Dry-type air-core reactor turn-to-turn short-circuit fault recognition method based on random matrix

Abstract

The application relates to a random matrix-based dry-type air-core reactor turn-to-turn short-circuit fault identification method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a real-time pulse response signal of a dry-type air-core reactor; the real-time pulse response signal comprises current electrical characteristic change information of the dry-type air-core reactor; a random matrix is constructed based on the real-time pulse response signal; eigenvalue analysis is performed on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the operating state of the dry-type air-core reactor; and the operating state characteristic parameter of the dry-type air-core reactor is input into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor. The method can accurately and efficiently detect a turn-to-turn short-circuit fault of the dry-type air-core reactor.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power systems, in particular to a dry-type air-core reactor inter-turn short-circuit fault identification method and device based on a random matrix, a computer device, a storage medium, and a computer program product. BACKGROUND

[0002] Dry-type air-core reactors are widely used in power systems. As a common power device in power systems, once a dry-type air-core reactor fails, it may pose a safety hazard.

[0003] Currently, methods for detecting inter-turn short-circuit faults of dry-type air-core reactors include probe coil methods, temperature monitoring methods, smoke monitoring methods, and equivalent impedance monitoring methods. The probe coil method is susceptible to spatial magnetic field influences, leading to state misjudgments. The temperature detection method relies on dry-type air-core reactor temperature changes for detection, but has low sensitivity in the early stages of a fault. The smoke monitoring method is affected by gases in the environment and can only provide early warnings for serious faults, with poor reliability. The equivalent impedance monitoring method uses the equivalent impedance transformation characteristics of the reactor to monitor inter-turn short-circuit faults, but has low monitoring sensitivity in the early stages of a fault due to small impedance changes.

[0004] Therefore, the conventional technology has the problem of being unable to accurately and efficiently detect inter-turn short-circuit faults of dry-type air-core reactors. SUMMARY

[0005] Therefore, it is necessary to provide a dry-type air-core reactor inter-turn short-circuit fault identification method and device based on a random matrix that can accurately and efficiently detect inter-turn short-circuit faults of dry-type air-core reactors, a computer device, a computer readable storage medium, and a computer program product.

[0006] A dry-type air-core reactor inter-turn short-circuit fault identification method based on a random matrix, the method comprising:

[0007] Obtaining a real-time impulse response signal of the dry-type air-core reactor; the real-time impulse response signal includes current electrical characteristic change information of the dry-type air-core reactor;

[0008] Based on the real-time impulse response signal, a random matrix is constructed;

[0009] Performing eigenvalue analysis on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the current operating state of the dry-type air-core reactor;

[0010] Inputting the operating state characteristic parameter of the dry-type air-core reactor into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor.

[0011] In an example embodiment, the random matrix is constructed based on the real-time impulse response signal, including:

[0012] The real-time impulse response signal is filtered to obtain a filtered response signal;

[0013] The filtered response signal is equally spaced sampled to obtain an equally spaced sampled signal;

[0014] The random matrix is constructed based on the equally spaced sampled signal.

[0015] In an example embodiment, the random matrix is constructed based on the equally spaced sampled signal, including:

[0016] An initial matrix is constructed based on the equally spaced sampled signal; the number of rows of the initial matrix corresponds to the dimension of the equally spaced sampled signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equally spaced sampling;

[0017] The initial matrix is transformed to obtain the random matrix; the row and column ratio of the random matrix is greater than the row and column ratio of the initial matrix.

[0018] In an example embodiment, the operating state characteristic parameter includes the average spectral radius of the eigenvalue of the random matrix and the eigenvalue of the covariance matrix corresponding to the random matrix, and the operating state characteristic parameter of the dry-type air-core reactor is input into the pre-trained fault recognition model to obtain the fault recognition result of the dry-type air-core reactor, including:

[0019] The average spectral radius of the eigenvalue and the eigenvalue of the covariance matrix are combined to form an input instance, which is input into the pre-trained fault recognition model, and the pre-trained fault recognition model determines a preset number of training instances most adjacent to the input instance in the training instance set through the pre-trained fault recognition model; the pre-trained fault recognition model is obtained by training the training instance set;

[0020] According to the fault degree corresponding to each training instance in the preset number of training instances, the fault degree of the dry-type air-core reactor is determined.

[0021] A training method of a fault recognition model, the method comprising:

[0022] Obtaining original impulse response signals corresponding to the dry-type air-core reactor under different working conditions; the original impulse response signals include electrical characteristic change information of the dry-type air-core reactor;

[0023] Based on each original impulse response signal, a random matrix corresponding to the dry-type air-core reactor under different working conditions is constructed;

[0024] The eigenvalue analysis is performed on each random matrix, and the operating state characteristic parameters of the dry-type air-core reactor corresponding to different operating conditions are obtained; and the operating state characteristic parameters corresponding to any operating condition represent the operating state of the dry-type air-core reactor under the operating condition.

[0025] Based on the operating state characteristic parameters of the dry-type air-core reactor under different operating conditions, a training instance set is generated.

[0026] The training instance set is used to train the fault identification model to be trained.

[0027] In an exemplary embodiment, the original impulse response signals of the dry-type air-core reactor corresponding to different operating conditions are obtained, including:

[0028] The pulse signal generator is used to inject pulse signals with different frequencies and different amplitudes into the dry-type air-core reactor.

[0029] The response signals of the dry-type air-core reactor under the pulse signals with different frequencies and different amplitudes are collected as the original impulse response signals of the dry-type air-core reactor under different operating conditions.

[0030] A dry-type air-core reactor turn-to-turn short-circuit fault identification device based on a random matrix, the device comprising:

[0031] An acquisition module is configured to acquire a real-time impulse response signal of the dry-type air-core reactor; the real-time impulse response signal includes current electrical characteristic change information of the dry-type air-core reactor;

[0032] A construction module is configured to construct a random matrix based on the real-time impulse response signal;

[0033] An analysis module is configured to perform eigenvalue analysis on the random matrix, and obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents a current operating state of the dry-type air-core reactor;

[0034] An identification module is configured to input the operating state characteristic parameter of the dry-type air-core reactor into a pre-trained fault identification model, and obtain a fault identification result of the dry-type air-core reactor.

