Transformer vibration characteristic selection method, system and equipment and storage medium
By calculating the mutual information and entropy value of the transformer vibration characteristics, the most relevant and important features are selected, and the problem of ignoring redundancy in the prior art is solved, the most preferred selection of the transformer vibration characteristics is achieved, and the accuracy of monitoring is improved degree and efficiency.
Patent Information
- Application Number
- CN202411144473.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-15
AI Technical Summary
The existing transformer vibration feature dimensionality reduction method only considers the correlation between the features and the original data, ignores the redundancy between the vibration feature to be selected and the selected features, resulting in the inability to achieve the most preferred choice of the vibration feature, affecting online monitoring. accuracy and efficiency.
By obtaining the different vibration characteristics of the transformer, calculating mutual information and entropy values, obtaining the characteristic correlation and importance, filtering out the most relevant and important features, gradually deleting the redundant features, and realizing the target vibration characteristics of the transformer.
The redundancy between vibration characteristics is reduced, the most preferred choice of the transformer vibration characteristics is achieved, and the accuracy and efficiency of online monitoring is improved.
Smart Images

Figure CN118964944A_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to the field of transformer vibration feature selection, and in particular to a transformer vibration feature selection method, system, device and storage medium. [Background technology]
[0002] The vibration feature data of transformer online monitoring is huge, and there is a large amount of redundant and unrelated data between vibration features. If the vibration features are not selected differently, the accuracy and efficiency of transformer online monitoring will be affected. The existing vibration feature dimensionality reduction method only considers the correlation between the feature and the original data, but ignores the redundancy between the vibration features to be selected and the selected features, so it is impossible to achieve the optimal selection of vibration features. [Summary of the invention]
[0003] In view of this, the present invention provides a transformer vibration feature selection method, system, device and storage medium.
[0004] The specific technical scheme of the first embodiment of the present invention is: a transformer vibration feature selection method, the method comprising: obtaining different vibration features of the transformer to construct a first vibration feature set; obtaining the mutual information between the vibration features in the first vibration feature set; obtaining the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; obtaining the vibration feature with the largest first feature correlation as the first target vibration feature; obtaining the feature importance of all vibration features in the second vibration feature set; the second vibration feature set is the feature set after deleting the target vibration feature from the first vibration feature set; obtaining a target feature set in the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature; the first target vibration feature and the target feature set constitute the target vibration feature of the transformer.
[0005] In a specific embodiment, obtaining a target feature set from the second vibration feature set according to the feature importance includes: obtaining a vibration feature with the largest feature importance in the second vibration feature set as a vibration feature to be selected; determining whether the feature importance of the vibration feature to be selected is greater than or equal to a preset threshold; if greater, the vibration feature to be selected is a second target vibration feature; removing the second target vibration feature from the second vibration feature set to obtain a third vibration feature set; using the third vibration feature set as the second vibration feature set, and returning to the step of obtaining the feature importance of all vibration features in the second vibration feature set, until the feature importance of the vibration feature to be selected is less than the preset threshold, then all second target vibration features constitute the target feature set.
[0006] Preferably, acquiring the feature importance of all vibration features in the second vibration feature set includes: obtaining the feature importance of each vibration feature according to the second feature correlation and feature redundancy of each vibration feature in the second vibration feature set.
[0007] Preferably, the second feature correlation is obtained using the following formula:
[0008]
[0009] Among them, FRL (f I ) is the vibration characteristic f I The characteristic correlation, H(f I ) is the vibration characteristic f I The entropy value of n is the total number of vibration features, R(f I ;f J ) is the vibration characteristic f I With vibration characteristics f J The mutual information of .
[0010] Preferably, the feature redundancy is obtained using the following formula:
[0011]
[0012] Among them, FRD(f I ) is the vibration characteristic f I The feature redundancy, R(f I ; U i ) is the vibration characteristic f I The vibration feature f is removed from the second vibration feature set I The mutual information of all vibrational features except
[0013] Preferably, the feature importance is obtained using the following formula:
[0014]
[0015] Among them, FI(f I ) is the vibration characteristic f I The feature importance, FRL(f I ) is the vibration characteristic f I The feature correlation, FRD(f I ) is the vibration characteristic f I The feature redundancy, FRL(f J ) is the vibration characteristic f J The feature correlation, FRD(f J ) is the vibration characteristic f J feature redundancy.
