CN114492542A - Method and device for detecting running state of on-load tap-changer - Google Patents
Method and device for detecting running state of on-load tap-changer Download PDFInfo
- Publication number
- CN114492542A CN114492542A CN202210352722.8A CN202210352722A CN114492542A CN 114492542 A CN114492542 A CN 114492542A CN 202210352722 A CN202210352722 A CN 202210352722A CN 114492542 A CN114492542 A CN 114492542A
- Authority
- CN
- China
- Prior art keywords
- changer
- load tap
- fault
- sample
- vibration signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
Abstract
The application provides a method and a device for detecting the running state of an on-load tap-changer, comprising the following steps: extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates; extracting a feature vector of the vibration signal; inputting the feature vector into each Gaussian mixture model to determine the probability distribution condition of the feature vector; determining the actual running state of the on-load tap-changer by using a Bayesian classifier based on the probability distribution condition determined by the characteristic vector and each Gaussian mixture model; and when the determined operation state is the preset error-prone fault type, inputting the characteristic vector into the support vector machine model so as to re-determine the fault type determined by the support vector machine model as the operation state of the on-load tap-changer. Therefore, the operating state of the on-load tap-changer can be detected more timely and accurately, the recognition capability of fault vibration signals is improved, and the probability of misjudgment is reduced.
Description
Technical Field
The present invention relates to the field of on-load tap-changer technologies, and in particular, to a method and a device for detecting an operating state of an on-load tap-changer.
Background
The on-load tap changer is very important power transmission and distribution equipment in a power system, wherein the on-load tap changer is a core component of the on-load tap changer. Through the step-by-step action of the on-load tap-changer, the on-load voltage regulating transformer can realize the on-load voltage regulation of the high-voltage transmission and distribution network, so that the system voltage of industrial power supply and residential power supply is kept stable. Therefore, the performance condition of the on-load tap-changer is directly related to the safe operation of the on-load tap-changer.
Currently, the main workload of the on-load tap-changer operation and maintenance is concentrated on the on-load tap-changer. The existing detection method of the on-load tap-changer mainly takes the running time and the operation times of the on-load tap-changer as the basis, and maintenance is carried out through regular preventive tests, regular inspection and fault overhaul. Therefore, the existing detection method is difficult to detect the fault problem of the on-load tap-changer in time; in addition, if the on-load tap-changer does not fail, the maintenance will cause unnecessary power outages.
Disclosure of Invention
In view of the above, an object of the present application is to provide a method and a device for detecting an operating state of an on-load tap-changer, wherein a vibration signal of the on-load tap-changer is utilized, and a gaussian mixture model and a bayes classifier are combined to determine the operating state of the on-load tap-changer; and when the determined operation state is the error-prone fault type, the fault type determined by the support vector machine model is determined as the operation state of the on-load tap-changer again. Therefore, the operation state of the on-load tap-changer can be detected more timely and accurately by combining a Gaussian mixture model, a Bayesian classification classifier and a support vector machine model, the recognition capability of fault vibration signals is improved, and the probability of misjudgment is reduced.
The embodiment of the application provides a method for detecting the running state of an on-load tap-changer, wherein the on-load tap-changer is included in an on-load tap-changer, and the method comprises the following steps:
extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates;
extracting a feature vector of the vibration signal by using a principal component analysis method;
inputting the feature vector into each pre-trained Gaussian mixture model to determine the probability distribution condition of the feature vector through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed;
determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the feature vector and each Gaussian mixture model;
and when the actual operating state of the on-load tap-changer is determined to be the fault-prone fault category preset in the fault categories, inputting the characteristic vector into a pre-trained support vector machine model so as to re-determine the fault category determined by the support vector machine model as the actual operating state of the on-load tap-changer.
Further, the Gaussian mixture model is trained by:
obtaining multi-class sample vibration signals of the on-load tap-changer; wherein the multi-class sample vibration signals comprise sample vibration signals when the on-load tap-changer is not in fault and sample vibration signals when the on-load tap-changer is in fault of the fault class for each fault class;
aiming at each type of sample vibration signals in the multi-type sample vibration signals, training an initial Gaussian mixture model by using a sample characteristic vector of the type of sample vibration signals and adopting an expectation-maximization algorithm to obtain a Gaussian mixture model corresponding to the type of sample vibration signals; and determining initial parameters of the initial Gaussian mixture model through a K-means clustering algorithm.
Further, the error-prone fault category is determined by:
dividing the multi-class sample vibration signals into a first group of sample vibration signals and a second group of sample vibration signals;
training an initial Bayes classifier by using the first group of sample vibration signals and each Gaussian mixture model to obtain a trained Bayes classifier;
aiming at any second vibration signal in the second group of sample vibration signals, testing a trained Bayesian classifier by using the second vibration signal and a Gaussian mixture model, and determining the predicted operation state of the on-load tap-changer when the second vibration signal is generated by the Bayesian classifier;
comparing the preset sample label of the second vibration signal with the predicted operation state determined by the Bayesian classifier, and judging whether the operation state is detected correctly or incorrectly by the Bayesian classifier; the preset sample label of the second vibration signal is the real running state of the on-load tap-changer when the second vibration signal is generated;
respectively carrying out data statistics on the operation states of the Bayesian classifier with detection errors and detection correctness to determine a group of fault categories which are classified and confused with each other when the Bayesian classifier detects the operation states;
determining each fault category in the set of fault categories as the error-prone fault category.
Further, training the support vector machine model by:
screening a plurality of second vibration signals from the second group of sample vibration signals in a first screening mode based on a preset sample label of each second vibration signal in the second group of sample vibration signals; the first screening mode is to screen out all second vibration signals of which preset sample labels belong to the error-prone fault category;
screening a plurality of second vibration signals from the plurality of second vibration signals screened by the first screening way by a second screening way; the second screening mode is that the Bayesian classifier detects all second vibration signals with correct operation state in the process of testing the Bayesian classifier;
dividing the plurality of second vibration signals screened in the second screening mode into a plurality of types of target vibration signals according to preset sample labels; each second vibration signal in each type of target vibration signal has the same preset sample label;
and training an initial support vector machine model by using the multi-class target vibration signals respectively to obtain the support vector machine model.