[0035] A computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0036] A computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the above method.

[0037] A computer program product includes a computer program, and the computer program is executed by a processor to implement the steps of the above method.

[0038] The random matrix-based dry-type air-core reactor inter-turn short-circuit fault recognition method, device, computer device, storage medium and computer program product, by acquiring a real-time pulse response signal of the dry-type air-core reactor; the real-time pulse response signal includes current electrical characteristic change information of the dry-type air-core reactor; based on the real-time pulse response signal, a random matrix is constructed; eigenvalue analysis is performed on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the current operating state of the dry-type air-core reactor; the operating state characteristic parameter of the dry-type air-core reactor is input into a pre-trained fault recognition model to obtain a fault recognition result of the dry-type air-core reactor; in this way, the fault can be quickly recognized and a diagnosis result can be provided by real-time collection and processing of the response signal of the dry-type air-core reactor, the time for fault detection and processing is shortened, and the operating efficiency and reliability of the power system are improved. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0040] Figure 1 An application environment diagram of a dry-type air-core reactor inter-turn short-circuit fault recognition method based on a random matrix in an embodiment;

[0041] Figure 2 A flowchart of a dry-type air-core reactor inter-turn short-circuit fault recognition method based on a random matrix in an embodiment;

[0042] Figure 3 A flowchart of a step of constructing a random matrix in an embodiment;

[0043] Figure 4 A single-ring law spectrum distribution diagram in an embodiment;

[0044] Figure 5 A flowchart of a training method of a fault recognition model in an embodiment;

[0045] Figure 6 A flowchart of detecting a dry-type air-core reactor inter-turn short-circuit fault by using a random matrix theory method in an embodiment;

[0046] Figure 7 A flowchart of reading response signal data and the like when pulsed in an embodiment.

[0047] Figure 8 , Figure 9 is a schematic diagram of the response signal of a dry-type air-core reactor with normal and inter-turn short circuit in an embodiment;

[0048] Figure 10 is a schematic diagram of the single ring law generated by the average spectral radius of the random matrix constructed based on the response signal waveform of the dry-type air-core reactor winding in a healthy state in an embodiment;

[0049] Figure 11 is a schematic diagram of the single ring law generated by the average spectral radius of the random matrix constructed based on the response signal waveform of the dry-type air-core reactor winding in an inter-turn short circuit fault in an embodiment;

[0050] Figure 12 is a schematic diagram of a flow of a dry-type air-core reactor inter-turn short circuit fault identification method based on a random matrix in another embodiment;

[0051] Figure 13 is a structural block diagram of a dry-type air-core reactor inter-turn short circuit fault identification device based on a random matrix in an embodiment;

[0052] Figure 14 is a structural block diagram of a training device of a fault identification model in an embodiment;

[0053] Figure 15 is an internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0055] The dry-type air-core reactor of the present application uses air as the cooling medium, has the advantages of simple structure, convenient maintenance, safety and environmental protection, etc., is usually composed of a conductive coil and a support structure, and does not contain an iron core, thereby avoiding the problems of iron core loss and saturation, and the coil is wrapped with insulating material to prevent short circuit and improve voltage withstand capability. The dry-type air-core reactor inter-turn short circuit fault identification method based on a random matrix provided in the embodiments of the present application can be applied to, for example, Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. Server 104 acquires the real-time pulse response signal of the dry-type air-core reactor; the real-time pulse response signal includes information on the current electrical characteristic changes of the dry-type air-core reactor; server 104 constructs a random matrix based on the real-time pulse response signal; server 104 performs eigenvalue analysis on the random matrix to obtain the operating state characteristic parameters of the dry-type air-core reactor; the operating state characteristic parameters characterize the current operating state of the dry-type air-core reactor; server 104 inputs the operating state characteristic parameters of the dry-type air-core reactor into a pre-trained fault identification model to obtain the fault identification result of the dry-type air-core reactor. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0056] In one exemplary embodiment, such as Figure 2 As shown, a method for identifying inter-turn short-circuit faults in dry-type air-core reactors based on random matrices is provided, and this method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 202 to 206. Wherein:

[0057] Step 202: Obtain the real-time pulse response signal of the dry-type air-core reactor.

[0058] Among them, the real-time pulse response signal is the electrical characteristic change information of the dry air-core reactor, which is collected in real time by high-precision current and voltage sensors. The electrical characteristic change information refers to the waveform of electrical parameters such as voltage, current, and power changing over time.

[0059] Optionally, the server acquires the real-time pulse response signal of the dry-type air-core reactor.

[0060] Step 204: Construct a random matrix based on the real-time impulse response signal.

[0061] Here, a random matrix can refer to a high-dimensional random matrix.

[0062] Optionally, the server constructs a random matrix based on the real-time impulse response signal.

[0063] Step 206: Perform eigenvalue analysis on the random matrix to obtain the operating state characteristic parameters of the dry-type air-core reactor.

[0064] The operation state characteristic parameter represents a current operation state of the dry-type air-core reactor.

[0065] Optionally, the server performs eigenvalue analysis on the random matrix to obtain a characteristic parameter that can reflect the operation state of the dry-type air-core reactor.

[0066] In step 208, the operation state characteristic parameter of the dry-type air-core reactor is input into the pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor.

[0067] The pre-trained fault identification model can be a neural network model that is pre-trained to identify whether the dry-type air-core reactor sends a fault or a fault state. In actual application, the fault identification model can be a KNN model based on instance learning, or a random forest model based on a decision tree.

[0068] The fault identification result can be whether a fault occurs or a fault degree.

[0069] Optionally, the server inputs the operation state characteristic parameter of the dry-type air-core reactor into the pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor.

