this embodiment, a random forest classifier is used to determine the state of the on-load tap changer. The random forest classifier adopts a parallel ensemble learning approach, where the base learners are independent of each other and have no dependencies. The output matrix of the base learners is as follows:

[0053]

[0054] in Let x be the probability that the i-th base classifier predicts that sample x belongs to the N-th class.

[0055] Random forest classifiers use a voting method for classification, which is fast to train and less prone to overfitting. The voting method works as follows: each sample is classified using different base learners, and the classification results are tallied. If a classification result receives more than 50% of the votes, that result is output.

[0056] In some embodiments, prior to step 103, the method further includes:

[0057] Step 104: Calculate the lag autocorrelation coefficients for each time-domain feature;

[0058] Step 105: Select a portion of the lagged autocorrelation coefficients from the lagged autocorrelation coefficients based on the magnitude of each lagged autocorrelation coefficient;

[0059] Step 103 includes:

[0060] The time-frequency features corresponding to each selected lag autocorrelation coefficient are input into the trained random forest classifier.

[0061] In this embodiment, the lag autocorrelation coefficient of a certain time-domain feature can reflect the correlation between that time-domain feature and the classification result. The larger the lag autocorrelation coefficient, the stronger the correlation between the time-domain feature and the classification result. Based on the magnitude of each lag autocorrelation coefficient, a portion of the lag autocorrelation coefficients can be selected to identify some time-frequency features with strong correlation to the classification result, reducing redundant input and thus reducing the possibility of overfitting in the random forest classifier during the classification process.

[0062] Furthermore, the time-frequency features in this embodiment can also be input into a random forest classifier in matrix form for classification.

[0063] In some embodiments, step 101 includes:

[0064] Step 1011: Obtain the current signal and vibration signal of the on-load tap changer;

[0065] Step 1012: Perform preliminary segmentation of the vibration signal based on the start and end times of the current signal;

[0066] Step 1013: Using the peak position of the initially intercepted vibration signal as the interception midpoint, a signal of a preset length is intercepted from the initially intercepted vibration signal to obtain the main vibration waveform corresponding to the switching action of the on-load tap changer.

[0067] In this embodiment, the current signal of the on-load tap changer can be acquired by a current sensor, and the vibration signal of the on-load tap changer can be acquired by an accelerometer. The accelerometer can be placed on the top cover of the on-load tap changer.

[0068] In this embodiment, the vibration signal is initially extracted based on the start and end times of the current signal, which can filter out vibration signals generated by the on-load tap changer outside of its operating time. After the initial extraction, the peak value of the vibration signal within the operating time is used as the midpoint of the extraction, which can extract the part of the vibration signal containing the most features. The preset length can be determined experimentally, typically ranging from 120ms to 250ms.

[0069] In this embodiment, multiple vibration signals can also be processed in batches. The specific steps are as follows:

[0070] Each vibration signal is initially extracted based on the start and end times of the current signal;

[0071] Align the peak positions of the initially extracted vibration signals on the time axis;

[0072] Using the peak position of each vibration signal after preliminary interception as the interception midpoint, a preset length of signal is intercepted from each vibration signal after preliminary interception to obtain multiple main vibration waveforms corresponding to the switching action of the on-load tap changer.

[0073] Correspondingly, after processing multiple vibration signals, the main vibration waveforms are input into the random forest classifier. The classification results of the random forest classifier can be verified according to the temporal relationship between the main vibration signals, thereby improving the classification accuracy.

[0074] In some embodiments, prior to step 103, the method further includes:

[0075] Step 106: Establish the initial random forest classifier;

[0076] Step 107: Use the sample set to train the initial random forest classifier to obtain the trained random forest classifier; the sample set includes multiple training samples, each training sample is the main vibration waveform corresponding to the switching action of the on-load tap changer, and the label of each training sample is the switching state of the on-load tap changer corresponding to that training sample.

[0077] In this embodiment, each training sample in the sample set represents the main vibration signal of an on-load tap changer under different switching states. When training the initial random forest classifier using the sample set, the sample set can be divided into a training set and a test set. First, the training samples in the training set are used to train the random forest classifier. Then, the training samples in the test set are input into the random forest classifier to test the training results. Notably, the training samples do not require normalization before being input into the random forest classifier.

[0078] In some embodiments, step 104 includes:

[0079] For each time-domain feature, the lag autocorrelation coefficient of that time-domain feature is calculated based on a preset formula; the formula for calculating the lag autocorrelation coefficient is as follows:

[0080]

[0081] Where n represents the length of the main vibration waveform, l represents the preset hysteresis value, and σ 2 X represents the variance of the amplitude of the principal vibration waveform, μ represents the mean of the amplitude of the principal vibration waveform, and X represents the variance of the amplitude of the principal vibration waveform. t X represents the value of this time-domain feature at time t. t-l This represents the value of the time-domain feature at time tl.