[0016] Preferably, the mutual information between the vibration features is obtained using the following formula:
[0017]
[0018] Among them, p() is the probability of the random variable under different possible values, H(f I ) is the vibration characteristic f I The entropy value, H(f I |f J ) is the vibration characteristic f J Vibration characteristics under the condition of f I The conditional entropy, f I,i is the vibration feature f in the i-th group of vibration data I The value of f J,j is the vibration feature f in the Jth group of vibration data J The value of .
[0019] The specific technical scheme of the second embodiment of the present invention is: a transformer vibration feature selection system, the system comprising: a vibration feature acquisition module, a mutual information acquisition module, a feature correlation acquisition module, a target vibration feature acquisition module, a feature importance acquisition module, a target feature set acquisition module and a feature output module; the vibration feature acquisition module is used to acquire different vibration features of the transformer and construct a first vibration feature set; the mutual information acquisition module is used to acquire the mutual information between the vibration features in the first vibration feature set; the feature correlation acquisition module is used to acquire the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; the target vibration feature acquisition module is used to acquire the vibration feature with the largest first feature correlation as the first target vibration feature; the feature importance acquisition module is used to acquire the feature importance of all vibration features in the second vibration feature set; the second vibration feature set is the feature set after deleting the target vibration feature from the first vibration feature set; the target feature set acquisition module is used to acquire the target feature set in the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature; the feature output module is used to form the target vibration feature of the transformer with the first target vibration feature and the target feature set.
[0020] The specific technical solution of the third embodiment of the present invention is: a transformer vibration feature selection device, including a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method described in any one of the first embodiments of the present application.
[0021] The specific technical solution of the fourth embodiment of the present invention is: a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to perform the steps of the method described in any one of the first embodiments of the present application.
[0022] Implementing the embodiments of the present invention will have the following beneficial effects:
[0023] The present invention obtains the mutual information between vibration features in a first vibration feature set; obtains the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; selects the vibration feature with the largest first feature correlation as the target vibration feature, and further screens the target vibration feature according to the feature importance among the remaining vibration features, thereby reducing the redundancy between the vibration features, achieving dimensionality reduction of the transformer vibration features, and obtaining the optimal vibration feature.
Brief Description of the Drawings
[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0025] Figure 1 A flow chart of the steps of the transformer vibration feature selection method;
[0026] Figure 2 The schematic diagram of the structure of the transformer vibration feature selection system;
[0027] Figure 3 A diagram of the internal structure of a computer device;
[0028] Among them, 201, vibration feature acquisition module; 202, mutual information acquisition module; 203, feature correlation acquisition module; 204, target vibration feature acquisition module; 205, feature importance acquisition module; 206, target feature set acquisition module; 207, feature output module. [Specific implementation method]
[0029] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0030] The terms "first", "second", etc. in the specification, claims and drawings of this application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally includes steps or modules that are not listed, or optionally includes other steps or modules that are inherent to these processes, methods, products or devices.
[0031] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0032] The vibration characteristics of the transformer mainly include the increase of noise level, spectrum change and vibration mode under specific conditions. Increase of noise level: Under DC bias conditions, the noise level of the transformer will increase significantly. For example, the noise is 69dB when the neutral point is not grounded, and the noise increases to 85.1dB after the neutral point is grounded, which is actually 16.1dB higher, which is basically consistent with the theoretical increase in noise level. Spectrum change: The vibration spectrum of the transformer will change under specific conditions. For example, under DC bias conditions, the noise spectrum is consistent with the typical spectrum of DC bias noise, which can be measured and analyzed by a spectrum analyzer. Vibration mode under specific conditions: Vibration characteristics under no-load conditions: Under no-load conditions, deformation or damage of the transformer winding core will cause changes in vibration characteristics. By online monitoring of the transformer body vibration, the winding and core conditions can be reflected. The no-load vibration characteristics are the basis for judging whether the core is faulty. Vibration characteristics under short-circuit conditions: Under short-circuit conditions, the vibration characteristics of the transformer will be affected by the short-circuit current amplitude, the winding natural frequency and the cumulative effect of the short circuit. As the short-circuit current increases, the short-circuit electromotive force on the winding increases, resulting in an increase in the nonlinear effect of the paperboard, an increase in the effective value of harmonic vibration and vibration entropy, and a decrease in the main frequency energy ratio. When the short-circuit current is large, the winding cumulative effect is obvious. As the number of short-circuit impacts increases, the vibration entropy decreases and the main frequency energy ratio increases. These changes in vibration characteristics can help diagnose potential problems inside the transformer, such as looseness or deformation of the winding or core, so that maintenance measures can be taken in a timely manner to ensure the safe operation of the transformer.