[0070] In the above dry-type air-core reactor inter-turn short circuit fault identification method based on a random matrix, a real-time pulse response signal of the dry-type air-core reactor is obtained; the real-time pulse response signal includes current electrical characteristic change information of the dry-type air-core reactor; a random matrix is constructed based on the real-time pulse response signal; eigenvalue analysis is performed on the random matrix to obtain an operation state characteristic parameter of the dry-type air-core reactor; the operation state characteristic parameter represents a current operation state of the dry-type air-core reactor; the operation state characteristic parameter of the dry-type air-core reactor is input into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor; in this way, the fault can be quickly identified and a diagnosis result can be provided by real-time collection and processing of the response signal of the dry-type air-core reactor, the time for fault detection and processing is shortened, and the operation efficiency and reliability of the power system are improved.

[0071] In one exemplary embodiment, as shown in FIG. 3, step 204 includes steps 302 to 306. Wherein: Figure 3

[0072] In step 302, the real-time pulse response signal is filtered to obtain a filtered response signal.

[0073] The filtered response signal can be a signal obtained by denoising the real-time pulse response signal using a wavelet transform method or a low-pass filter.

[0074] ​Optionally, the server filters the real-time impulse response signal to obtain a filtered response signal.

[0075] At step 304, the filtered response signal is equally spaced sampled to obtain an equally spaced sampled signal.

[0076] The equally spaced sampled signal can be a uniform sampled signal obtained by equally sampling a response curve corresponding to the filtered response signal. Equal interval sampling is a prerequisite for random matrix theory analysis. In order to construct a random matrix, the filtered response signal needs to be equally spaced sampled.

[0077] Optionally, the server equally spaces samples the filtered response signal to obtain an equally spaced sampled signal.

[0078] At step 306, a random matrix is constructed based on the equally spaced sampled signal.

[0079] Optionally, the server constructs a random matrix based on the equally spaced sampled signal.

[0080] In this embodiment, the real-time impulse response signal is filtered to obtain a filtered response signal, the filtered response signal is equally spaced sampled to obtain an equally spaced sampled signal, and a random matrix is constructed based on the equally spaced sampled signal. The real-time impulse response signal can be converted into a random matrix suitable for analysis using random matrix theory, thereby facilitating accurate analysis of the real-time impulse response signal of the dry-type air-core reactor and accurate identification of the fault condition of the dry-type air-core reactor.

[0081] In one exemplary embodiment, constructing a random matrix based on the equally spaced sampled signal includes: constructing an initial matrix based on the equally spaced sampled signal; the number of rows of the initial matrix corresponds to the dimension of the equally spaced sampled signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equally spaced sampling; transforming the initial matrix to obtain a random matrix; the row and column ratio of the random matrix is greater than that of the initial matrix.

[0082] The initial matrix can be a matrix first constructed based on the equally spaced sampled signal, and the number of rows and the number of columns of the matrix correspond to the dimension and the number of sampling points of the sampled signal, respectively.

[0083] Optionally, the server first constructs an initial matrix, and the number of rows and the number of columns of the matrix correspond to the dimension and the number of sampling points of the sampled signal, respectively. The server then transforms the initial matrix so that the row and column ratio of the transformed matrix is more suitable for extracting eigenvalues and other parameters, i.e., a high-dimensional random matrix is constructed.

[0084] In the embodiment, the initial matrix is constructed based on the equally spaced sampling signal; the number of rows of the initial matrix corresponds to the dimension of the equally spaced sampling signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equally spaced sampling; the initial matrix is transformed to obtain a random matrix; the row and column ratio of the random matrix is greater than that of the initial matrix; in this way, the matrix constructed for the first time can be transformed into a matrix with a more appropriate row and column ratio, thereby facilitating accurate analysis of the real-time pulse response signal of the dry-type air-core reactor, and further facilitating accurate identification of the fault condition of the dry-type air-core reactor.

[0085] In one exemplary embodiment, the operating state characteristic parameter includes an eigenvalue average spectral radius of the random matrix and an eigenvalue of a covariance matrix corresponding to the random matrix, and the operating state characteristic parameter of the dry-type air-core reactor is input into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor, including: inputting the eigenvalue average spectral radius and the eigenvalue of the covariance matrix into the pre-trained fault identification model to determine a preset number of training instances most adjacent to the input instance in the training instance set through the pre-trained fault identification model; the pre-trained fault identification model is obtained by training the training instance set; and determining the fault degree of the dry-type air-core reactor according to the fault degree corresponding to each training instance in the preset number of training instances.

[0086] wherein the random matrix can be represented as X, the eigenvalue average spectral radius MSR of the random matrix is represented as the mean value of the eigenvalue distribution of the matrix, and the calculation formula is:

[0087] ;

[0088] wherein, is the eigenvalue of the matrix.

[0089] In actual application, if the measured reactor MSR value is greatly different from the value of the normal reactor, it can be initially considered that the reactor is a fault reactor.

[0090] wherein the covariance matrix corresponding to the random matrix can be represented as S, , is the conjugate transpose matrix of X.

[0091] In actual application, eigenvalue solving on S can obtain the eigenvalue of S, which can also be used as a basis for judging whether the reactor has a turn-to-turn short circuit fault. The obtained eigenvalue is plotted on the complex plane to observe its distribution. According to the single ring law, the eigenvalue should present a circular ring distribution on the complex plane. Figure 4 is an ideal single ring law spectrum distribution graph, wherein the red solid line (the innermost circle solid line) is the theoretical inner diameter, i.e. , the red dotted line is the outer diameter (i.e. the outermost circle dotted line), normalized to 1, Figure 4 The blue dots in the figure are the eigenvalues, which are distributed between the inner diameter and the outer diameter, i.e. on the single ring, and the thick green line (i.e. the line between the outermost circle and the innermost circle) represents the mean of all eigenvalues calculated, denoted as .