[0082] In this embodiment, the preset lag value can be set according to actual conditions. The autocorrelation coefficient here is a function of the similarity between two observations and the time difference between them. The autocorrelation coefficient measures the degree of correlation between the same event at two different times; figuratively speaking, it measures the impact of one's past behavior on one's present.

[0083] In some embodiments, step 105 includes:

[0084] Sort the lagged autocorrelation coefficients by size;

[0085] Select a preset number of lag autocorrelation coefficients from largest to smallest.

[0086] In this embodiment, step 105 can also involve selecting a hysteresis autocorrelation coefficient greater than a preset threshold. The preset quantity and preset threshold in this embodiment can be determined experimentally.

[0087] In some embodiments, the base classifier of the random forest classifier is a C4.5 decision tree.

[0088] In this embodiment, the decision tree is divided based on the information gain ratio, and the formula for calculating the information gain ratio is:

[0089]

[0090] Where Gain(D,a) represents information gain, and the calculation formula is:

[0091]

[0092] Among them, the empirical entropy of the Ent(D) dataset D and the original entropy values ​​of the dataset for different classification results are included. Ent(D) represents the empirical conditional entropy of dataset D under different conditions for the current features. v ) represents the entropy of different cases under the current feature, and v represents the number of different cases.

[0093] Where IV(a) is the fixed gain value of the node and the entropy value of the current feature itself.

[0094] In one specific embodiment, the steps of training a random forest classifier and using the trained random forest classifier to classify vibration signals include:

[0095] An accelerometer was installed on the top cover of the on-load tap changer to collect vibration signals during tap changer switching. A total of 180 vibration signals were collected. The sum of these vibration signals is shown below. Figure 3 As shown. Referring to the internal structure of the on-load tap changer, odd-numbered taps for upshifting or downshifting (1-2, 3-2) are classified into one category, and even-numbered taps for upshifting or downshifting (2-3, 4-3) are classified into another category.

[0096] Based on the start and end times of the current signal, each vibration signal is initially truncated. The peak positions of each initially truncated vibration signal are aligned on the time axis, and the peak positions are used as the midpoints for truncating. After extracting signals of a preset length from each initially truncated vibration signal, 180 sets of main vibration waveforms are obtained. These waveforms are then superimposed as follows: Figure 4 As shown.

[0097] Ninety sets out of 180 main vibration waveforms were used as the training set, and 90 sets of vibration signals were used as the test set to train the initial random forest classifier. Both the training and test sets contained 45 sets of vibration signals with odd-to-even states and 45 sets with even-to-odd states.

[0098] After training the initial random forest classifier using the main vibration waveforms in the training set, time-frequency features are extracted from each main vibration waveform in the test set. The Fourier coefficient angle value, hysteresis autocorrelation coefficient, real part of the Fourier coefficient, imaginary part of the Fourier coefficient, and energy of each main vibration waveform are calculated respectively, resulting in a set of feature vectors for the vibration signals from odd to even gears and from even to odd gears, as shown in Table 1.

[0099] Table 1

[0100]

[0101] The trained random forest classifier was tested using the feature vectors in Table 1, and the results are as follows: Figure 5 The Receiver Operating Characteristic Curve (ROC) plot shown below uses the Area Under Curve (AUC) as an example. The horizontal axis represents the proportion of samples incorrectly predicted as positive (FPR), while the vertical axis represents the proportion of samples correctly predicted as positive (TPR). The closer the ROC curve is to the (0,1) point, the better the prediction performance. Simultaneously, the area under the curve and its diagonal will also be larger; this area is the AUC.

[0102] Table 2 shows the confusion matrix of the classification results obtained by testing the trained random forest classifier. As shown in Table 2, the trained random forest classifier achieved a precision of 97.7% on the test set. Precision indicates the percentage of truly positive data points among those predicted as positive. Recall in Table 1 represents the percentage of all positive data points successfully predicted as positive. The F1 score in Table 1 is a combination of precision and recall, compensating for the shortcomings of either alone. F1 score = 2 * Precision * Recall / (Precision + Recall). This result indicates that the results of on-load tap changer switching can be divided into two categories: odd-numbered taps – even-numbered taps, and even-numbered taps – odd-numbered taps.

