Method, device, system and electronic equipment for detecting transformer winding looseness

Abstract

The application provides a transformer winding loosening detection method, device, system and electronic equipment, wherein the method comprises the following steps: obtaining vibration information of a transformer collected by a detection device; determining a vibration feature matrix according to the vibration information; and calculating a loosening state parameter by using a clustering algorithm according to the vibration feature matrix, wherein the loosening state parameter is used to indicate the loosening state of the winding of the transformer. The detection result accuracy of the transformer winding loosening detection method is high.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, and in particular to a method, apparatus, system and electronic equipment for detecting loose transformer windings. Background Technology

[0002] Transformer windings generate electromagnetic forces under the influence of current and magnetic fields, causing vibrations. With increasing operating time and the influence of complex electrical, magnetic, and thermal factors, transformers are prone to mechanical failures such as loosening. Loosening of the transformer windings alters their modal characteristics. When the changing natural frequency approaches the frequency of the electromagnetic force, resonance can easily occur, significantly increasing vibration, further exacerbating the loosening, and even leading to coil disintegration and collapse, resulting in instability. Furthermore, winding loosening has a cumulative effect; prolonged loosening significantly reduces the winding's short-circuit withstand capability, posing a serious threat to the safe and stable operation of the transformer. Therefore, monitoring and diagnosing the loosening status of transformer windings to determine different degrees of loosening is essential for timely detection and repair of potential faults.

[0003] Fault diagnosis methods based on vibration signals have become the main means of monitoring transformer winding loosening. The basic idea is to extract vibration information that can reflect the mechanical state and further determine the degree of winding loosening based on changes in vibration information.

[0004] However, existing methods for determining the loosening state of windings based on changes in vibration information cannot effectively eliminate interference information in the vibration information. Interference information includes at least high temperature, electrical, magnetic and high noise interference in the transformer operating environment. The accuracy of the detection results obtained by existing methods for detecting the loosening state of transformer windings is low. Summary of the Invention

[0005] This invention provides a method, apparatus, system, and electronic device for detecting loose transformer windings, in order to solve the problem of low accuracy of existing methods for detecting the loose state of transformer windings.

[0006] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0007] In a first aspect, embodiments of the present invention provide a method for detecting loose transformer windings, comprising:

[0008] Obtain vibration information of the transformer collected by the detection equipment;

[0009] Based on the vibration information, determine the vibration feature matrix;

[0010] Based on the vibration feature matrix, a clustering algorithm is used to calculate the loosening state parameters, which are used to indicate the loosening state of the transformer windings.

[0011] Optionally,

[0012] Based on the vibration information, the vibration feature matrix is ​​determined, including:

[0013] Based on the vibration information, the vibration characteristic parameters are calculated;

[0014] The vibration characteristic matrix is ​​obtained based on the vibration characteristic parameters.

[0015] Optionally,

[0016] The vibration characteristic parameter includes at least one of the following:

[0017] Fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy ratio, energy matrix norm, and energy matrix entropy;

[0018] Based on the vibration characteristic parameters, the vibration characteristic matrix is ​​obtained, including:

[0019] The vibration characteristic parameters are combined to obtain the vibration characteristic matrix.

[0020] Optionally,

[0021] Based on the vibration feature matrix, the loosening state parameters are calculated using a clustering algorithm, including:

[0022] At least two initial cluster centers are randomly generated based on a preset initial cluster, wherein the number of preset initial clusters is equal to the number of initial cluster centers;

[0023] The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers.

[0024] Optionally,

[0025] Based on the vibration feature matrix and at least two sets of initial cluster centers, the loosening state parameters are calculated, including:

[0026] Calculation steps: Calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center;

[0027] Update steps: Divide the vibration feature matrices into types according to the total Euclidean distance value to obtain sample clusters, and obtain updated cluster centers based on the sample clusters;

[0028] A second check is performed to determine whether the updated cluster centers are consistent with the initial cluster centers;

[0029] If the result of the second detection is that the updated cluster center is inconsistent with the initial cluster center, the updated cluster center is used to replace the old initial cluster center to obtain a new cluster center, and the update step is returned until the result of the second detection is that the updated cluster center is consistent with the latest cluster center.