[0092] For higher accuracy, the average spectral radius (MSR) obtained from the above multiple data and the eigenvalues of the covariance matrix can also be used to form a one-dimensional sequence, where most of the sequences (90%) are used as the training set and the remaining data are used as the test set. The training set is used to test the K- nearest neighbor model, and the test set is used to evaluate the performance of the model. Through the trained K- nearest neighbor model, the model can determine the optimal K value, which can be more accurate, so as to realize the accurate judgment of the degree of turn-to-turn short circuit of the dry-type air-core reactor. In addition to the K- nearest neighbor model, the training set can also be used to test the random forest model, and the test set can be used to evaluate the performance of the model. During the training process, the random forest will randomly extract data and features to generate multiple decision trees, and each decision tree will learn according to the division of features. When new data enters the model, all decision trees of the random forest will predict the fault based on the eigenvalues, and the final fault category will be obtained by voting all tree prediction results, so as to realize the accurate judgment of the degree of turn-to-turn short circuit of the dry-type air-core reactor.

[0093] wherein the input instance is a pair of numbers composed of the eigenvalue average spectral radius and the eigenvalues of the covariance matrix.

[0094] wherein the preset number can be a value determined by the model.

[0095] Optionally, the server inputs the eigenvalue average spectral radius and the eigenvalues of the covariance matrix into a set of numerical pairs into the pre-trained fault identification model; if the fault identification model adopts a K-neighbor model, the pre-trained K-neighbor model determines a preset number of numerical pairs closest to the set of numerical pairs among the numerical pairs provided in the training stage, for example, 7 numerical pairs closest to the set of numerical pairs are determined, among the 7 numerical pairs, different numerical pairs correspond to different fault degrees, the number of numerical pairs with the most fault degrees is determined as the fault degree of the dry-type air-core reactor; if the fault identification model adopts a random forest model, the voting of the decision tree trained by each numerical pair in the training stage is used to determine the fault type, for example, the above-mentioned data is input, each data will be judged by all decision trees, and a prediction result is output, if most of the 100 trees predict "slight fault", the input is classified as "slight fault". In the fault identification process of the random forest model, each decision tree independently judges the fault according to the input instance. Then, the model determines the fault degree of the input instance according to the voting result of all decision trees. In this way, the fault degree of the dry-type air-core reactor can be quickly and accurately determined.

[0096] In the embodiment, the eigenvalue average spectral radius and the eigenvalues of the covariance matrix are input into the pre-trained fault identification model, the pre-trained fault identification model determines a preset number of training instances closest to the input instance in the training instance set; the pre-trained fault identification model is trained by using the training instance set; and the fault degree of the dry-type air-core reactor is determined according to the fault degrees corresponding to each training instance in the preset number of training instances, so that the fault degree of the dry-type air-core reactor can be quickly and accurately determined.

[0097] In one exemplary embodiment, as shown in Figure 5 , a training method of a fault identification model is provided, and the method is applied to the server 104 in Figure 1 for example, and includes the following steps 502 to 510. Wherein:

[0098] Step 502, obtaining the original pulse response signals corresponding to the dry-type air-core reactor under different working conditions; the original pulse response signals include the electrical characteristic change information of the dry-type air-core reactor.

[0099] Optionally, the server obtains the original pulse response signals corresponding to the dry-type air-core reactor under different working conditions.

[0100] Step 504, based on each original impulse response signal, a random matrix corresponding to the dry-type air-core reactor under different working conditions is constructed.

[0101] Optionally, the server constructs a random matrix corresponding to the dry-type air-core reactor under different working conditions based on each original impulse response signal.

[0102] Step 506, eigenvalue analysis is performed on each random matrix to obtain an operating state characteristic parameter corresponding to the dry-type air-core reactor under different working conditions; and the operating state characteristic parameter corresponding to any working condition represents the operating state of the dry-type air-core reactor under the working condition.

[0103] Optionally, the server performs eigenvalue analysis on each random matrix to obtain a characteristic parameter reflecting the operating state of the dry-type air-core reactor under different working conditions.

[0104] Step 508, based on the operating state characteristic parameter of the dry-type air-core reactor under different working conditions, a training instance set is generated.

[0105] Any training instance in the training instance set is a pair of numbers composed of the average spectral radius of the eigenvalue and the eigenvalue of the covariance matrix. The training instance corresponds to the input instance in the above embodiment, the training instance is the training data input to the fault recognition model to be trained in the training stage, and the input instance is the input data input to the pre-trained fault recognition model in the test stage.

[0106] Optionally, the server generates a training instance set based on the operating state characteristic parameter of the dry-type air-core reactor under different working conditions.

[0107] Step 510, the training instance set is used to train the fault recognition model to be trained.

[0108] The fault recognition model to be trained can be a random forest model or a K-neighbor model, and the trained fault recognition model can identify whether the dry-type air-core reactor has a fault and can identify the fault degree in the case of a fault.

[0109] Optionally, the server trains the fault recognition model to be trained using the training instance set to obtain a trained fault recognition model.

[0110] The training method of the fault identification model comprises the following steps: obtaining original pulse response signals of the dry-type air-core reactor under different working conditions; the original pulse response signals comprise electrical characteristic change information of the dry-type air-core reactor; constructing random matrices corresponding to the dry-type air-core reactor under different working conditions based on the original pulse response signals; performing eigenvalue analysis on the random matrices to obtain operating state characteristic parameters of the dry-type air-core reactor under different working conditions; the operating state characteristic parameters corresponding to any working condition represent the operating state of the dry-type air-core reactor under the working condition; generating a training instance set based on the operating state characteristic parameters of the dry-type air-core reactor under different working conditions; and training the fault identification model to be trained by using the training instance set. In this way, a large amount of original pulse response data of the dry-type air-core reactor under different working conditions can be obtained by collecting a large amount of original pulse response signals of the dry-type air-core reactor under different working conditions, and a random matrix is constructed based on the original pulse response data. The key characteristic parameters of a large number of random matrices, such as eigenvalues and covariance matrices, are extracted, and the key characteristic parameters of the large number of random matrices are used as input data of the fault identification model to train the fault identification model. The trained fault identification model can accurately detect the turn-to-turn short-circuit fault of the dry-type air-core reactor, and further identify the fault degree of the turn-to-turn short-circuit fault of the dry-type air-core reactor.