[0103] Table 2

[0104] Classification results Precision recall f1-score support Odd number - Even number 0.98 0.98 0.98 46 Even number - Odd number 0.98 0.98 0.98 44

[0105] The on-load tap changer state determination method provided in this embodiment of the invention includes: acquiring the main vibration waveform corresponding to the switching action of the on-load tap changer; extracting the time-frequency features of the main vibration waveform; and inputting the time-frequency features into a trained random forest classifier to obtain the on-load tap changer state determination result. This embodiment of the invention uses the time-frequency features of the vibration signal as the input value of the random forest classifier, which can accurately determine the switching state of the on-load tap changer.

[0106] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0107] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0108] Figure 2 A schematic diagram of the on-load tap changer status determination device provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0109] like Figure 2 As shown, the on-load tap changer status determination device 2 includes:

[0110] The acquisition module 21 is used to acquire the main vibration waveform corresponding to the switching action of the on-load tap changer;

[0111] Extraction module 22 is used to extract the time-frequency characteristics of the main vibration waveform;

[0112] The judgment module 23 is used to input the time-frequency features into the trained random forest classifier to obtain the state judgment result of the on-load tap changer.

[0113] In some embodiments, the extraction module 22 is specifically used for:

[0114] Before inputting the time-frequency features into the trained random forest classifier, calculate the lag autocorrelation coefficients of each time-domain feature;

[0115] Based on the magnitude of each lagged autocorrelation coefficient, a subset of lagged autocorrelation coefficients are selected from each lagged autocorrelation coefficient.

[0116] Module 23 is specifically used for:

[0117] The time-frequency features corresponding to each selected lag autocorrelation coefficient are input into the trained random forest classifier.

[0118] In some embodiments, the acquisition module 21 is specifically used for:

[0119] Acquire the current and vibration signals of the on-load tap changer;

[0120] The vibration signal is initially extracted based on the start and end times of the current signal;

[0121] Using the peak position of the initially intercepted vibration signal as the midpoint of the interception, a signal of a preset length is intercepted from the initially intercepted vibration signal to obtain the main vibration waveform corresponding to the switching action of the on-load tap changer.

[0122] In some embodiments, the determination module 23 is specifically used for:

[0123] Before inputting the time-frequency features into the trained random forest classifier, an initial random forest classifier is built.

[0124] The initial random forest classifier is trained using a sample set to obtain a trained random forest classifier. The sample set includes multiple training samples, each of which is the main vibration waveform corresponding to the switching action of an on-load tap changer, and the label of each training sample is the switching state of the on-load tap changer corresponding to that training sample.

[0125] In some embodiments, the extraction module 22 is specifically used for:

[0126] For each time-domain feature, the lag autocorrelation coefficient of that time-domain feature is calculated based on a preset formula; the formula for calculating the lag autocorrelation coefficient is as follows:

[0127]

[0128] Where n represents the length of the main vibration waveform, l represents the preset hysteresis value, and σ2 X represents the variance of the amplitude of the principal vibration waveform, μ represents the mean of the amplitude of the principal vibration waveform, and X represents the variance of the amplitude of the principal vibration waveform. t X represents the value of this time-domain feature at time t. t-l This represents the value of the time-domain feature at time tl.

[0129] In some embodiments, the extraction module 22 is specifically used for:

[0130] Sort the lagged autocorrelation coefficients by size;

[0131] Select a preset number of lag autocorrelation coefficients from largest to smallest.

[0132] In some embodiments, the base classifier of the random forest classifier is a C4.5 decision tree.

[0133] The on-load tap changer state determination device provided in this embodiment of the invention includes: an acquisition module for acquiring the main vibration waveform corresponding to the switching action of the on-load tap changer; an extraction module for extracting the time-frequency features of the main vibration waveform; and a determination module for inputting the time-frequency features into a trained random forest classifier to obtain the state determination result of the on-load tap changer. This embodiment of the invention uses the time-frequency features of the vibration signal as the input value of the random forest classifier, which can accurately determine the switching state of the on-load tap changer.

[0134] Figure 6 This is a schematic diagram of a terminal provided in an embodiment of the present invention. Figure 6 As shown, the terminal 6 in this embodiment includes: a processor 60, a memory 61, and a computer program 62 stored in the memory 61 and executable on the processor 60. When the processor 60 executes the computer program 62, it implements the steps in the above embodiments of the on-load tap changer state determination method, for example... Figure 1 Steps 101 to 103 are shown. Alternatively, when the processor 60 executes the computer program 62, it implements the functions of each module in the above-described device embodiments, for example... Figure 2 The functions of modules 21 to 23 are shown.

[0135] For example, the computer program 62 can be divided into one or more modules, which are stored in the memory 61 and executed by the processor 60 to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 62 in the terminal 6. For example, the computer program 62 can be divided into... Figure 2 Modules 21 to 23 are shown.