Further, the determining the actual operating state of the on-load tap-changer by using a pre-established bayesian classifier based on the probability distribution determined by the feature vector to be detected and each gaussian mixture model includes:
determining the probability of the on-load tap-changer in each running state based on the feature vector to be detected and the probability distribution condition determined by each Gaussian mixture model by using the Bayesian classifier;
and comparing the probability of the on-load tap-changer in each running state, and determining the running state with the highest probability as the actual running state of the on-load tap-changer.
Further, the extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates includes:
performing dimensionality reduction processing on the mixed vibration signal by using a principal component analysis method;
dividing the mixed vibration signal subjected to the dimensionality reduction into a plurality of independent sub-component signals by using a Robust independent component analysis method;
reconstructing a periodic signal in the mixed vibration signal according to the plurality of independent sub-component signals by using an orthogonal matching pursuit algorithm based on wavelet transformation; the periodic signal comprises a periodic signal which is generated by vibration of a winding, an iron core, an oil pump and a fan in the on-load tap changing transformer and is related to the frequency of a power grid when the on-load tap changing transformer runs;
and removing the periodic signal from the mixed vibration signal to obtain a vibration signal of the on-load tap-changer.
Further, the fault detection method further includes:
when the actual operation state of the on-load tap-changer is determined to be not the fault-prone fault category preset in the fault categories, determining the fault category determined by the Bayesian classifier as the actual operation state of the on-load tap-changer; wherein the fault categories include at least one or more of: contact looseness, contact wear, spindle deformation, and main spring weakening.
The embodiment of the present application further provides a detection device for an operating state of an on-load tap-changer, the on-load tap-changer includes in an on-load tap-changer, the detection device includes:
the first extraction module is used for extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates;
the second extraction module is used for extracting the characteristic vector of the vibration signal by using a principal component analysis method;
the input module is used for inputting the feature vectors into the Gaussian mixture model aiming at each pre-trained Gaussian mixture model so as to determine the probability distribution condition of the feature vectors through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed;
the first determination module is used for determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the characteristic vector and each Gaussian mixture model;
and the second determination module is used for inputting the characteristic vector into a pre-trained support vector machine model when the actual operation state of the on-load tap-changer is determined to be a fault-prone fault type preset in the fault types, so that the fault type determined by the support vector machine model is determined to be the actual operation state of the on-load tap-changer again.
Further, the detection device further comprises a first training module; the first training module trains the Gaussian mixture model by:
obtaining multi-class sample vibration signals of the on-load tap-changer; wherein the multi-class sample vibration signals comprise sample vibration signals when the on-load tap-changer is not in fault and sample vibration signals when the on-load tap-changer is in fault of the fault class for each fault class;
aiming at each type of sample vibration signals in the multi-type sample vibration signals, training an initial Gaussian mixture model by using a sample characteristic vector of the type of sample vibration signals and adopting an expectation-maximization algorithm to obtain a Gaussian mixture model corresponding to the type of sample vibration signals; and determining initial parameters of the initial Gaussian mixture model through a K-means clustering algorithm.
Further, the detection device further comprises a second training module; the second training module determines the error-prone fault category by:
dividing the multi-class sample vibration signals into a first group of sample vibration signals and a second group of sample vibration signals;
training an initial Bayes classifier by using the first group of sample vibration signals and each Gaussian mixture model to obtain a trained Bayes classifier;
aiming at any one second vibration signal in the second group of sample vibration signals, testing a trained Bayesian classifier by using the second vibration signal and a Gaussian mixture model, and determining the predicted operation state of the on-load tap-changer when the second vibration signal is generated by the Bayesian classifier;
comparing the preset sample label of the second vibration signal with the predicted operation state determined by the Bayesian classifier, and judging whether the operation state is detected correctly or incorrectly by the Bayesian classifier; the preset sample label of the second vibration signal is the real running state of the on-load tap-changer when the second vibration signal is generated;
respectively carrying out data statistics on the operation states of the Bayesian classifier with detection errors and detection correctness to determine a group of fault categories which are classified and confused with each other when the Bayesian classifier detects the operation states;
determining each fault category in the set of fault categories as the error-prone fault category.
Further, the detection device further comprises a third training module; the third training module trains the support vector machine model by:
screening a plurality of second vibration signals from the second group of sample vibration signals in a first screening mode based on a preset sample label of each second vibration signal in the second group of sample vibration signals; the first screening mode is to screen out all second vibration signals of which preset sample labels belong to the error-prone fault category;
screening a plurality of second vibration signals from the plurality of second vibration signals screened by the first screening mode by a second screening mode; the second screening mode is that the Bayesian classifier detects all second vibration signals with correct operation state in the process of testing the Bayesian classifier;
dividing the plurality of second vibration signals screened in the second screening mode into a plurality of types of target vibration signals according to preset sample labels; each second vibration signal in each type of target vibration signal has the same preset sample label;
and training an initial support vector machine model by using the multi-class target vibration signals respectively to obtain the support vector machine model.
Further, when the first determining module is configured to determine the actual operating state of the on-load tap-changer by using a pre-trained bayesian classifier based on the feature vectors and the probability distribution determined by each gaussian mixture model, the first determining module is configured to:
determining the probability of the on-load tap-changer in each running state based on the feature vector to be detected and the probability distribution condition determined by each Gaussian mixture model by using the Bayesian classifier;
and comparing the probability of the on-load tap-changer in each running state, and determining the running state with the highest probability as the actual running state of the on-load tap-changer.