[0111] In an exemplary embodiment, the original pulse response signals of the dry-type air-core reactor under different working conditions are obtained by injecting pulse signals of different frequencies and amplitudes into the dry-type air-core reactor through a pulse signal generator, and collecting response signals of the dry-type air-core reactor under the pulse signals of different frequencies and amplitudes as the original pulse response signals of the dry-type air-core reactor under different working conditions.

[0112] Optionally, the server generates pulse signals of different frequencies and amplitudes through a pulse signal generator, injects the pulse signals into the dry-type air-core reactor, and collects response signals of the dry-type air-core reactor under different working conditions using high-precision current sensors and voltage sensors. The response signals contain electrical characteristic change information of the dry-type air-core reactor under various fault states.

[0113] In this embodiment, pulse signals of different frequencies and amplitudes are injected into the dry-type air-core reactor through a pulse signal generator, and response signals of the dry-type air-core reactor under the pulse signals of different frequencies and amplitudes are collected as the original pulse response signals of the dry-type air-core reactor under different working conditions. A large amount of original pulse response signals of the dry-type air-core reactor under different working conditions can be quickly obtained, which is beneficial to improving the accuracy of the fault identification model.

[0114] For the convenience of those skilled in the art to understand, the following exemplary provides a kind of dry type air-core reactor interturn short circuit fault diagnosis method based on random matrix theory, specific steps include:

[0115] Step one: generate pulse signals of different frequencies and amplitudes through pulse signal generator, and inject these pulse signals into dry type air-core reactor. Under different working conditions of dry type air-core reactor, use high-precision current sensor and voltage sensor to collect the response signals of dry type air-core reactor, which contain the electrical characteristic change information of dry type air-core reactor under various fault states.

[0116] Step two: input the collected original pulse response signals into data processing module, first filter the signals to remove noise and interference, then uniformly sample the signals to ensure that the time-domain signals have consistent sampling intervals on time axis, and the uniformly sampled time-domain signals are more suitable for subsequent matrix feature extraction and analysis.

[0117] Step three: use the preprocessed uniformly sampled time-domain signals to construct high-dimensional random matrix. First, construct an initial matrix, the number of rows and columns of the matrix correspond to the dimension of the sampled signal and the number of sampling points respectively; then transform the initial matrix to make the matrix row-column ratio more suitable for extracting eigenvalue and other parameters, thereby constructing high-dimensional random matrix.

[0118] Step four: analyze the eigenvalues of the constructed high-dimensional random matrix through random matrix theory, and extract the characteristic parameters reflecting the running state of dry type air-core reactor. Use the extracted characteristic parameters as input, and use random forest model or K-nearest neighbor model for classification and identification. To diagnose whether fault occurs and the degree of fault.

[0119] The above method applies transient pulse signals to dry type air-core reactor, collects output current, and obtains high-dimensional random matrix, then calculates eigenvalues of the matrix, and uses eigenvalue comparison to detect whether interturn short circuit fault of reactor occurs. For the convenience of those skilled in the art to understand, the following provides two embodiments to further illustrate the above method.

[0120] In embodiment one, referring to Figure 6 , a flowchart for detecting interturn short circuit fault of dry type air-core reactor using random matrix theory method is provided.

[0121] In step one, a certain 35kV reactor is taken as experimental object in this embodiment, and its capacity System voltage U = 35 kV, 20 layers of coil. Square wave pulse as the excitation signal, that is, the stable square wave pulse voltage signal is input at one end of the measured dry type air core reactor, and the output response signal is detected at the other end. Under different degrees of turn-to-turn short circuit fault, the output response signal waveform will have a gap under the same excitation pulse, and the dry type air core reactor can be preliminarily judged whether it has a fault.

[0122] In step two, under the influence of various factors, noise and interference will inevitably exist in the output data. In order to avoid the influence of these factors on the results in subsequent data processing, the noise and interference should be removed when processing the data. At the same time, since the equal interval sampling is the precondition of the random matrix theory analysis, the response curve should be equally sampled to obtain the uniformly sampled data.

[0123] Figure 7 The flow chart for reading the response signal and other data under pulse excitation.

[0124] In step three, the uniformly sampled data obtained in step two is used to form a high-dimensional matrix, so that the eigenvalue and covariance matrix of the matrix can be calculated. By repeatedly sampling multiple times, multiple equally spaced sampling data vectors are obtained, and each equally spaced data vector obtained by sampling is used as a row or a column of the matrix to construct an initial matrix. For example, assuming that there are N experiments, and T frequency domain data points are obtained in each experiment, then the size of the initially constructed matrix X is T*N. However, if the number of electrical quantities sampled is small, that is, the sampling rate is large or the sampling time window is long, the row and column ratio of the formed matrix will be very small, which is not conducive to calculation, so the matrix needs to be transformed to make it have a more appropriate row and column ratio. For example, there is a matrix X' with an inappropriate row and column ratio. The T column elements of X' are split into i blocks in order, and then X' is divided into i sub-matrices. The i sub-matrices are sequentially translated to the original X' below. Since the random matrix only requires that the data is large enough, its row and column ratio is appropriate.

[0125] In step four, the eigenvalue extraction of the random matrix is performed, mainly the mean spectral radius (MSR) and the eigenvalue of the covariance matrix. The mean spectral radius (MSR) is represented as the mean value of the eigenvalue distribution of the matrix, and its calculation formula is:

[0126] ;

[0127] Where, is the eigenvalue of the matrix.