[0030] If the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center, the loosening state parameter is determined based on the updated cluster center.

[0031] Secondly, embodiments of the present invention provide a device for detecting loose transformer windings, comprising:

[0032] The acquisition module is used to acquire vibration information of the transformer collected by the detection equipment;

[0033] An execution module is used to determine a vibration feature matrix based on the vibration information;

[0034] The execution module is further configured to calculate loosening state parameters based on the vibration feature matrix using a clustering algorithm, the loosening state parameters being used to indicate the loosening state of the transformer windings.

[0035] Optionally,

[0036] The execution module is also used to calculate vibration characteristic parameters based on the vibration information;

[0037] The execution module is further configured to obtain the vibration feature matrix based on the vibration feature parameters.

[0038] Thirdly, embodiments of the present invention provide a detection system for loose transformer windings, comprising:

[0039] A vibration sensor for monitoring loose transformer windings is installed on the transformer.

[0040] A transformer winding loosening detection device is connected to the vibration acquisition sensor and is used to determine a vibration feature matrix based on the vibration information of the monitored transformer obtained by the vibration acquisition sensor; based on the vibration feature matrix, a clustering algorithm is used to calculate loosening state parameters, which are used to indicate the loosening state of the transformer winding.

[0041] Fourthly, embodiments of the present invention provide an electronic device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps in the transformer winding loosening detection method as described in any one of the first aspects.

[0042] Fifthly, embodiments of the present invention provide a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps in the transformer winding loosening detection method as described in any one of the first aspects.

[0043] In this embodiment of the invention, vibration information of the transformer collected by a detection device is acquired; a vibration feature matrix is ​​determined based on the vibration information; and a loosening state parameter is calculated using a clustering algorithm based on the vibration feature matrix. Compared with existing technologies that directly analyze vibration information, the vibration feature matrix can effectively eliminate interference factors in the vibration information directly obtained through detection, and can accurately characterize the loosening condition of the transformer windings. Furthermore, the clustering algorithm based on the vibration feature matrix can accurately obtain the loosening state parameter reflecting the loosening condition of the transformer windings. The method for detecting the loosening state of transformer windings in this embodiment of the invention has a high accuracy rate. Attached Figure Description

[0044] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0045] Figure 1 This is one of the flowcharts illustrating the method for detecting loose transformer windings according to an embodiment of the present invention;

[0046] Figure 2 This is a second schematic flowchart of the method for detecting loose transformer windings according to an embodiment of the present invention;

[0047] Figure 3 This is a schematic diagram showing the installation location (measuring point) of the vibration sensor;

[0048] Figure 4 Waveforms and spectrum diagrams for different loosening states;

[0049] Figure 5 This is a schematic diagram illustrating the classification results obtained by using the K-means clustering algorithm to classify different degrees of looseness;

[0050] Figure 6This is the third flowchart illustrating the method for detecting loose transformer windings according to an embodiment of the present invention;

[0051] Figure 7 This is the fourth flowchart illustrating the method for detecting loose transformer windings according to an embodiment of the present invention.

[0052] Figure 8 This is a schematic block diagram of a transformer winding loosening detection device according to an embodiment of the present invention;

[0053] Figure 9 This is a schematic diagram of the transformer winding loosening detection system according to an embodiment of the present invention;

[0054] Figure 10 This is a schematic block diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] This invention provides a method for detecting loose transformer windings. See [link to relevant documentation]. Figure 1 As shown, Figure 1 This is one of the flowcharts illustrating a method for detecting loose transformer windings according to an embodiment of the present invention, including:

[0057] Step 11: Obtain the vibration information of the transformer collected by the detection equipment;

[0058] Step 12: Determine the vibration characteristic matrix based on the vibration information;

[0059] Step 13: Based on the vibration feature matrix, the loosening state parameters are calculated using a clustering algorithm. The loosening state parameters are used to indicate the loosening state of the windings of the monitored transformer.