Further, when the first extraction module is configured to extract the vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates, the first extraction module is configured to:
performing dimensionality reduction processing on the mixed vibration signal by using a principal component analysis method;
dividing the mixed vibration signal subjected to the dimensionality reduction into a plurality of independent sub-component signals by using a Robust independent component analysis method;
reconstructing a periodic signal in the mixed vibration signal according to the plurality of independent sub-component signals by using an orthogonal matching pursuit algorithm based on wavelet transformation; the periodic signal comprises a periodic signal which is generated by vibration of a winding, an iron core, an oil pump and a fan in the on-load tap changing transformer and is related to the frequency of a power grid when the on-load tap changing transformer runs;
and removing the periodic signal from the mixed vibration signal to obtain a vibration signal of the on-load tap-changer.
Further, the detection device further comprises a third determination module; the third determining module is to:
when the actual operating state of the on-load tap-changer is determined to be not the fault-prone fault category preset in the fault categories, determining the fault category determined by the Bayesian classifier as the actual operating state of the on-load tap-changer; wherein the fault categories include at least one or more of: contact looseness, contact wear, spindle deformation, and main spring weakening.
An embodiment of the present application further provides an electronic device, including: a processor, a memory and a bus, the memory storing machine readable instructions executable by the processor, the processor and the memory communicating via the bus when the electronic device is operating, the machine readable instructions when executed by the processor performing the steps of a method of detecting an operating state of an on-load tap-changer as described above.
Embodiments of the present application further provide a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, performs the steps of the method for detecting an operating state of an on-load tap-changer.
According to the detection method and the detection device for the operating state of the on-load tap-changer, provided by the embodiment of the application, the vibration signal of the on-load tap-changer is extracted from the mixed vibration signal generated when the on-load tap-changer operates; extracting a feature vector of the vibration signal by using a principal component analysis method; inputting the feature vector into each pre-trained Gaussian mixture model to determine the probability distribution condition of the feature vector through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed; determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the feature vector and each Gaussian mixture model; and when the actual operating state of the on-load tap-changer is determined to be the fault-prone fault category preset in the fault categories, inputting the characteristic vector into a pre-trained support vector machine model so as to re-determine the fault category determined by the support vector machine model as the actual operating state of the on-load tap-changer.
By the mode, the operating state of the on-load tap-changer can be determined by using the vibration signal of the on-load tap-changer and combining a Gaussian mixture model and a Bayesian classifier; and when the determined operation state is the error-prone fault type, the fault type determined by the support vector machine model is determined as the operation state of the on-load tap-changer again. Therefore, the operation state of the on-load tap-changer can be detected more timely and accurately by combining a Gaussian mixture model, a Bayesian classification classifier and a support vector machine model, the recognition capability of fault vibration signals is improved, and the probability of misjudgment is reduced.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
Fig. 1 shows a flow chart of a method for detecting an operating state of an on-load tap changer according to an embodiment of the present application;
fig. 2 shows one of the schematic structural diagrams of the detection device for the operating state of the on-load tap-changer provided by the embodiment of the present application;
fig. 3 shows a second schematic structural diagram of a device for detecting an operating state of an on-load tap changer according to an embodiment of the present application;
fig. 4 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. Every other embodiment that can be obtained by a person skilled in the art without making creative efforts based on the embodiments of the present application falls within the protection scope of the present application.
First, an application scenario to which the present application is applicable will be described. The application can be applied to the technical field of on-load tap-changers.
Research shows that at present, the main workload of the on-load tap-changer operation and maintenance is concentrated on the on-load tap-changer. The existing detection method of the on-load tap-changer mainly takes the running time and the operation times of the on-load tap-changer as the basis, and maintenance is carried out through regular preventive tests, regular checks and fault overhaul. Therefore, the existing detection method is difficult to detect the fault problem of the on-load tap-changer in time; in addition, if the on-load tap-changer does not fail, the maintenance will cause unnecessary power outages.
Based on this, the embodiment of the application provides a method and a device for detecting the operating state of an on-load tap-changer, so as to detect the operating state of the on-load tap-changer more timely and accurately, improve the recognition capability of a fault vibration signal, and reduce the probability of misjudgment.
Referring to fig. 1, fig. 1 is a flowchart illustrating a method for detecting an operating state of an on-load tap changer according to an embodiment of the present disclosure. The on-load tap-changer is included in an on-load tap-changer, as shown in fig. 1, a detection method provided by the embodiment of the present application includes:
s101, extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates.
The mixed vibration signal generated when the on-load tap changer operates can be collected through a sensor arranged on a transformer oil tank.
It should be noted that the vibration signal acquired by the sensor includes the vibration signal of the winding, the iron core, the oil pump, and the fan, in addition to the vibration signal of the on-load tap-changer itself. Because the vibration signals of the winding, the iron core, the oil pump and the fan are periodic signals related to the frequency of a power grid, and the action of the on-load tap-changer generates non-periodic impact signals, the two types of signals are independent from each other, and the independent vibration signals during the action of the on-load tap-changer can be extracted.
In one possible implementation, step S101 includes:
and S1011, performing dimensionality reduction on the mixed vibration signal by using a principal component analysis method.
It should be noted that, in order to ensure the integrity and accuracy of the sampled data, a higher sampling frequency may be used to sample the mixed vibration signal, which may result in a larger amount of sampled data. In order to avoid the problem of large calculation amount caused by large data volume and ensure the integrity of the acquired data, the acquired vibration signals can be subjected to dimensionality reduction by using a principal component analysis method according to the prior knowledge of the actual condition. Before the dimension reduction processing is carried out on the mixed vibration signal, the mixed vibration signal can be converted into a high-dimensional subspace signal through zero filling.
And S1012, dividing the mixed vibration signal subjected to the dimensionality reduction into a plurality of independent sub-component signals by using a Robust independent component analysis method.
Here, the number of iterations can be reduced by adopting a Robust independent component analysis method, the convergence speed is increased, and the calculation amount is greatly reduced; the Robust independent component analysis method can also improve the robustness when the signal has dead points and pseudo local extrema; under a small sample space, the mean square error of the Robust independent component analysis method is obviously superior to that of the common independent component analysis method.
And S1013, reconstructing a periodic signal in the mixed vibration signal according to the plurality of independent sub-component signals by using an orthogonal matching pursuit algorithm based on wavelet transformation.