[0128] In practical applications, if the measured value of the MSR of the reactor is significantly different from the value of the normal reactor, it can be preliminarily considered that the dry-type air-core reactor is a faulty reactor. If the empirical spectral distribution (ESD) is used to describe the distribution of the eigenvalues of the matrix, the difference in the average spectral radius can be more intuitively observed. For a random matrix X, the corresponding covariance matrix S:

[0129]

[0130] wherein, is the conjugate transpose matrix of X. The eigenvalues of S obtained by solving the eigenvalues of S are , which can also be used as a basis for judging whether the reactor has an inter-turn short-circuit fault. The obtained eigenvalues can be plotted on the complex plane to observe their distribution. According to the single-ring law, the eigenvalues should present a circular ring distribution on the complex plane. See Figure 4 and the corresponding description of Figure 4 above, in order to achieve higher accuracy, the average spectral radius (MSR) and the eigenvalues of the covariance matrix obtained by using the above multiple data form a one-dimensional sequence, wherein most of the sequences (90%) are used as a training set, and the remaining data are used as a test set. The training set is used to test the random forest model or the K-neighbor model, and the test set is used to evaluate the performance of the model. Through the trained random forest model, the voting classification of multiple decision trees can be performed, so as to more accurately judge the degree of inter-turn short-circuit fault of the dry-type air-core reactor; or through the trained K-neighbor model, the best K value can be determined, so as to more accurately judge the degree of inter-turn short-circuit fault of the dry-type air-core reactor.

[0131] In Example Two, Step One takes a certain 35kV reactor as the experimental object, with a capacity of , a system voltage U = 35kV, and a total of 20 layers of coils. A square wave pulse is used as the excitation signal, and the pulse data used are: low level V1 = 0V, high level V2 = 500V, pulse delay TD = 1us, pulse rise time TR = 30ns, pulse fall time TF = 30ns, pulse width PW = 300ns, and pulse period PER = 1s. The response signals of the normal and inter-turn short-circuit dry-type air-core reactors measured under this pulse are shown in Figure 8 , Figure 9 .

[0132] In Step Two, the measured data can be de-noised using a low-pass filter or wavelet transform to de-noise the obtained original data, remove noise components, and retain useful signal information. In Example Two, wavelet transform is used, and its mathematical expression is:

[0133] ;​

[0134] in: It is a signal The wavelet coefficients at scale a and position b, where a is the scale parameter that determines the expansion and contraction of the wavelet function, and b is the translation parameter that determines the position of the wavelet function. It is the mother wavelet function. It is the original signal.

[0135] Wavelet transform decomposes a signal into wavelet coefficients at different scales and locations. These coefficients contain the signal's local features and frequency information. The wavelet coefficients are then thresholded, with coefficients below a certain threshold set to zero (hard thresholding) or softened (soft thresholding) to remove noise components. The processed wavelet coefficients are then subjected to inverse transform to obtain the denoised signal. After denoising, the signal is sampled at equal intervals. The sampled data are then organized to construct a high-dimensional matrix.

[0136] In step three, for the high-dimensional matrix constructed by the equally spaced sampling in step two, a row-column transformation is performed if the row-column ratio is inappropriate. For example, to obtain the waveform of a dry-type air-core reactor before and after a fault, a total of one cycle is obtained. The sampling points form the original data matrix X′, as follows:

[0137] ;

[0138] The first element in matrix X′ represents the output current data of the first sampling point, i.e. Its row-to-column ratio is The original data matrix X′ is divided into rows and columns into 50 blocks, which are then shifted and expanded to form the state data matrix X′ as follows:

[0139] ;

[0140] The ratio of the number of rows to the number of columns in the matrix becomes 1 / 4. Using the above method, X′ is transformed into matrix X. After constructing a high-dimensional matrix from the measured data, the average spectral radius of the eigenvalues ​​of the random matrix of the dry-type air-core reactor under non-fault conditions is calculated according to the above transformation and the formula for calculating the average spectral radius. The value is r = 0.0736. The mean spectral radius of the eigenvalues ​​of the random matrix of a dry-type air-core reactor experiencing an inter-turn short-circuit fault is... The average spectral radius of the random matrix constructed from the response signal waveforms of the dry-type air-core reactor winding under healthy conditions and when an inter-turn short-circuit fault occurs can generate a schematic diagram of the single-loop law that meets the requirements, as shown below. Figure 10 , Figure 11 As shown.

[0141] In step four, in order to more accurately test the fault degree of the turn-to-turn short circuit of the dry-type air-core reactor, a random matrix model (or a K-neighbor model) is adopted. Specifically, the average spectral radius (MSR) calculated above and the eigenvalue sequence of the covariance matrix are combined to form a new sequence, 90% of the new sequence is used as training of the random forest algorithm (or the K-neighbor model), and the remaining part is used as identification test. In this example, the Euclidean distance is selected as the distance metric. For the eigenvalue spectrum distribution of the waveform of any two types of signals, the Euclidean distance of the MSRs is wherein:

[0142] ;

[0143] The present application can extract more accurate and comprehensive feature parameters such as the average spectral radius (MSR) and the eigenvalue of the covariance matrix from high-dimensional data by introducing the random matrix theory, which can more accurately reflect the operating state of the reactor and improve the accuracy of fault detection. Meanwhile, the extracted feature parameters are classified and identified in combination with the random forest model (or the K-neighbor model), which can effectively distinguish different degrees of turn-to-turn short circuit faults and improve the accuracy of fault identification. Since the random matrix theory has significant advantages in processing high-dimensional data and noise, it can extract effective feature parameters in a complex and disturbed environment, improve the anti-interference ability of the system, and solve the problems of low precision, high cost, complicated method, and great environmental influence of the current turn-to-turn short circuit fault detection technology of the dry-type air-core reactor. By comparing the matrix eigenvalues, the data processing time is reduced. The intelligent detection system based on the random matrix theory can realize the intelligentization of the turn-to-turn short circuit fault detection. Through the collection and processing of the reactor response signals, the system can quickly identify the fault and provide the diagnosis result, shorten the fault detection and processing time, and improve the operation efficiency and reliability of the power system.