[0136] The terminal 6 can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The terminal 6 may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art will understand that... Figure 6 This is merely an example of terminal 6 and does not constitute a limitation on terminal 6. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0137] The processor 60 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0138] The memory 61 can be an internal storage unit of the terminal 6, such as a hard disk or memory of the terminal 6. The memory 61 can also be an external storage device of the terminal 6, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the terminal 6. Furthermore, the memory 61 can include both internal storage units and external storage devices of the terminal 6. The memory 61 is used to store the computer program and other programs and data required by the terminal. The memory 61 can also be used to temporarily store data that has been output or will be output.

[0139] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0140] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0141] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0142] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0143] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0144] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0145] If the integrated module is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above embodiments of the on-load tap changer state determination method. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately added to or subtracted from the content as required by the legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium may not include electrical carrier signals and telecommunication signals.

[0146] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for determining the state of an on-load tap changer, characterized in that, include: Obtain the main vibration waveform corresponding to the switching action of the on-load tap changer; Extract the time-frequency characteristics of the main vibration waveform; The time-frequency features are input into a trained random forest classifier to obtain the state judgment result of the on-load tap changer; wherein, the state judgment result includes switching from an odd position to an even position and switching from an even position to an odd position. Before inputting the time-frequency features into the trained random forest classifier, the method further includes: Calculate the lag autocorrelation coefficients for each time-domain feature; Based on the magnitude of each lagged autocorrelation coefficient, a subset of lagged autocorrelation coefficients are selected from each lagged autocorrelation coefficient. The step of inputting the time-frequency features into the trained random forest classifier includes: The time-frequency features corresponding to each selected lag autocorrelation coefficient are input into the trained random forest classifier.

2. The method for determining the state of an on-load tap changer according to claim 1, characterized in that, The acquisition of the main vibration waveform corresponding to the switching action of the on-load tap changer includes: Acquire the current and vibration signals of the on-load tap changer; The vibration signal is initially extracted based on the start and end times of the current signal; Using the peak position of the initially intercepted vibration signal as the midpoint of the interception, a signal of a preset length is intercepted from the initially intercepted vibration signal to obtain the main vibration waveform corresponding to the switching action of the on-load tap changer.

3. The method for determining the state of an on-load tap changer according to claim 1, characterized in that, Before inputting the time-frequency features into the trained random forest classifier, the method further includes: Establish an initial random forest classifier; The initial random forest classifier is trained using a sample set to obtain a trained random forest classifier. The sample set includes multiple training samples, each training sample being the main vibration waveform corresponding to the switching action of an on-load tap changer, and each training sample being labeled with the switching state of the on-load tap changer corresponding to that training sample.

4. The method for determining the state of an on-load tap changer according to claim 1, characterized in that, The calculation of the lag autocorrelation coefficient for each time-domain feature includes: For each time-domain feature, the lag autocorrelation coefficient of that time-domain feature is calculated based on a preset formula; the formula for calculating the lag autocorrelation coefficient is as follows: Where n represents the length of the main vibration waveform, l represents the preset hysteresis value, and σ 2 X represents the variance of the amplitude of the main vibration waveform, μ represents the mean of the amplitude of the main vibration waveform, and X represents the variance of the amplitude of the main vibration waveform. t X represents the value of this time-domain feature at time t. t-l This represents the value of the time-domain feature at time tl.

5. The method for determining the state of an on-load tap changer according to claim 1, characterized in that, The selection of a subset of lagged autocorrelation coefficients based on the magnitude of each lagged autocorrelation coefficient includes: Sort the lagged autocorrelation coefficients by size; Select a preset number of lag autocorrelation coefficients from largest to smallest.

6. The method for determining the state of an on-load tap changer according to any one of claims 1 to 5, characterized in that, The base classifier of the random forest classifier is a C4.5 decision tree.

7. A state determination device for an on-load tap changer, characterized in that, include: The acquisition module is used to acquire the main vibration waveform corresponding to the switching action of the on-load tap changer; An extraction module is used to extract the time-frequency characteristics of the main vibration waveform; The judgment module is used to input the time-frequency features into a trained random forest classifier to obtain the state judgment result of the on-load tap changer; The extraction module is specifically used for: Before inputting the time-frequency features into the trained random forest classifier, the lag autocorrelation coefficient of each time-domain feature is calculated; Based on the magnitude of each lagged autocorrelation coefficient, a subset of lagged autocorrelation coefficients are selected from each lagged autocorrelation coefficient. The judgment module is specifically used for: The time-frequency features corresponding to each selected lag autocorrelation coefficient are input into the trained random forest classifier.

8. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6 above.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6 above.

Images