And the periodic signal comprises a periodic signal which is generated by vibration of a winding, an iron core, an oil pump and a fan in the on-load tap changing transformer and is related to the frequency of a power grid when the on-load tap changing transformer operates. Specifically, the normalized kurtosis value of each independent subcomponent signal can be calculated, and the independent subcomponent signal corresponding to the smallest normalized kurtosis value is screened out; and reconstructing the periodic signal by using an orthogonal matching pursuit algorithm based on wavelet transformation.
And S1014, removing the periodic signal from the mixed vibration signal to obtain a vibration signal of the on-load tap-changer.
And S102, extracting the feature vector of the vibration signal by using a principal component analysis method.
In the step, the vibration signal of the on-load tap-changer is subjected to feature extraction through a principal component analysis method to obtain a W-dimensional feature vector.
S103, inputting the feature vectors into the Gaussian mixture model aiming at each pre-trained Gaussian mixture model, and determining the probability distribution condition of the feature vectors through the Gaussian mixture model.
The Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating condition includes that the on-load tap-changer is not faulty and a fault category of the on-load tap-changer. Illustratively, the fault category may include at least one or more of: contact looseness, contact wear, spindle deformation, and main spring weakening.
In one possible implementation, the gaussian mixture model may be trained by:
and step 1, obtaining multi-class sample vibration signals of the on-load tap-changer.
The types of the sample vibration signals correspond to the operating states of the on-load tap-changer one by one, that is, the sample vibration signals respectively include the sample vibration signals when the on-load tap-changer is not in fault and the sample vibration signals when the on-load tap-changer is in fault of the fault type for each fault type. Corresponding to the above example, five types of sample vibration signals of the on-load tap-changer in five operating states of no fault, loose contact, wear of the contact, deformation of the main shaft and weakening of the main spring can be obtained respectively.
And 2, aiming at each type of sample vibration signals in the multi-type sample vibration signals, training an initial Gaussian mixture model by using the sample characteristic vector of the type of sample vibration signals and adopting an expectation-maximization algorithm to obtain the Gaussian mixture model corresponding to the type of sample vibration signals.
In the step, firstly, a sample feature vector of each sample vibration signal in any type of sample vibration signals is extracted
(ii) a Secondly, determining initialization parameters of the Gaussian mixture model by using a K-means clustering algorithm
To construct an initial Gaussian mixture model; thirdly, obtaining new model parameters of the Gaussian mixture model by using the sample characteristic vector of the sample vibration signal and an expectation maximization algorithm
The constraint conditions are as follows:
wherein, in the step (A),
likelihood functions that are gaussian mixture models; the expectation-maximization algorithm can correct the model parameters calculated in the last iteration through multiple iterative calculations to obtain new model parameters until the Gaussian mixture model meets the preset convergence condition. At this time, the training of the gaussian mixture model corresponding to the type of sample vibration signal (i.e., the operating state of the on-load tap-changer at the time of generation of the type of sample vibration signal) is completed.
By adopting the method, the Gaussian mixture model corresponding to the sample vibration signals is trained respectively according to the sample vibration signals. In this way, the trained gaussian mixture models correspond to the operating states one by one, and corresponding to the above example, five gaussian mixture models for describing the probability distribution of the feature vector of the vibration signal can be trained respectively when the on-load tap-changer is in the five operating states of no fault, loose contact, abrasion of the contact, deformation of the main shaft and weakening of the main spring.
And S104, determining the actual running state of the on-load tap-changer by using a pre-trained Bayes classifier based on the probability distribution condition determined by the feature vector and each Gaussian mixture model.
In one possible implementation, step S104 includes:
and S1041, determining the probability of the on-load tap-changer in each running state by using the Bayesian classifier based on the probability distribution condition determined by the feature vector to be detected and each Gaussian mixture model.
S1042, comparing the probability of the on-load tap-changer in each operation state, and determining the operation state with the maximum probability as the actual operation state of the on-load tap-changer.
Here, the pre-trained gaussian mixture model can provide prior probability required by decision of the bayesian classifier, and the bayesian classifier determines the probability of the on-load tap-changer in each operating state according to the feature vector to be detected and the probability distribution condition determined by each gaussian mixture model; and determining the operation state with the highest probability as the actual operation state of the on-load tap-changer by comparing the probability of the on-load tap-changer in each operation state. The Gaussian mixture model not only serves as a reference model for identification and classification, but also can perform clustering on the feature vectors to be detected, namely the feature vectors are more easily classified after being subjected to Gaussian mixture clustering, so that the identification capability of the vibration signals is improved.
And S105, when the actual operation state of the on-load tap-changer is determined to be the fault-prone fault type preset in the fault types, inputting the characteristic vector into a pre-trained support vector machine model, and re-determining the fault type determined by the support vector machine model as the actual operation state of the on-load tap-changer.
The preset fault-prone classification is determined according to the classification condition of the trained Bayes classifier in the process of training and testing the Bayes classifier. In particular, in one possible implementation, the error-prone fault category may be determined by:
step 1, dividing the multi-class sample vibration signals into a first group of sample vibration signals and a second group of sample vibration signals.
In this step, the multi-class sample vibration signals may be divided into a first group of sample vibration signals and a second group of sample vibration signals according to a certain quantitative ratio. Wherein, each group of sample vibration signals comprises vibration signals generated when the on-load tap-changer is in each operation state.
And 2, training the initial Bayes classifier by using the first group of sample vibration signals and each Gaussian mixture model to obtain the trained Bayes classifier.
And 3, aiming at any second vibration signal in the second group of sample vibration signals, testing the trained Bayesian classifier by using the second vibration signal and the Gaussian mixture model, and determining the predicted operation state of the on-load tap-changer when the second vibration signal is generated by the Bayesian classifier.
And 4, comparing the preset sample label of the second vibration signal with the predicted operation state determined by the Bayes classifier, and judging whether the operation state is detected correctly or incorrectly by the Bayes classifier.
The preset sample label of the second vibration signal is the real running state of the on-load tap-changer when the second vibration signal is generated.