[0033] See also Figure 1, is a flowchart of the steps of a transformer vibration feature selection method in the first embodiment of the present application, thereby reducing the redundancy between vibration features and achieving dimensionality reduction of transformer vibration features. The method includes:
[0034] Step 101: Acquire different vibration characteristics of the transformer and construct a first vibration characteristic set;
[0035] Step 102, obtaining mutual information between vibration features in the first vibration feature set;
[0036] Step 103: acquiring a first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature;
[0037] Step 104: Acquire the vibration feature with the largest first feature correlation as the first target vibration feature;
[0038] Step 105, obtaining the feature importance of all vibration features in a second vibration feature set; the second vibration feature set is a feature set obtained by deleting the target vibration feature from the first vibration feature set;
[0039] Step 106: acquiring a target feature set from the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature;
[0040] Step 107: The first target vibration feature and the target feature set constitute a target vibration feature of the transformer.
[0041] Specifically, the first vibration feature set F = {f1, f2, ..., f n}, a total of n vibration features, each vibration feature f i Containing m groups of vibration data; obtaining the mutual information R(f I ;f J ); obtain the feature correlation FRL of all vibration features according to the mutual information and the entropy value of each vibration feature; put the vibration feature with the largest feature correlation into the empty set U, and the empty set is used as the vibration feature subset U, and delete the vibration feature with the largest feature correlation in the first vibration feature set; screen the remaining vibration features in the first vibration feature set again according to the feature importance of each vibration feature, and put the vibration features obtained after screening into the vibration feature subset U. After the screening, the vibration features in the vibration feature subset U are all the target vibration features of the transformer.
[0042] Mutual information is a useful information measure in information theory. It can be viewed as the amount of information contained in a random variable about another random variable, or the uncertainty of a random variable reduced by knowing another random variable.
[0043] Feature relevance refers to the degree of association between different features in a data set. It is a common method in data analysis to understand the relationship between features, so as to better understand the structure of the data set and the importance of features. Feature relevance can be measured by correlation coefficients. Commonly used correlation coefficients include Pearson correlation coefficient, Spearman correlation coefficient, and Chebyshev correlation coefficient. In machine learning, feature relevance is an important concept that involves how to select and utilize features in a data set to optimize the performance of the model. Some features may be highly relevant to the problem to be solved, while other features may be redundant or irrelevant to the problem. Attributes that are useful for the current learning task are called relevant features, while useless attributes are called irrelevant features. In addition, relevance is also a key concept in search engine optimization, which refers to the degree of association between search requests (such as keywords) and search results (such as web pages). Search engines provide the most relevant results to user queries by calculating relevance. Relevance calculation is one of the most important technologies of search engines. By using calculation methods such as vector space models or probability models, search engines can improve the accuracy of user queries. In summary, feature relevance is not only a concept in data analysis, but also an important tool in machine learning and search engine optimization. It is used to measure and optimize the relationship between features and target variables in a data set, thereby improving the accuracy of model predictions and the relevance of search engine results.
[0044] The method in this embodiment obtains the mutual information between the vibration features in the first vibration feature set; obtains the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; and selects the vibration feature with the largest first feature correlation as the target vibration feature, and further screens the target vibration feature among the remaining vibration features according to the feature importance, thereby reducing the redundancy between the vibration features, achieving dimensionality reduction of the transformer vibration features, and obtaining the optimal vibration feature.
[0045] In a specific embodiment, obtaining a target feature set from the second vibration feature set according to the feature importance includes: obtaining a vibration feature with the largest feature importance in the second vibration feature set as a vibration feature to be selected; determining whether the feature importance of the vibration feature to be selected is greater than or equal to a preset threshold; if greater, the vibration feature to be selected is a second target vibration feature; removing the second target vibration feature from the second vibration feature set to obtain a third vibration feature set; using the third vibration feature set as the second vibration feature set, and returning to the step of obtaining the feature importance of all vibration features in the second vibration feature set, until the feature importance of the vibration feature to be selected is less than the preset threshold, then all second target vibration features constitute the target feature set.