[0060] In some embodiments of the present invention, optionally, the detection device is a piezoelectric acceleration vibration sensor, which is magnetically attached to the surface of the transformer tank to collect the vibration information of the transformer.

[0061] In some embodiments of the present invention, the detection device is optionally a vibration sensor. According to the winding height of the transformer whose vibration information is to be collected, vibration sensors are arranged at five positions: the bottom of the winding, 1 / 4 of the winding height, 2 / 4 of the winding height, 3 / 4 of the winding height, and the top of the winding. These are named measuring point 1, measuring point 2, ..., measuring point 5, that is, five vibration sensors are arranged for each winding.

[0062] In some embodiments of the present invention, optionally, in step 12, a vibration feature matrix is ​​determined based on the vibration information; the vibration feature matrix includes at least one of the following parameters: fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy ratio, energy matrix norm, and energy matrix entropy.

[0063] Cluster analysis, also known as group analysis, is a statistical analysis method for studying the classification of samples or indicators, and it is also an important algorithm in data mining. Cluster analysis consists of several patterns, typically a vector of measurements or a point in a multidimensional space. Cluster analysis is based on similarity; patterns within a cluster are more similar than patterns in different clusters.

[0064] In some embodiments of the present invention, optionally, loosening state parameters are calculated using a clustering algorithm based on the vibration feature matrix, including:

[0065] Based on the vibration feature matrix, the loosening state parameters are calculated using the k-means clustering algorithm.

[0066] The k-means algorithm takes an input k and then divides n data objects into k clusters such that the resulting clusters satisfy the following conditions: objects within the same cluster have high similarity, while objects in different clusters have low similarity. Cluster similarity is calculated using a "centroid" (center of gravity) obtained from the mean of the objects in each cluster.

[0067] The working process of the k-means algorithm is explained below:

[0068] First, select k objects from n data objects as initial cluster centers; then, for the remaining objects, assign them to the clusters most similar to them (represented by the cluster centers) based on their similarity (distance) to these cluster centers.

[0069] Then calculate the cluster center of each newly obtained cluster (the mean of all objects in that cluster); repeat this process until the standard measure function begins to converge.

[0070] The standard deviation is generally used as the standard measure function. k clusters have the following characteristics: each cluster is as compact as possible, while the clusters are as far apart as possible.

[0071] In this embodiment of the invention, vibration information of the transformer collected by a detection device is acquired; a vibration feature matrix is ​​determined based on the vibration information; and a loosening state parameter is calculated using a clustering algorithm based on the vibration feature matrix. Compared with existing technologies that directly analyze vibration information, the vibration feature matrix can effectively eliminate interference factors in the vibration information directly obtained through detection, and can accurately characterize the loosening condition of the transformer windings. Furthermore, the clustering algorithm based on the vibration feature matrix can accurately obtain the loosening state parameter reflecting the loosening condition of the transformer windings. The method for detecting the loosening state of transformer windings in this embodiment of the invention has a high accuracy rate.

[0072] In some embodiments of the present invention, see optionally, see Figure 2 As shown, Figure 2 This is a second schematic flowchart of the transformer winding loosening detection method according to an embodiment of the present invention. Based on vibration information, the vibration characteristic matrix is ​​determined, including:

[0073] Step 21: Calculate the vibration characteristic parameters based on the vibration information;

[0074] Step 22: Obtain the vibration characteristic matrix based on the vibration characteristic parameters.

[0075] In some embodiments of the present invention, the vibration characteristic parameters optionally include at least one of the following:

[0076] Fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy ratio, energy matrix norm, and energy matrix entropy;

[0077] Based on the vibration characteristic parameters, the vibration characteristic matrix is ​​obtained, including:

[0078] By combining the vibration characteristic parameters, a vibration characteristic matrix is ​​obtained.

[0079] In some embodiments of the present invention, optionally, in step 21, vibration characteristic parameters are calculated based on vibration information, and the calculation process is as follows:

[0080] 1) Fundamental frequency amplitude A 100Hz The amplitude of the 100Hz vibration signal is extracted by Fourier transforming the time-domain signal.