[0144] In another embodiment, as shown in Figure 12 , a random matrix-based dry-type air-core reactor turn-to-turn short circuit fault identification method is provided. Taking a server in Figure 1 as an example, the method includes the following steps:

[0145] Step 1202, acquiring a real-time pulse response signal of the dry-type air-core reactor; the real-time pulse response signal includes current electrical characteristic change information of the dry-type air-core reactor.

[0146] Step 1204, filtering the real-time pulse response signal to obtain a filtered response signal.

[0147] Step 1206, equally sampling the filtered response signal to obtain an equally sampled signal.

[0148] Step 1208, based on the equally spaced sampling signal, an initial matrix is constructed; the number of rows of the initial matrix corresponds to the dimension of the equally spaced sampling signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equally spaced sampling.

[0149] Step 1210, the initial matrix is transformed to obtain a random matrix; the row and column ratio of the random matrix is greater than that of the initial matrix.

[0150] Step 1212, eigenvalue analysis is performed on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the operating state of the dry-type air-core reactor.

[0151] Step 1214, the operating state characteristic parameter of the dry-type air-core reactor is input into the pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor.

[0152] It should be noted that the specific limitations of the above steps can be referred to the specific limitations of the above-mentioned method for identifying turn-to-turn short circuit fault of dry-type air-core reactor based on random matrix.

[0153] It should be understood that, although each step in the flowchart involved in each of the above-mentioned embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each of the above-mentioned embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or steps or stages in other steps.

[0154] Based on the same inventive concept, the embodiments of the present application also provide a device for implementing the above-mentioned method for identifying turn-to-turn short circuit fault of dry-type air-core reactor based on random matrix, and a training device for implementing the above-mentioned training method of fault identification model. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more device embodiments for identifying turn-to-turn short circuit fault of dry-type air-core reactor based on random matrix provided below can be referred to the limitations of the method for identifying turn-to-turn short circuit fault of dry-type air-core reactor based on random matrix, and the specific limitations in one or more training device embodiments of fault identification model provided below can be referred to the limitations of the training method of fault identification model, which will not be repeated here.

[0155] In one example embodiment, as shown in Figure 13 A random matrix-based dry-type air-core reactor turn-to-turn short-circuit fault recognition device is provided, including: an acquisition module 1302, a construction module 1304, an analysis module 13013, and a recognition module 1308, wherein:

[0156] The acquisition module 1302 is configured to acquire a real-time impulse response signal of the dry-type air-core reactor; the real-time impulse response signal includes current electrical characteristic change information of the dry-type air-core reactor;

[0157] The construction module 1304 is configured to construct a random matrix based on the real-time impulse response signal;

[0158] The analysis module 13013 is configured to perform eigenvalue analysis on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents an operating state of the dry-type air-core reactor;

[0159] The recognition module 1308 is configured to input the operating state characteristic parameter of the dry-type air-core reactor into a pre-trained fault recognition model to obtain a fault recognition result of the dry-type air-core reactor.

[0160] In one example embodiment, the construction module 1304 is specifically configured to perform filtering processing on the real-time impulse response signal to obtain a filtered response signal; perform equidistant sampling on the filtered response signal to obtain an equidistant sampling signal; and construct the random matrix based on the equidistant sampling signal.

[0161] In one example embodiment, the construction module 1304 is specifically configured to construct an initial matrix based on the equidistant sampling signal; a row number of the initial matrix corresponds to a dimension of the equidistant sampling signal; a column number of the initial matrix corresponds to a sampling point number of the equidistant sampling; perform transformation on the initial matrix to obtain the random matrix; and a row-column ratio of the random matrix is greater than a row-column ratio of the initial matrix.

[0162] In an example embodiment, the operating state characteristic parameter includes an eigenroot average spectral radius of the random matrix and an eigenvalue of a covariance matrix corresponding to the random matrix, the identification module 1308 is specifically configured to input the eigenroot average spectral radius and the eigenvalue of the covariance matrix into an input instance into the pre-trained fault identification model, determine a preset number of training instances most adjacent to the input instance in a training instance set through the pre-trained fault identification model; the pre-trained fault identification model is obtained by training the training instance set; and determine the fault degree of the dry-type air-core reactor according to the fault degree corresponding to each training instance in the preset number of training instances.

[0163] In an example embodiment, as shown in Figure 14 A training device of a fault identification model is provided, including: an original data acquisition module 1402, a random matrix construction module 1404, a random matrix analysis module 1406, a training instance set generation module 1408, and a fault identification model training module 1410. Wherein:

[0164] The original data acquisition module 1402 is configured to acquire original pulse response signals corresponding to the dry-type air-core reactor under different working conditions; the original pulse response signals include electrical characteristic change information of the dry-type air-core reactor under each fault state;

[0165] The random matrix construction module 1404 is configured to construct random matrices corresponding to the dry-type air-core reactor under different working conditions based on each original pulse response signal;

[0166] The random matrix analysis module 1406 is configured to perform eigenvalue analysis on each random matrix to obtain operating state characteristic parameters corresponding to the dry-type air-core reactor under different working conditions; the operating state characteristic parameter corresponding to any working condition represents the operating state of the dry-type air-core reactor under the working condition;

[0167] The training instance set generation module 1408 is configured to generate a training instance set based on the operating state characteristic parameters of the dry-type air-core reactor under different working conditions;

[0168] The fault identification model training module 1410 is configured to train a fault identification model to be trained using the training instance set.

[0169] In an example embodiment, the original data acquisition module 1402 is specifically configured to inject pulse signals of different frequencies and different amplitudes into the dry-type air-core reactor through a pulse signal generator; and collect response signals of the dry-type air-core reactor under the pulse signals of different frequencies and different amplitudes as corresponding original pulse response signals of the dry-type air-core reactor under different working conditions.

[0170] Each of the above-mentioned modules of the dry-type air-core reactor turn-to-turn short-circuit fault recognition device and the training device of the fault recognition model can be realized by software, hardware, and a combination thereof, in whole or in part. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to each of the above-mentioned modules.