[0046] Specifically, the vibration feature f with the largest feature importance in the second vibration feature set is obtained. I , vibration characteristics f I When the feature importance FI is the largest and greater than 1, the vibration feature f I Add to the vibration feature subset U, and increase the feature subset dimension by 1; remove the vibration feature f in the second vibration feature set I , obtain a third vibration feature set, and repeatedly obtain the vibration feature with the largest feature importance in the third vibration feature set, until there is no vibration feature with a feature importance greater than 1, or there is no vibration feature in the vibration feature set, the vibration feature subset U is the target feature set.
[0047] In a specific embodiment, obtaining the feature importance of all vibration features in the second vibration feature set includes: obtaining the feature importance of each vibration feature according to the second feature correlation and feature redundancy of each vibration feature in the second vibration feature set. Specifically, the feature correlation is obtained according to the feature correlation and feature redundancy, so as to accurately screen the importance of the vibration features.
[0048] In a specific embodiment, the second feature correlation is obtained using the following formula:
[0049]
[0050] Among them, FRL (f I ) is the vibration characteristic f I The characteristic correlation, H(f I ) is the vibration characteristic f I The entropy value of n is the total number of vibration features, R(f I ;f J ) is the vibration characteristic f I With vibration characteristics f J The mutual information of .
[0051] In a specific embodiment, the feature redundancy is obtained using the following formula:
[0052]
[0053] Among them, FRD(f I ) is the vibration characteristic f I The feature redundancy, R(f I ; U i ) is the vibration characteristic f I The vibration feature f is removed from the second vibration feature set I The mutual information of all vibrational features except
[0054] In a specific embodiment, the feature importance is obtained using the following formula:
[0055]
[0056] Among them, FI(f I ) is the vibration characteristic f I The feature importance, FRL(f I ) is the vibration characteristic f I The feature correlation, FRD(f I ) is the vibration characteristic f I The feature redundancy, FRL(f J ) is the vibration characteristic f J The feature correlation, FRD(f J ) is the vibration characteristic f J feature redundancy.
[0057] In a specific embodiment, the mutual information between the vibration features is obtained using the following formula:
[0058]
[0059] Among them, p() is the probability of the random variable under different possible values, H(f I ) is the vibration characteristic f I The entropy value, H(f I |f J ) is the vibration characteristic f J Vibration characteristics under the condition of f I The conditional entropy, f I,i is the vibration feature f in the i-th group of vibration data I The value of f J,j is the vibration feature f in the Jth group of vibration data J The value of .
[0060] In a specific embodiment, the specific process of the method is:
[0061] Obtain the first vibration feature set F = {f1, f2, ..., fn} of the transformer, with a total of n vibration features, and each vibration feature value f i Contains m sets of vibration data. Initialize feature subset U to an empty set;
[0062] Calculate the mutual information R(f I ;f J ), forming the mutual information matrix R, the calculation formula is as follows:
[0063]
[0064] In the formula, p() is the probability of the random variable under different possible values, H(f I ) is the vibration characteristic f IThe entropy value, H(f I |f J ) is the vibration characteristic f J Vibration characteristics under the condition of f I The conditional entropy, f I,i is the vibration feature f in the i-th group of vibration data I The value of f J,j is the vibration feature f in the Jth group of vibration data J The value of .
[0065] Calculate the feature correlation FRL of all vibration features. The calculation formula is as follows:
[0066]
[0067] Among them, FRL (f I ) is the vibration characteristic f I The characteristic correlation, H(f I ) is the vibration characteristic f I The entropy value of n is the total number of vibration features, R(f I ;f J ) is the vibration characteristic f I With vibration characteristics f J The mutual information of .
[0068] The feature with the largest FRL value among the n vibration features is selected as the first feature U1 in the feature subset U, the dimension of the feature subset is x=1, the first vibration feature set F is updated, the features added to the feature subset are deleted, and n-1 vibration features are left, and the second vibration feature set is obtained;
[0069] The second vibration feature set is selected, and the feature redundancy FRD of each vibration feature in the second vibration feature set is calculated. The calculation formula is as follows:
[0070]
[0071] In the formula, FRD(f I ) is the vibration characteristic f I The feature redundancy, R(f I ; U i ) is the vibration characteristic f I The vibration feature f is removed from the second vibration feature set I After obtaining the FRL and FRD values of all vibration feature quantities in the second vibration feature set, the feature importance FI of each feature in the second vibration feature set can be calculated, and the calculation formula is as follows:
[0072]
[0073] Among them, FI(fI ) is the vibration characteristic f I The feature importance, FRL(f I ) is the vibration characteristic f I The feature correlation, FRD(f I ) is the vibration characteristic f I The feature redundancy, FRL(f J ) is the vibration characteristic f J The feature correlation, FRD(f J ) is the vibration characteristic f J feature redundancy. If the feature importance FI of all vibration features is calculated, the vibration feature with the largest FI and greater than 1 is selected and added to the vibration feature subset U, and the feature subset dimension x is increased by 1; the second vibration feature set is updated, that is, the vibration feature with the largest feature importance is removed to obtain the third vibration feature set, and the screening is repeated in the third vibration feature set until the calculated feature importance is less than 1 or the vibration feature set is an empty set, the loop is terminated, and finally the feature subset U is selected.
[0074] In the specific embodiments, see Figure 2 , a schematic diagram of the structure of a transformer vibration feature selection system is provided for the second embodiment of the present application, the system comprising: a vibration feature acquisition module 201, a mutual information acquisition module 202, a feature correlation acquisition module 203, a target vibration feature acquisition module 204, a feature importance acquisition module 205, a target feature set acquisition module 206 and a feature output module 207; the vibration feature acquisition module 201 is used to acquire different vibration features of the transformer and construct a first vibration feature set; the mutual information acquisition module 202 is used to acquire the mutual information between the vibration features in the first vibration feature set; the feature correlation acquisition module 203 is used to acquire the first vibration feature of each vibration feature according to the mutual information and the entropy value of each vibration feature. A feature correlation; the target vibration feature acquisition module 204 is used to obtain the vibration feature with the largest first feature correlation as the first target vibration feature; the feature importance acquisition module 205 is used to obtain the feature importance of all vibration features in the second vibration feature set; the second vibration feature set is the feature set after deleting the target vibration feature from the first vibration feature set; the target feature set acquisition module 206 is used to obtain a target feature set in the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature; the feature output module 207 is used to form the target vibration feature of the transformer with the first target vibration feature and the target feature set.
[0075] The system in this embodiment obtains the mutual information between the vibration features in the first vibration feature set; obtains the first feature correlation of each vibration feature based on the mutual information and the entropy value of each vibration feature; and selects the vibration feature with the largest first feature correlation as the target vibration feature, and further screens the target vibration feature among the remaining vibration features based on the feature importance, thereby reducing the redundancy between the vibration features, achieving dimensionality reduction of the transformer vibration features, and obtaining the optimal vibration feature.
[0076] In a specific embodiment, the third embodiment of the present application provides a transformer vibration feature selection device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described in any one of the first embodiments of the present application. The device in this embodiment obtains the mutual information between the vibration features in the first vibration feature set; obtains the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; and selects the vibration feature with the largest first feature correlation as the target vibration feature, and further screens the target vibration feature according to the feature importance among the remaining vibration features, thereby reducing the redundancy between the vibration features, achieving dimensionality reduction of the transformer vibration features, and obtaining the optimal vibration feature.
[0077] In a specific embodiment, the fourth embodiment of the present application provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the processor executes the steps of the method described in any one of the first embodiments of the present application. The storage medium in this embodiment obtains the mutual information between the vibration features in the first vibration feature set; obtains the first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; and selects the vibration feature with the largest first feature correlation as the target vibration feature, and further screens the target vibration feature according to the feature importance among the remaining vibration features, thereby reducing the redundancy between the vibration features, achieving dimensionality reduction of the transformer vibration features, and obtaining the optimal vibration feature.
[0078] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0079] Figure 3 The internal structure of a computer device in one embodiment is shown. The computer device can be a terminal or a server. Figure 3The computer device includes a processor, a memory, a boot link setting module, a jump module, a reservation module, a display module, and a confirmation module connected through a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement the method in this embodiment. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can implement the method in this embodiment. Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0080] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
[0081] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.