And 5, respectively carrying out data statistics on the operation states of the Bayesian classifier with the detection errors and the detection correctness, and determining a group of fault categories with the confused classifications when the Bayesian classifier detects the operation states.
In this step, the detection accuracy of the bayesian classifier for each sample label can be respectively counted. Illustratively, as shown in table 1 below, table 1 is a statistical table of the detection results of the bayesian classifier in one experiment.
TABLE 1 Bayes classifier test result statistical table
Illustratively, data statistics are respectively carried out on the operation states of the Bayesian classifier with detection errors and detection correctness, and a detection result statistical table is obtained as shown in the table; taking a sample vibration signal with a sample label of "contact wear" in the second group of sample signals as an example (that is, when the sample vibration signal is generated, the real operating state of the on-load tap-changer is "contact wear"), through the detection of the bayesian classifier, in 50 sample vibration signals with a sample label of "contact wear", 40 sample vibration signals are correctly detected, and 10 sample vibration signals are erroneously detected as "contact looseness", so that the classification accuracy of the bayesian classifier for the "contact wear" type vibration signals is 80%; likewise, of the 50 sample vibration signals labeled "loose tip", 41 sample vibration signals were detected correctly, while 9 sample vibration signals were erroneously detected as "worn tip". Therefore, the data statistics result shows that the contact looseness and the contact abrasion are a group of fault categories which are easy to be classified and mixed when the Bayesian classifier detects the operation state.
And 6, determining each fault category in the group of fault categories as the error-prone fault category.
Corresponding to the above example, both "contact looseness" and "contact wear" may be determined as error prone fault categories.
It is noted that the error-prone fault category refers to a group of fault categories which are confused with each other in classification when the Bayesian classifier detects the operation state; that is, the error-prone fault category is not only a fault category which is easy to classify by the bayesian classifier screened according to the classification accuracy, but also a fault category which belongs to the error-prone fault category, and a vibration signal of any fault category is easy to be classified by the bayesian classifier to other fault categories in the group; corresponding to the above example, the vibration signal of "contact loose" is easily classified as "contact wear" by the bayesian classifier, and the vibration signal of "contact wear" is easily classified as "contact loose" by the bayesian classifier, which means that "contact loose" and "contact wear" constitute a group of error-prone fault categories.
This is because certain classes of vibration signals are more similar in their characteristics, for example, vibration signals with loose contacts and worn contacts are very similar in their characteristics, both spike amplitude attenuation signals. The Bayesian classifier cannot effectively distinguish the vibration signals of the categories due to the self limitation, so that the classification confusion is easy to occur, and the misjudgment phenomenon is caused.
Therefore, in order to improve the accuracy of fault diagnosis and reduce the error rate, when the actual operating state of the on-load tap-changer is determined to be the preset error-prone fault type by using the bayesian classifier, the feature vector can be input into the pre-trained support vector machine model again, so that the fault type determined by the support vector machine model can be determined to be the actual operating state of the on-load tap-changer again.
Therefore, the operating state of the on-load tap-changer can be detected more accurately by utilizing the characteristics that the support vector machine model is good in linear inseparable condition and the number of required samples is small. Through the screening of the Bayes classifier, the vibration signal which is easy to misjudge is combined with the SVM model, the misjudge problem when the characteristics are similar in the Bayes algorithm is solved, the storage occupation resources of the processor are reduced, the recognition capability of the fault signal is improved, and the misjudge probability is reduced.
In one possible implementation, the support vector machine model may be trained by:
step 1, screening a plurality of second vibration signals from the second group of sample vibration signals in a first screening mode based on a preset sample label of each second vibration signal in the second group of sample vibration signals.
And the first screening mode is used for screening all second vibration signals of which preset sample labels belong to the error-prone fault category.
And 2, screening a plurality of second vibration signals from the plurality of second vibration signals screened by the first screening mode by a second screening mode.
And the second screening mode is that the Bayesian classifier detects all second vibration signals with correct operation state in the process of testing the Bayesian classifier.
Therefore, interference samples in the sample vibration signals belonging to the error-prone fault category can be removed through twice screening, and the interference to the model is reduced, so that the support vector machine model trained by the sample vibration signals obtained after twice screening can be used for better detecting the error-prone fault category. Experiments prove that the misjudgment times of the support vector machine model trained in the way on the vibration signals of the fault-prone types can be obviously reduced.
And 3, dividing the plurality of second vibration signals screened in the second screening mode into a plurality of types of target vibration signals according to preset sample labels.
And the second vibration signals in each type of target vibration signals have the same preset sample label. In the step, the plurality of second vibration signals screened by the second screening mode are classified according to different preset sample labels. Corresponding to the above example, the plurality of second vibration signals screened out in the second screening manner may be divided into two types of target vibration signals, where the preset sample label of each vibration signal in one type of target vibration signal is "loose contact", and the preset sample label of each vibration signal in another type of target vibration signal is "worn contact".
And 4, training an initial support vector machine model by using the multi-class target vibration signals respectively to obtain the support vector machine model.
For example, the initial support vector machine model may be trained by using a sample labeled "loose tip" and a bayesian classifier detecting a correct sample vibration signal as a training sample of the loose tip class in the support vector machine model.
Further, the detection method further comprises: and S106, when the actual operation state of the on-load tap-changer is determined not to be the fault-prone type preset in the fault types, determining the fault type determined by the Bayesian classifier as the actual operation state of the on-load tap-changer.
The method for detecting the operating state of the on-load tap-changer provided by the embodiment of the application comprises the following steps: extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates; extracting a feature vector of the vibration signal by using a principal component analysis method; inputting the feature vector into each pre-trained Gaussian mixture model to determine the probability distribution condition of the feature vector through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed; determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the feature vector and each Gaussian mixture model; and when the actual operating state of the on-load tap-changer is determined to be the fault-prone fault category preset in the fault categories, inputting the characteristic vector into a pre-trained support vector machine model so as to re-determine the fault category determined by the support vector machine model as the actual operating state of the on-load tap-changer.