[0081] 2) Fundamental frequency weight P 100Hz The ratio of the amplitude of the 100Hz vibration signal to the sum of the amplitudes of all 100×i (i=1, 2, 3, …, n)Hz vibration signals after Fourier transform of the time-domain signal.

[0082]

[0083] In the formula: n is taken as 20.

[0084] 3) The proportion of wavelet energy at the dominant frequency P w The frequency at which the maximum amplitude is determined after Fourier transform is the main frequency. The frequency range of wavelet transform is reasonably divided according to the sampling frequency and the main frequency, and the wavelet energy ratio of the main frequency range is calculated.

[0085] The vibration signal was determined to have a dominant frequency of 100×i Hz after Fourier transform.

[0086] The wavelet transform formula is:

[0087]

[0088] In the formula: a is the scale factor, τ is the translation amount, and f(t) is the collected vibration signal on the surface of the oil tank.

[0089] Given a fixed dominant frequency and sampling frequency, the wavelet transform signal is reconstructed. Different decomposition levels are defined based on the principle that the dominant frequency exists in the low-frequency band. Assuming the wavelet transform decomposition level is m levels, m+1 frequency bands are generated, resulting in m+1 reconstructed signals.

[0090] The reconstructed signal is

[0091]

[0092] In the formula: X Lm For low-frequency reconstructed signals, X Hi It is a high-frequency reconstructed signal.

[0093] Calculate the energy of each reconstructed signal

[0094]

[0095] In the formula: E Lm For wavelet low-frequency energy; E Hi It represents the high-frequency energy of wavelets.

[0096] Calculate the proportion of wavelet energy P at the dominant frequency w

[0097]

[0098] 4) Energy matrix norm E n : The sum of squares of the wavelet energy proportions.

[0099]

[0100] In the formula: P HiThis represents the proportion of high-frequency wavelet energy.

[0101] 5) Entropy value of the energy matrix E e The calculation is as follows:

[0102]

[0103] In some embodiments of the present invention, optionally, the vibration characteristic parameters are combined to obtain a vibration characteristic matrix, including: combining the fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy proportion, energy matrix norm, and energy matrix entropy value to obtain a vibration characteristic matrix [A]. 100Hz P 100Hz P w E n E e ].

[0104] For example, for a three-phase transformer, based on the number of windings, 15 acceleration vibration sensors are arranged at five locations from bottom to top on windings A, B, and C: the bottom of the winding, 1 / 4 of the winding height, 2 / 4 of the winding height, 3 / 4 of the winding height, and the top of the winding. These sensors are named sequentially as measuring point A-1, measuring point A-2, ..., measuring point A-5, measuring point B-1, measuring point B-2, ..., measuring point B-5, measuring point C-1, measuring point C-2, ..., measuring point C-5. See [reference needed] Figure 3 As shown, Figure 3 This is a schematic diagram of the installation location (measuring points) of the vibration sensor. The circular dots on the winding are the measuring points. The three columns of measuring points from left to right are A, B and C, and each column is numbered 1-5 from top to bottom. For example, the leftmost column of measuring points is column A. The measuring point at the top of column A is measuring point A-1. The measuring points below measuring point A-1 in the same column are measuring points A-2, A-3, A-4 and A-5, respectively.

[0105] The factory preload of the winding is used as the standard preload. Simultaneously, the clamping pins of the tightly pressed winding are loosened to achieve the purpose of loosening the winding. Based on the actual situation, the standard preload, 90% standard preload, and 80% standard preload (i.e., normal state, 10% loosening, and 20% loosening) are named State 1, State 2, and State 3, respectively. A vibration testing system is used to collect vibration signals from the surface of the oil tank under different loosening states. To accumulate a large number of data samples, multiple sets of data are collected under the same state. See [link / reference] Figure 4 As shown, Figure 4The waveforms and spectra of different loosening states are shown, with state 1, state 2, and state 3 represented from top to bottom. It can be seen that as the degree of loosening increases, the vibration amplitude gradually increases, harmonic distortion gradually decreases, and the waveform becomes smoother, closer to a sine wave. Spectrally, the frequencies in the three states are mainly distributed between 0 and 1000 Hz. Under normal conditions, due to the larger preload, the vibration is smaller, but the noise and harmonic content are larger. As the degree of loosening increases and the vibration increases, the spectrum becomes simpler, mainly dominated by the 100 Hz fundamental frequency. Furthermore, to further determine the degree of loosening, it is often necessary to propose vibration characteristic parameters.

[0106] Therefore, the original vibration signal is processed to extract the fundamental frequency amplitude A. 100Hz Fundamental frequency weight P 100Hz The proportion of wavelet energy at the dominant frequency (P) w Energy matrix norm E n Entropy value of energy matrix E e There are a total of 5 vibration characteristic parameters, thus forming a vibration characteristic matrix [A] 100Hz P 100Hz P w E n E e ].

[0107] Finally, the vibration characteristic matrices for three states (state 1, state 2, and state 3) were obtained: P1 = [0.06, 36.73, 85.64, 0.86, 93.28], P2 = [0.08, 46.17, 86.99, 0.87, 8.55], and P3 = [0.88, 76.86, 91.23, 0.92, 141.06]. The vibration characteristic matrices show that as the degree of loosening increases, all characteristic parameters increase, especially in state 3 compared to states 1 and 2, where the increases are more significant. However, while the vibration characteristic parameters in state 2 increase compared to state 1, the increases are smaller, and under the influence of various factors, misjudgments are more likely.

[0108] The K-means clustering algorithm was used to classify different degrees of loosening. A K value of 3 was determined, indicating the existence of 3 cluster centers and 3 different loosening states. (See [link to relevant documentation]). Figure 5 As shown, Figure 5 The diagram shows the classification effect obtained by using the K-means clustering algorithm to classify different degrees of loosening. As can be seen from the diagram, the K-means clustering algorithm with multiple vibration feature parameters can distinguish different loosening states very well.

[0109] In some embodiments of the present invention, see optionally, see Figure 6 As shown, Figure 6This is the third flowchart of the transformer winding loosening detection method according to an embodiment of the present invention. Based on the vibration feature matrix, a clustering algorithm is used to calculate the loosening state parameters, including:

[0110] Step 31: Randomly generate at least two sets of initial cluster centers based on the preset initial clusters, where the preset number of initial clusters is equal to the number of initial cluster centers;

[0111] Step 32: Calculate the loosening state parameters based on the vibration feature matrix and at least two initial cluster centers.

[0112] In some embodiments of the present invention, optionally, the preset initial cluster may be an initial cluster estimated by the user, used to generate initial cluster centers based on the initial cluster, so as to achieve the initial conditions for using the clustering algorithm.

[0113] In some embodiments of the present invention, optionally, the preset initial cluster can be an empirical estimate summarized by the user for a specific transformer model and / or specific transformer usage, used to generate initial cluster centers based on the initial cluster, so as to achieve the initial conditions for using the clustering algorithm.

[0114] Understandably, the closer the empirical estimate is to the actual loosening type, the fewer iterations are needed in the subsequent process of calculating the loosening state parameters using clustering algorithms, and the lower the time and computing power costs are.

[0115] In some embodiments of the present invention, see optionally, see Figure 7 As shown, Figure 7 This is a schematic flowchart of the transformer winding loosening detection method according to an embodiment of the present invention. Based on the vibration feature matrix and at least two sets of initial cluster centers, loosening state parameters are calculated, including:

[0116] Step 41: Calculation steps: Calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the corresponding value of the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center.

[0117] Step 42: Update step: Divide the vibration feature matrix into different types according to the total Euclidean distance, obtain sample clusters, and obtain updated cluster centers based on the sample clusters;

[0118] Step 43: Perform a second check to determine whether the updated cluster centers are consistent with the initial cluster centers;

[0119] Step 44: If the result of the second detection is that the updated cluster centers are inconsistent with the initial cluster centers, replace the old initial cluster centers with the updated cluster centers to obtain new cluster centers, return to the update step, and continue until the result of the second detection is that the updated cluster centers are consistent with the latest cluster centers.

[0120] Step 45: If the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center, determine the loosening state parameters based on the updated cluster center.

[0121] In mathematics, Euclidean distance, or Euclidean metric, is the "ordinary" (i.e., straight-line) distance between two points in Euclidean space. Using this distance, Euclidean space becomes a metric space. The associated norm is called the Euclidean norm. Earlier literature referred to it as the Pythagorean metric.

[0122] In this embodiment of the invention, the preset initial cluster has a value corresponding to each vibration feature parameter in the vibration feature matrix when it is preset. To calculate the total Euclidean distance between a vibration feature matrix and a preset initial cluster, it is necessary to calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the preset initial cluster one by one, and then add the obtained Euclidean distances together.

[0123] For example, the preset initial clustering includes [a, b], and the vibration feature matrix is ​​[A]. 100Hz P 100Hz ], where a and A 100Hz Correspondence, b and P 100Hz Correspondingly, the total Euclidean distance between the preset initial cluster and the feature matrix is ​​calculated, which is: obtaining a and A 100Hz The Euclidean distance between b and P 100Hz The Euclidean distance between them, and then the Euclidean distance between a and A. 100Hz The Euclidean distance between b and P 100Hz The Euclidean distances between the two clusters are summed to obtain the total Euclidean distance between the preset initial cluster and the feature matrix.

[0124] In this embodiment of the invention, the total Euclidean distance is used to characterize the similarity between each group of data in the vibration feature matrix and the initial cluster center, and the cluster center with the highest similarity is taken as the sample cluster.

[0125] For example, the loosening state parameters are calculated using the k-means clustering algorithm based on the vibration feature matrix, including the following steps (a to f):

[0126] a. Determine K initial clusters based on different degrees of looseness in the windings;

[0127] b. Randomly generate K values ​​as initial cluster centers;

[0128] c. Calculate the Euclidean distance from each data point to the K cluster centers;

[0129] The formula for calculating Euclidean distance is:

[0130] In the formula: i is the number of cluster points; x is the cluster point; y is the coordinate of the random cluster center.

[0131] d. Select the cluster center with the highest similarity as the cluster of the samples;

[0132] e. Calculate the average value for each cluster, and then update it as the new cluster centers;

[0133] New cluster centers (average)

[0134] In the formula: i is the number of cluster points; y is the object;

[0135] f. Repeat steps c to e until the cluster center no longer changes, and finally determine the K different loose states of the winding.

[0136] This invention provides a device for detecting loose transformer windings, see [link to relevant documentation]. Figure 8 As shown, Figure 8 This is a schematic block diagram of a transformer winding loosening detection device according to an embodiment of the present invention. The transformer winding loosening detection device 80 includes:

[0137] The acquisition module 81 is used to acquire the vibration information of the transformer collected by the detection equipment;

[0138] Execution module 82 is used to determine the vibration feature matrix based on the vibration information;

[0139] The execution module 82 is also used to calculate the loosening state parameters based on the vibration feature matrix using a clustering algorithm. The loosening state parameters are used to indicate the loosening state of the windings of the monitored transformer.

[0140] In some embodiments of the present invention, optionally,

[0141] The execution module 82 is also used to calculate the vibration characteristic parameters based on the vibration information;

[0142] The execution module 82 is also used to obtain the vibration feature matrix based on the vibration feature parameters.

[0143] In some embodiments of the present invention, the vibration characteristic parameters optionally include at least one of the following:

[0144] Fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy ratio, energy matrix norm, and energy matrix entropy;

[0145] The execution module 82 is also used to combine the vibration characteristic parameters to obtain the vibration characteristic matrix.

[0146] In some embodiments of the present invention, optionally,

[0147] The execution module 82 is further configured to randomly generate at least two sets of initial cluster centers based on a preset initial cluster, wherein the number of preset initial clusters is equal to the number of initial cluster centers;

[0148] The execution module 82 is further configured to calculate the loosening state parameters based on the vibration feature matrix and at least two sets of the initial cluster centers.

[0149] In some embodiments of the present invention, optionally,

[0150] The execution module 82 is further configured to perform the following calculation steps: calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center;

[0151] The execution module 82 is also used for the update step: classifying the types of each vibration feature matrix according to the total Euclidean distance value to obtain sample clusters, and obtaining updated cluster centers according to the sample clusters;

[0152] The execution module 82 is further configured to perform a second detection on whether the updated cluster centers are consistent with the initial cluster centers;

[0153] The execution module 82 is further configured to, if the detection result of the second detection is that the updated cluster center is inconsistent with the initial cluster center, replace the old initial cluster center with the updated cluster center to obtain a new cluster center, return to the update step, and continue until the detection result of the second detection is consistent with the latest cluster center.

[0154] The execution module 82 is further configured to determine the loosening state parameter based on the updated cluster center if the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center.

[0155] The transformer winding loosening detection device provided in this application embodiment can achieve Figures 1 to 7 The various processes implemented in the method embodiments achieve the same technical effect, and will not be described again here to avoid repetition.

[0156] This invention provides a detection system for loose transformer windings, see [link to relevant documentation]. Figure 9 As shown, Figure 9 This is a schematic diagram of a transformer winding loosening detection system according to an embodiment of the present invention. The transformer winding loosening detection system 90 includes:

[0157] Vibration sensor 91, which detects loosening of transformer windings, is installed on the transformer.

[0158] The transformer winding loosening detection device 92 is connected to a vibration acquisition sensor. It is used to determine the vibration feature matrix based on the vibration information of the monitored transformer obtained by the vibration acquisition sensor. Based on the vibration feature matrix, a clustering algorithm is used to calculate the loosening state parameters, which are used to indicate the loosening state of the winding of the monitored transformer.

[0159] The transformer winding loosening detection system provided in this application embodiment can achieve Figures 1 to 7 The various processes implemented in the method embodiments achieve the same technical effect, and will not be described again here to avoid repetition.

[0160] This invention provides an electronic device 100, see [link to documentation]. Figure 10 As shown, Figure 10 This is a schematic block diagram of an electronic device 100 according to an embodiment of the present invention, including a processor 101, a memory 102, and a program or instructions stored in the memory 102 and executable on the processor 101. When the program or instructions are executed by the processor, they implement the steps in any of the transformer winding loosening detection methods of the present invention.

[0161] This invention provides a readable storage medium on which a program or instruction is stored. When the program or instruction is executed by a processor, it implements the various processes of the embodiment of the transformer winding loosening detection method as described above, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0162] The readable storage medium mentioned above includes, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0163] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A method for detecting loose transformer windings, characterized in that, include: Obtain vibration information of the transformer collected by the detection equipment; Based on the vibration information, determine the vibration feature matrix; Based on the vibration feature matrix, a clustering algorithm is used to calculate the loosening state parameters, which are used to indicate the loosening state of the transformer windings. The loosening state parameters are calculated using a clustering algorithm based on the vibration feature matrix, including: At least two initial cluster centers are randomly generated based on a preset initial cluster, wherein the number of preset initial clusters is equal to the number of initial cluster centers; The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers. The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers, including: Calculation steps: Calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center; Update steps: Divide the vibration feature matrices into types according to the total Euclidean distance value to obtain sample clusters, and obtain updated cluster centers based on the sample clusters; A second check is performed to determine whether the updated cluster centers are consistent with the initial cluster centers; If the result of the second detection is that the updated cluster center is inconsistent with the initial cluster center, the updated cluster center is used to replace the old initial cluster center to obtain a new cluster center, and the update step is returned until the result of the second detection is that the updated cluster center is consistent with the latest cluster center. If the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center, the loosening state parameter is determined based on the updated cluster center.

2. The method for detecting loose transformer windings according to claim 1, characterized in that: Based on the vibration information, the vibration feature matrix is ​​determined, including: Based on the vibration information, the vibration characteristic parameters are calculated; The vibration characteristic matrix is ​​obtained based on the vibration characteristic parameters.

3. The method for detecting loose transformer windings according to claim 2, characterized in that: The vibration characteristic parameter includes at least one of the following: Fundamental frequency amplitude, fundamental frequency weight, dominant frequency wavelet energy ratio, energy matrix norm, and energy matrix entropy; Based on the vibration characteristic parameters, the vibration characteristic matrix is ​​obtained, including: The vibration characteristic parameters are combined to obtain the vibration characteristic matrix.

4. A device for detecting loose transformer windings, characterized in that, include: The acquisition module is used to acquire vibration information of the transformer collected by the detection equipment; An execution module is used to determine a vibration feature matrix based on the vibration information; The execution module is further configured to calculate loosening state parameters based on the vibration feature matrix using a clustering algorithm, the loosening state parameters being used to indicate the loosening state of the transformer windings; The loosening state parameters are calculated using a clustering algorithm based on the vibration feature matrix, including: At least two initial cluster centers are randomly generated based on a preset initial cluster, wherein the number of preset initial clusters is equal to the number of initial cluster centers; The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers. The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers, including: Calculation steps: Calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center; Update steps: Divide the vibration feature matrices into types according to the total Euclidean distance value to obtain sample clusters, and obtain updated cluster centers based on the sample clusters; A second check is performed to determine whether the updated cluster centers are consistent with the initial cluster centers; If the result of the second detection is that the updated cluster center is inconsistent with the initial cluster center, the updated cluster center is used to replace the old initial cluster center to obtain a new cluster center, and the update step is returned until the result of the second detection is that the updated cluster center is consistent with the latest cluster center. If the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center, the loosening state parameter is determined based on the updated cluster center.

5. The transformer winding loosening detection device according to claim 4, characterized in that: The execution module is also used to calculate vibration characteristic parameters based on the vibration information; The execution module is further configured to obtain the vibration feature matrix based on the vibration feature parameters.

6. A detection system for loose transformer windings, characterized in that, include: A vibration sensor for monitoring loose transformer windings is installed on the transformer. A transformer winding loosening detection device is connected to the vibration acquisition sensor and is used to determine a vibration feature matrix based on the vibration information of the monitored transformer acquired by the vibration acquisition sensor; based on the vibration feature matrix, a clustering algorithm is used to calculate loosening state parameters, which are used to indicate the loosening state of the transformer windings; The loosening state parameters are calculated using a clustering algorithm based on the vibration feature matrix, including: At least two initial cluster centers are randomly generated based on a preset initial cluster, wherein the number of preset initial clusters is equal to the number of initial cluster centers; The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers. The loosening state parameters are calculated based on the vibration feature matrix and at least two sets of initial cluster centers, including: Calculation steps: Calculate the Euclidean distance between each vibration feature parameter in the vibration feature matrix and the value corresponding to the vibration feature parameter in each initial cluster center, and add the Euclidean distances between each vibration feature parameter in the same vibration feature matrix and the same initial cluster center to obtain the total Euclidean distance value corresponding one-to-one with the initial cluster center; Update steps: Divide the vibration feature matrices into types according to the total Euclidean distance value to obtain sample clusters, and obtain updated cluster centers based on the sample clusters; A second check is performed to determine whether the updated cluster centers are consistent with the initial cluster centers; If the result of the second detection is that the updated cluster center is inconsistent with the initial cluster center, the updated cluster center is used to replace the old initial cluster center to obtain a new cluster center, and the update step is returned until the result of the second detection is that the updated cluster center is consistent with the latest cluster center. If the detection result of the second detection is that the updated cluster center is consistent with the initial cluster center, the loosening state parameter is determined based on the updated cluster center.

7. An electronic device, characterized in that: It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps in the method for detecting loose transformer windings as described in any one of claims 1 to 3.

8. A readable storage medium, characterized in that: The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps in the method for detecting loose transformer windings as described in any one of claims 1 to 3.

Images