Claims
1. A transformer vibration feature selection method, characterized in that: The method comprises: Acquire different vibration characteristics of the transformer and construct a first vibration characteristic set; Obtaining mutual information between vibration features in the first vibration feature set; Acquire a first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; Acquire the vibration feature with the greatest first feature correlation as the first target vibration feature; Acquiring feature importances of all vibration features in a second vibration feature set; the second vibration feature set is a feature set obtained by deleting the target vibration feature from the first vibration feature set; Acquire a target feature set from the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature; The first target vibration feature and the target feature set constitute a target vibration feature of the transformer.
2. The transformer vibration feature selection method according to claim 1, characterized in that: The acquiring a target feature set from the second vibration feature set according to the feature importance comprises: Acquire a vibration feature with the greatest feature importance in the second vibration feature set as a vibration feature to be selected; Determining whether the feature importance of the selected vibration feature is greater than or equal to a preset threshold; If it is greater than, the vibration feature to be selected is the second target vibration feature; Remove the second target vibration feature from the second vibration feature set to obtain a third vibration feature set; The third vibration feature set is used as the second vibration feature set, and the step of obtaining the feature importance of all vibration features in the second vibration feature set is returned until the feature importance of the selected vibration feature is less than the preset threshold, and all second target vibration features constitute the target feature set.
3. The transformer vibration feature selection method according to claim 2, characterized in that: Obtaining the feature importance of all vibration features in the second vibration feature set, including: The feature importance of each vibration feature is obtained according to the second feature correlation and feature redundancy of each vibration feature in the second vibration feature set.
4. The transformer vibration feature selection method according to claim 3, characterized in that: The second feature relevance is obtained using the following formula: Among them, FRL (f I ) is the vibration characteristic f I The characteristic correlation, H(f I ) is the vibration characteristic f I The entropy value of n is the total number of vibration features, R(f I ;f J ) is the vibration characteristic f I With vibration characteristics f J The mutual information of .
5. The transformer vibration feature selection method according to claim 3, characterized in that: The feature redundancy is obtained using the following formula: Among them, FRD(f I ) is the vibration characteristic f I The feature redundancy, R(f I ; U i ) is the vibration characteristic f I The vibration feature f is removed from the second vibration feature set I The mutual information of all vibrational features except 6. The transformer vibration feature selection method according to claim 3, characterized in that: The feature importance is obtained using the following formula: Among them, FI(f I ) is the vibration characteristic f I The feature importance, FRL(f I ) is the vibration characteristic f I The feature correlation, FRD(f I ) is the vibration characteristic f I The feature redundancy, FRL(f J ) is the vibration characteristic f J The feature correlation, FRD(f J ) is the vibration characteristic f J feature redundancy.
7. The transformer vibration feature selection method according to claim 1, characterized in that: The mutual information between the vibration features is obtained using the following formula: Among them, p() is the probability of the random variable under different possible values, H(f I ) is the vibration characteristic f I The entropy value, H(f I |f J ) is the vibration characteristic f J Vibration characteristics under the condition of f I The conditional entropy, f I,i is the vibration feature f in the i-th group of vibration data I The value of f J,j is the vibration feature f in the Jth group of vibration data J The value of .
8. A transformer vibration feature selection system, characterized in that: The system comprises: a vibration feature acquisition module, a mutual information acquisition module, a feature correlation acquisition module, a target vibration feature acquisition module, a feature importance acquisition module, a target feature set acquisition module and a feature output module; The vibration feature acquisition module is used to acquire different vibration features of the transformer and construct a first vibration feature set; The mutual information acquisition module is used to acquire mutual information between vibration features in the first vibration feature set; The feature correlation acquisition module is used to acquire a first feature correlation of each vibration feature according to the mutual information and the entropy value of each vibration feature; The target vibration feature acquisition module is used to acquire the vibration feature with the largest first feature correlation as the first target vibration feature; The feature importance acquisition module is used to acquire the feature importance of all vibration features in a second vibration feature set; the second vibration feature set is a feature set obtained by deleting the target vibration feature from the first vibration feature set; The target feature set acquisition module is used to acquire a target feature set from the second vibration feature set according to the feature importance; the target feature set includes at least one target vibration feature; The feature output module is used to form the target vibration feature of the transformer by combining the first target vibration feature and the target feature set.
9. A transformer vibration feature selection device, comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.
Citation Information
Patent Citations
An unsupervised feature selection method based on mutual information and fractal dimension
CN108985462A
Unsupervised feature selection method for vibration data of transformer
CN110598760A
Mutual information transformer vibration signal feature selection method based on minimum redundancy
CN116561583A