Therefore, the operation state of the on-load tap-changer can be detected more timely and accurately by combining a Gaussian mixture model, a Bayesian classification classifier and a support vector machine model, the recognition capability of fault vibration signals is improved, and the probability of misjudgment is reduced.
Referring to fig. 2 and fig. 3, fig. 2 is a first schematic structural diagram of an on-load tap-changer operation state detection device according to an embodiment of the present disclosure, and fig. 3 is a second schematic structural diagram of an on-load tap-changer operation state detection device according to an embodiment of the present disclosure. As shown in fig. 2, the detection apparatus 200 includes:
a first extraction module 210, configured to extract a vibration signal of the on-load tap changer from a mixed vibration signal generated when the on-load tap changer operates;
a second extraction module 220, configured to extract a feature vector of the vibration signal by using a principal component analysis method;
an input module 230, configured to input the feature vector into a pre-trained gaussian mixture model for each gaussian mixture model, so as to determine a probability distribution of the feature vector through the gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed;
a first determining module 240, configured to determine, based on the feature vector and the probability distribution determined by each gaussian mixture model, an actual operating state of the on-load tap-changer by using a pre-trained bayesian classifier;
and a second determining module 250, configured to, when it is determined that the actual operating state of the on-load tap-changer is the fault-prone fault category preset in the fault categories, input the feature vector into a pre-trained support vector machine model, so as to re-determine the fault category determined by the support vector machine model as the actual operating state of the on-load tap-changer.
Further, as shown in fig. 3, the detection apparatus 200 further includes a first training module 260; the first training module 260 trains the gaussian mixture model by:
acquiring multi-type sample vibration signals of the on-load tap-changer; wherein the multi-class sample vibration signals comprise sample vibration signals when the on-load tap-changer is not in fault and sample vibration signals when the on-load tap-changer is in fault of the fault class for each fault class;
aiming at each type of sample vibration signals in the multi-type sample vibration signals, training an initial Gaussian mixture model by using a sample characteristic vector of the type of sample vibration signals and adopting an expectation-maximization algorithm to obtain a Gaussian mixture model corresponding to the type of sample vibration signals; and determining initial parameters of the initial Gaussian mixture model through a K-means clustering algorithm.
Further, as shown in fig. 3, the detection apparatus 200 further includes a second training module 270; the second training module 270 determines the error-prone fault category by:
dividing the multi-class sample vibration signals into a first group of sample vibration signals and a second group of sample vibration signals;
training an initial Bayes classifier by using the first group of sample vibration signals and each Gaussian mixture model to obtain a trained Bayes classifier;
aiming at any one second vibration signal in the second group of sample vibration signals, testing a trained Bayesian classifier by using the second vibration signal and a Gaussian mixture model, and determining the predicted operation state of the on-load tap-changer when the second vibration signal is generated by the Bayesian classifier;
comparing the preset sample label of the second vibration signal with the predicted operation state determined by the Bayesian classifier, and judging whether the operation state is detected correctly or incorrectly by the Bayesian classifier; the preset sample label of the second vibration signal is the real running state of the on-load tap-changer when the second vibration signal is generated;
respectively carrying out data statistics on the operation states of the Bayesian classifier with detection errors and detection correctness to determine a group of fault categories which are classified and confused with each other when the Bayesian classifier detects the operation states;
determining each fault category in the set of fault categories as the error-prone fault category.
Further, as shown in fig. 3, the detection apparatus 200 further includes a third training module 280; the third training module 280 trains the support vector machine model by:
screening a plurality of second vibration signals from the second group of sample vibration signals in a first screening mode based on a preset sample label of each second vibration signal in the second group of sample vibration signals; the first screening mode is to screen out all second vibration signals of which preset sample labels belong to the error-prone fault category;
screening a plurality of second vibration signals from the plurality of second vibration signals screened by the first screening mode by a second screening mode; the second screening mode is that the Bayesian classifier detects all second vibration signals with correct operation state in the process of testing the Bayesian classifier;
dividing the plurality of second vibration signals screened in the second screening mode into a plurality of types of target vibration signals according to preset sample labels; each second vibration signal in each type of target vibration signal has the same preset sample label;
and training an initial support vector machine model by using the multi-class target vibration signals respectively to obtain the support vector machine model.
Further, when the first determining module 240 is configured to determine the actual operating state of the on-load tap changer by using a pre-trained bayesian classifier based on the feature vectors and the probability distribution determined by each gaussian mixture model, the first determining module 240 is configured to:
determining the probability of the on-load tap-changer in each running state based on the feature vector to be detected and the probability distribution condition determined by each Gaussian mixture model by using the Bayesian classifier;
and comparing the probability of the on-load tap-changer in each running state, and determining the running state with the highest probability as the actual running state of the on-load tap-changer.
Further, when the first extraction module 210 is configured to extract the vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates, the first extraction module 210 is configured to:
performing dimensionality reduction processing on the mixed vibration signal by using a principal component analysis method;
dividing the mixed vibration signal subjected to the dimensionality reduction into a plurality of independent sub-component signals by using a Robust independent component analysis method;
reconstructing a periodic signal in the mixed vibration signal according to the plurality of independent sub-component signals by using an orthogonal matching pursuit algorithm based on wavelet transformation; the periodic signal comprises a periodic signal which is generated by vibration of a winding, an iron core, an oil pump and a fan in the on-load tap changing transformer and is related to the frequency of a power grid when the on-load tap changing transformer operates;
and removing the periodic signal from the mixed vibration signal to obtain a vibration signal of the on-load tap-changer.
Further, as shown in fig. 3, the detecting apparatus 200 further includes a third determining module 290; the third determining module 290 is configured to:
when the actual operation state of the on-load tap-changer is determined to be not the fault-prone fault category preset in the fault categories, determining the fault category determined by the Bayesian classifier as the actual operation state of the on-load tap-changer; wherein the fault categories include at least one or more of: contact looseness, contact wear, spindle deformation, and main spring weakening.