[0171] In an example embodiment, a computer device is provided, which can be a server, and an internal structure diagram thereof can be as shown in Figure 15 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the computer device is configured to store dry-type air-core reactor turn-to-turn short-circuit fault recognition data based on a random matrix and training data of a fault recognition model. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a dry-type air-core reactor turn-to-turn short-circuit fault recognition method based on a random matrix.

[0172] Those skilled in the art can understand that Figure 15 The structure shown in the above-mentioned figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. Specifically, the computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0173] In an embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, the computer program being executed by the processor to cause the processor to perform the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix and the steps of the method for training a fault recognition model. The steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix can be the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix in each of the above embodiments, and the steps of the method for training a fault recognition model can be the steps of the method for training a fault recognition model in each of the above embodiments.

[0174] In an embodiment, a computer readable storage medium is provided, storing a computer program, the computer program being executed by a processor to cause the processor to perform the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix and the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix. The steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix can be the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix in each of the above embodiments, and the steps of the method for training a fault recognition model can be the steps of the method for training a fault recognition model in each of the above embodiments.

[0175] In an embodiment, a computer program product is provided, comprising a computer program, the computer program being executed by a processor to cause the processor to perform the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix and the steps of the method for training a fault recognition model. The steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix can be the steps of the method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix in each of the above embodiments, and the steps of the method for training a fault recognition model can be the steps of the method for training a fault recognition model in each of the above embodiments.

[0176] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0177] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0178] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for recognizing turn-to-turn short-circuit fault of dry-type air-core reactor based on random matrix, characterized in that, The method comprises: acquiring a real-time impulse response signal of a dry-type air-core reactor; the real-time impulse response signal comprises current electrical characteristic change information of the dry-type air-core reactor; filtering the real-time impulse response signal to obtain a filtered response signal; equidistantly sampling the filtered response signal to obtain an equidistantly sampled signal; constructing an initial matrix based on the equidistantly sampled signal; the number of rows of the initial matrix corresponds to the dimension of the equidistantly sampled signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equidistantly sampled signal; transforming the initial matrix to obtain a random matrix; the row-to-column ratio of the random matrix is greater than the row-to-column ratio of the initial matrix; performing eigenvalue analysis on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the current operating state of the dry-type air-core reactor; inputting the operating state characteristic parameter of the dry-type air-core reactor into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor.

2. The method of claim 1, wherein, The operating state characteristic parameter comprises an average spectral radius of eigenvalues of the random matrix and eigenvalues of a covariance matrix corresponding to the random matrix, and the inputting of the operating state characteristic parameter of the dry-type air-core reactor into the pre-trained fault identification model to obtain the fault identification result of the dry-type air-core reactor comprises: inputting the average spectral radius of eigenvalues and the eigenvalues of the covariance matrix into an input instance into the pre-trained fault identification model, determining a preset number of training instances most adjacent to the input instance in a training instance set through the pre-trained fault identification model; the pre-trained fault identification model is obtained by training the training instance set; determining the fault degree of the dry-type air-core reactor according to the fault degrees corresponding to each of the training instances in the preset number of training instances.

3. A training method for a fault identification model, characterized in that, The method comprises: acquiring original impulse response signals corresponding to different working conditions of a dry-type air-core reactor; the original impulse response signals comprise electrical characteristic change information of the dry-type air-core reactor; filtering the original impulse response signals to obtain filtered response signals; equidistantly sampling the filtered response signals to obtain equidistantly sampled signals; constructing an initial matrix based on the equidistantly sampled signals; the number of rows of the initial matrix corresponds to the dimension of the equidistantly sampled signals; the number of columns of the initial matrix corresponds to the number of sampling points of the equidistantly sampled signals; transforming the initial matrices constructed based on each of the original impulse response signals to construct random matrices corresponding to different working conditions of the dry-type air-core reactor; the row-to-column ratio of the random matrix is greater than the row-to-column ratio of the initial matrix; performing eigenvalue analysis on each of the random matrices to obtain operating state characteristic parameters corresponding to different working conditions of the dry-type air-core reactor; the operating state characteristic parameter corresponding to any of the working conditions represents the operating state of the dry-type air-core reactor under the working condition; Generate a training instance set based on the operating state characteristic parameters of the dry-type air-core reactor under different working conditions; Train the fault identification model to be trained using the training instance set.

4. The method of claim 3, wherein, The method includes: Injecting pulse signals of different frequencies and different amplitudes into the dry-type air-core reactor through a pulse signal generator; Collecting response signals of the dry-type air-core reactor under the pulse signals of different frequencies and different amplitudes as the original pulse response signals of the dry-type air-core reactor under different working conditions.

5. A random matrix based dry-type air-core reactor inter-turn short circuit fault recognition device, characterized in that, The device includes: An acquisition module configured to acquire a real-time pulse response signal of a dry-type air-core reactor; the real-time pulse response signal includes current electrical characteristic change information of the dry-type air-core reactor; A construction module configured to filter the real-time pulse response signal to obtain a filtered response signal; perform equidistant sampling on the filtered response signal to obtain an equidistant sampling signal; construct an initial matrix based on the equidistant sampling signal; the number of rows of the initial matrix corresponds to the dimension of the equidistant sampling signal; the number of columns of the initial matrix corresponds to the number of sampling points of the equidistant sampling signal; transform the initial matrix to obtain a random matrix; the row and column ratio of the random matrix is greater than the row and column ratio of the initial matrix; An analysis module configured to perform eigenvalue analysis on the random matrix to obtain an operating state characteristic parameter of the dry-type air-core reactor; the operating state characteristic parameter represents the current operating state of the dry-type air-core reactor; An identification module configured to input the operating state characteristic parameter of the dry-type air-core reactor into a pre-trained fault identification model to obtain a fault identification result of the dry-type air-core reactor. 6.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-5 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the method of any one of claims 1 to 4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 4.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 4.

Images