Referring to fig. 4, fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. As shown in fig. 4, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.
The memory 420 stores machine-readable instructions executable by the processor 410, when the electronic device 400 runs, the processor 410 and the memory 420 communicate through the bus 430, and when the machine-readable instructions are executed by the processor 410, the steps of the method for detecting the operating state of the on-load tap-changer in the embodiment of the method shown in fig. 1 may be executed.
An embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the step of the method for detecting an operating state of an on-load tap-changer in the method embodiment shown in fig. 1 may be executed.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present application and are intended to be covered by the appended claims. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method of detecting an operating condition of an on-load tap changer, the on-load tap changer being included in an on-load tap changer, the method comprising:
extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates;
extracting a feature vector of the vibration signal by using a principal component analysis method;
inputting the feature vector into each pre-trained Gaussian mixture model to determine the probability distribution condition of the feature vector through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed;
determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the feature vector and each Gaussian mixture model;
and when the actual operating state of the on-load tap-changer is determined to be the fault-prone fault category preset in the fault categories, inputting the characteristic vector into a pre-trained support vector machine model so as to re-determine the fault category determined by the support vector machine model as the actual operating state of the on-load tap-changer.
2. The detection method according to claim 1, wherein the gaussian mixture model is trained by:
obtaining multi-class sample vibration signals of the on-load tap-changer; wherein the multi-class sample vibration signals comprise sample vibration signals when the on-load tap-changer is not in fault and sample vibration signals when the on-load tap-changer is in fault of the fault class for each fault class;
aiming at each type of sample vibration signals in the multi-type sample vibration signals, training an initial Gaussian mixture model by using a sample characteristic vector of the type of sample vibration signals and adopting an expectation-maximization algorithm to obtain a Gaussian mixture model corresponding to the type of sample vibration signals; and determining initial parameters of the initial Gaussian mixture model through a K-means clustering algorithm.
3. The detection method according to claim 2, wherein the error-prone fault category is determined by:
dividing the multi-class sample vibration signals into a first group of sample vibration signals and a second group of sample vibration signals;
training an initial Bayes classifier by using the first group of sample vibration signals and each Gaussian mixture model to obtain a trained Bayes classifier;
aiming at any one second vibration signal in the second group of sample vibration signals, testing a trained Bayesian classifier by using the second vibration signal and a Gaussian mixture model, and determining the predicted operation state of the on-load tap-changer when the second vibration signal is generated by the Bayesian classifier;
comparing the preset sample label of the second vibration signal with the predicted operation state determined by the Bayesian classifier, and judging whether the operation state is detected correctly or incorrectly by the Bayesian classifier; the preset sample label of the second vibration signal is the real running state of the on-load tap-changer when the second vibration signal is generated;
respectively carrying out data statistics on the operation states of the Bayesian classifier with detection errors and detection correctness to determine a group of fault categories which are classified and confused with each other when the Bayesian classifier detects the operation states;
determining each fault category in the set of fault categories as the error-prone fault category.
4. The detection method according to claim 3, wherein the support vector machine model is trained by:
screening a plurality of second vibration signals from the second group of sample vibration signals in a first screening mode based on a preset sample label of each second vibration signal in the second group of sample vibration signals; the first screening mode is to screen out all second vibration signals of which preset sample labels belong to the error-prone fault category;
screening a plurality of second vibration signals from the plurality of second vibration signals screened by the first screening mode by a second screening mode; the second screening mode is that the Bayesian classifier detects all second vibration signals with correct operation state in the process of testing the Bayesian classifier;
dividing the plurality of second vibration signals screened out in the second screening mode into a plurality of types of target vibration signals according to preset sample labels; each second vibration signal in each type of target vibration signal has the same preset sample label;
and training an initial support vector machine model by using the multi-class target vibration signals respectively to obtain the support vector machine model.
5. The detection method according to claim 1, wherein the determining the actual operating state of the on-load tap-changer by using a pre-constructed bayesian classifier based on the probability distribution determined by the feature vector to be detected and each gaussian mixture model comprises:
determining the probability of the on-load tap-changer in each running state based on the feature vector to be detected and the probability distribution condition determined by each Gaussian mixture model by using the Bayesian classifier;
and comparing the probability of the on-load tap-changer in each running state, and determining the running state with the highest probability as the actual running state of the on-load tap-changer.
6. The method for detecting according to claim 1, wherein the extracting the vibration signal of the on-load tap-changer from the mixed vibration signal generated by the on-load tap-changer during operation comprises:
performing dimensionality reduction processing on the mixed vibration signal by using a principal component analysis method;
dividing the mixed vibration signal subjected to the dimensionality reduction into a plurality of independent sub-component signals by using a Robust independent component analysis method;
reconstructing a periodic signal in the mixed vibration signal according to the plurality of independent sub-component signals by using an orthogonal matching pursuit algorithm based on wavelet transformation; the periodic signal comprises a periodic signal which is generated by vibration of a winding, an iron core, an oil pump and a fan in the on-load tap changing transformer and is related to the frequency of a power grid when the on-load tap changing transformer runs;
and removing the periodic signal from the mixed vibration signal to obtain a vibration signal of the on-load tap-changer.
7. The detection method according to claim 1, wherein the fault detection method further comprises:
when the actual operation state of the on-load tap-changer is determined to be not the fault-prone fault category preset in the fault categories, determining the fault category determined by the Bayesian classifier as the actual operation state of the on-load tap-changer; wherein the fault categories include at least one or more of: contact looseness, contact wear, spindle deformation, and main spring weakening.
8. An on-load tap changer operating condition detection device, wherein the on-load tap changer is comprised in an on-load tap changer, the detection device comprising:
the first extraction module is used for extracting a vibration signal of the on-load tap-changer from a mixed vibration signal generated when the on-load tap-changer operates;
the second extraction module is used for extracting the characteristic vector of the vibration signal by using a principal component analysis method;
the input module is used for inputting the feature vectors into the Gaussian mixture model aiming at each pre-trained Gaussian mixture model so as to determine the probability distribution condition of the feature vectors through the Gaussian mixture model; the Gaussian mixed model is used for describing the probability distribution condition of the characteristic vector if the on-load tap-changer is in the running state corresponding to the Gaussian mixed model; the operating state comprises that the on-load tap-changer has not failed and a fault category of the on-load tap-changer having failed;
the first determination module is used for determining the actual running state of the on-load tap-changer by utilizing a pre-trained Bayes classifier based on the probability distribution condition determined by the characteristic vector and each Gaussian mixture model;
and the second determination module is used for inputting the characteristic vector into a pre-trained support vector machine model when the actual operation state of the on-load tap-changer is determined to be a fault-prone fault type preset in the fault types, so that the fault type determined by the support vector machine model is determined to be the actual operation state of the on-load tap-changer again.
9. An electronic device, comprising: a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory communicating over the bus when the electronic device is operating, the machine-readable instructions when executed by the processor performing the steps of a method of detecting an operating condition of an on-load tap-changer according to any of claims 1 to 7.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of a method for detecting an operating state of an on-load tap changer according to one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210352722.8A CN114492542B (en) | 2022-04-06 | 2022-04-06 | Method and device for detecting running state of on-load tap-changer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210352722.8A CN114492542B (en) | 2022-04-06 | 2022-04-06 | Method and device for detecting running state of on-load tap-changer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114492542A true CN114492542A (en) | 2022-05-13 |
| CN114492542B CN114492542B (en) | 2022-07-12 |
Family
ID=81487468
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210352722.8A Active CN114492542B (en) | 2022-04-06 | 2022-04-06 | Method and device for detecting running state of on-load tap-changer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114492542B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007294543A (en) * | 2006-04-21 | 2007-11-08 | Toshiba Corp | Failure diagnostic device for on-load tap changer and its method |
| CN101364408A (en) * | 2008-10-07 | 2009-02-11 | 西安成峰科技有限公司 | Sound image combined monitoring method and system |
| CN107273585A (en) * | 2017-05-25 | 2017-10-20 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch fault detection method and device |
| CN107328560A (en) * | 2017-05-26 | 2017-11-07 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch diagnostic method and device |
| CN112014047A (en) * | 2020-08-27 | 2020-12-01 | 华侨大学 | Mechanical fault diagnosis method for on-load tap-changer |
| CN113553930A (en) * | 2021-07-14 | 2021-10-26 | 董跃周 | Method for diagnosing mechanical fault of on-load tap-changer of transformer |
-
2022
- 2022-04-06 CN CN202210352722.8A patent/CN114492542B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007294543A (en) * | 2006-04-21 | 2007-11-08 | Toshiba Corp | Failure diagnostic device for on-load tap changer and its method |
| CN101364408A (en) * | 2008-10-07 | 2009-02-11 | 西安成峰科技有限公司 | Sound image combined monitoring method and system |
| CN107273585A (en) * | 2017-05-25 | 2017-10-20 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch fault detection method and device |
| CN107328560A (en) * | 2017-05-26 | 2017-11-07 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch diagnostic method and device |
| CN112014047A (en) * | 2020-08-27 | 2020-12-01 | 华侨大学 | Mechanical fault diagnosis method for on-load tap-changer |
| CN113553930A (en) * | 2021-07-14 | 2021-10-26 | 董跃周 | Method for diagnosing mechanical fault of on-load tap-changer of transformer |
Non-Patent Citations (2)
| Title |
|---|
| 张琛亮: "基于混合高斯模型的变压器振动状态诊断方法", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 胡东等: "基于有效频带特征的有载分接开关机械状态诊断方法", 《电气应用》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114492542B (en) | 2022-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3680639B1 (en) | Abnormality model learning device, method, and program | |
| US10805151B2 (en) | Method, apparatus, and storage medium for diagnosing failure based on a service monitoring indicator of a server by clustering servers with similar degrees of abnormal fluctuation | |
| US20110296244A1 (en) | Log message anomaly detection | |
| CN110826648A (en) | Method for realizing fault detection by utilizing time sequence clustering algorithm | |
| CN110991657A (en) | Abnormal sample detection method based on machine learning | |
| CN108171064B (en) | Sample format protection method and device for gray box fuzz test | |
| CN111291783A (en) | Intelligent fault diagnosis method, system, terminal and storage medium for gas pressure regulating equipment | |
| CN111966569A (en) | Hard disk health degree evaluation method and device and computer readable storage medium | |
| Prasojo et al. | Precise transformer fault diagnosis via random forest model enhanced by synthetic minority over-sampling technique | |
| CN111124732A (en) | Disk fault prediction method, system, device and storage medium | |
| Choeikiwong et al. | Software defect prediction in imbalanced data sets using unbiased support vector machine | |
| US11132790B2 (en) | Wafer map identification method and computer-readable recording medium | |
| CN115879017A (en) | Automatic classification and grading method and device for power sensitive data and storage medium | |
| CN114240817A (en) | Data analysis method and device, electronic equipment and storage medium | |
| WO2023273249A1 (en) | Tsvm-model-based abnormality detection method for automatic verification system of smart electricity meter | |
| CA3035387C (en) | Digitization of industrial inspection sheets by inferring visual relations | |
| CN114492542B (en) | Method and device for detecting running state of on-load tap-changer | |
| JP5905375B2 (en) | Misclassification detection apparatus, method, and program | |
| CN115729761B (en) | Hard disk fault prediction method, system, equipment and medium | |
| CN112085064A (en) | Transformer fault diagnosis method based on multi-classification probability output of support vector machine | |
| CN113722239B (en) | Airborne embedded software quality detection method, device, medium and electronic equipment | |
| Jia Yong et al. | Classification of fault and stray gassing in transformer by using duval pentagon and machine learning algorithms | |
| Niu et al. | Application of data mining technology based on RVM for power transformer fault diagnosis | |
| CN113569957A (en) | Object type identification method and device of business object and storage medium | |
| CN111258788A (en) | Disk failure prediction method, device and computer readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |