CN109033612B - Transformer fault diagnosis method based on vibration noise and BP neural network - Google Patents
Transformer fault diagnosis method based on vibration noise and BP neural network Download PDFInfo
- Publication number
- CN109033612B CN109033612B CN201810805455.9A CN201810805455A CN109033612B CN 109033612 B CN109033612 B CN 109033612B CN 201810805455 A CN201810805455 A CN 201810805455A CN 109033612 B CN109033612 B CN 109033612B
- Authority
- CN
- China
- Prior art keywords
- neural network
- vibration
- transformer
- neurons
- transformer fault
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Geometry (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention discloses a transformer fault diagnosis method based on vibration noise and BP neural network, which relates to the technical field of transformer fault treatment and comprises the following steps: s1, collecting vibration noise sound pressure signals of all areas of a transformer through a noise source identification module, and obtaining an area where a maximum noise source is located according to the vibration noise sound pressure signals; s2, collecting vibration signals from the area where the maximum noise source is located through a vibration signal measurement module; and S3, performing transformer fault diagnosis on the vibration signal by adopting a BP neural network algorithm. According to the invention, through the collection of the noise sound pressure signals of the transformer by the S1 and the S2 and the combination of the transformer fault diagnosis of the vibration signals by the BP neural network algorithm, the fault diagnosis precision is greatly improved, and the defect of difficult transformer fault diagnosis by the vibration signals of the transformer state information is overcome.
Description
Technical Field
The invention belongs to the technical field of transformer fault processing, and particularly relates to a transformer fault diagnosis method based on vibration noise and a BP neural network.
Background
Power transformers are one of the most important devices in power systems, and once a transformer fails, it will have a tremendous impact on the grid. Therefore, the maintenance and inspection work of the transformer is very important. However, the transformer has a complex structure, large maintenance workload and high difficulty. And the frequent disassembly and assembly of the transformer also easily damages components and parts, and reduces the operation reliability of the transformer. Live fault diagnosis is particularly urgent for transformer fault diagnosis. Among the methods for diagnosing faults of transformers, the vibration method is a reliable fault diagnosis technique for transformers which has been developed more rapidly in recent years and has a good prospect. The vibration method performs fault diagnosis by analyzing a mailbox vibration signal of the transformer. The transformer vibration signal contains fault information of the transformer, but it is a great difficulty to obtain a vibration signal of the overall transformer state information due to the large size of the transformer. Therefore, there is an urgent need for a more efficient method for fault diagnosis of transformers in the power grid.
Disclosure of Invention
The invention aims to provide a transformer fault diagnosis method, which solves the defect that the existing transformer fault diagnosis is difficult by a vibration signal of transformer state information.
In order to achieve the above purpose, the invention provides a transformer fault diagnosis method based on vibration noise and BP neural network, comprising the following steps:
s1, collecting vibration noise sound pressure signals of all areas of a transformer through a noise source identification module, and obtaining an area where a maximum noise source is located according to the vibration noise sound pressure signals;
s2, collecting vibration signals from the area where the maximum noise source is located through a vibration signal measurement module;
and S3, performing transformer fault diagnosis on the vibration signal by adopting a BP neural network algorithm.
Further, the step S1 includes:
s10, carrying out Fourier change on the vibration noise sound pressure signal to obtain a sound pressure level maximum frequency band;
s11, carrying out beam forming algorithm calculation on the vibration noise sound pressure signal corresponding to the maximum frequency band of the sound pressure level to obtain the position of the maximum noise source of the vibration noise sound pressure of each sound source area on the surface of the transformer box body;
s12, determining the area where the maximum noise source is located according to the position where the maximum noise source is located.
Furthermore, the vibration signal measuring module collects the vibration signals by adopting an acceleration sensor array.
Further, the step S3 includes:
s30, obtaining training sample data of a plurality of groups of vibration signals and transformer fault types corresponding to the training sample data according to the vibration signals;
s31, performing FFT processing on each group of training sample data to obtain characteristic quantities of vibration signals, wherein the characteristic quantities are used as neurons of an input layer of a BP neural network, and the quantity of the characteristic quantities is m;
s32, carrying out normalization processing on the characteristic quantities of each group to obtain normalized data of the characteristic quantities of each group;
s33, encoding the transformer fault types to obtain the number of the transformer fault types, wherein the transformer fault types are neurons of an BP neural network output layer, and the number of the transformer fault types is the number n of the neurons of the output layer;
s34, obtaining the hidden layer neuron number h of the BP neural network through the input layer neuron number m and the output layer neuron number n;
s35, constructing an initial BP neural network according to the number of the neurons of the input layer being m, the number of the neurons of the output layer being n and the number of the neurons of the hidden layer of the BP neural network being h;
s36, training the BP neural network according to the characteristic quantity of each group and the fault type of the transformer corresponding to the characteristic quantity of each group to obtain a trained BP neural network;
and S37, extracting the characteristic quantity of the transformer vibration signal to be diagnosed, and diagnosing by adopting the BP neural network obtained after training to obtain the fault type of the transformer vibration signal.
Further, the characteristic amount of the vibration signal includes: fundamental frequency specific gravity, basic amplitude, dominant frequency specific gravity, dominant frequency amplitude and vibration entropy.
Further, the transformer fault types include: no faults, core faults and winding faults.
Further, the step S34 includes:
s3400, calculating the number of hidden layer neurons by adopting the following steps:
wherein h is the number of neurons of an hidden layer, a is an adjustment constant between 1 and 10, m is the number of neurons of an input layer, and n is the number of neurons of an output layer;
s3401, from h min Initially, the number of neurons is increased one by one until h max Will [ h ] min ,h max ]Respectively performing training verification;
s3402, selecting the number of neurons corresponding to the optimal verification result in the training verification result, wherein the number of neurons is the number h of neurons of an hidden layer of the BP neural network.
Further, the BP neural network training in S36 includes the following steps:
s3600, inputting the characteristic quantity of each group of training sample data obtained in the step S31 into the BP neural network to obtain an output result and an error of the output result;
s3601, judging whether the error is smaller than a preset threshold value, and if the error is larger than the preset threshold value, entering step S3602; if the error is smaller than the preset threshold, step S3603 is performed;
s3602, reversely transferring the initial BP neural network by adopting a gradient descent method, continuously adjusting the weight and the threshold value of the network along the steepest descent direction of the sum of squares of relative errors, reversely calculating by an output layer, an implicit layer and an input layer, outputting the result of each layer, and returning to the step S35;
s3603, constructing a trained BP neural network by adopting the output result.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the transformer fault diagnosis method based on the vibration noise and the BP neural network, the area where the maximum noise source of the transformer is located is obtained through the noise source identification module, signals are collected through the vibration signal measurement module, and finally the transformer fault diagnosis is carried out on the vibration signals through the BP neural network fault diagnosis module by adopting the BP neural network algorithm. The BP neural network is calculated through various characteristic quantities and various data in the BP neural network fault diagnosis module, so that the defect that the transformer fault diagnosis is difficult through the vibration signals of the transformer state information is overcome; the combination of the noise source identification module, the vibration signal measurement module and the BP neural network fault diagnosis module greatly improves the fault diagnosis precision.
2. The vibration signal measuring module provided by the invention adopts the acceleration sensor array to collect the vibration signal of the area where the maximum noise source is located, and the acceleration sensor array is arranged at the position of the transformer shell where the maximum noise source is located, which is nearest to the vibration noise source, so that the vibration signal with the most abundant information can be collected, and a data base is provided for fault diagnosis.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawing in the description below is only one embodiment of the present invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a transformer fault diagnosis method based on vibration noise and a BP neural network according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made more apparent and fully by reference to the accompanying drawings, in which it is shown, however, only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the transformer fault diagnosis method based on vibration noise and BP neural network provided by the invention comprises the following steps:
s1, collecting vibration noise sound pressure signals P of all areas of a transformer by adopting a microphone array through a noise source identification module n And obtaining the area where the maximum noise source is located according to the vibration noise sound pressure signal. The noise source identification module is adopted to intuitively present the position and the intensity of the vibration noise source in the form of an image. S1 comprises the following steps:
s10, the vibration noise sound pressure signal P n Fourier transform is carried out to obtain the maximum frequency band f of sound pressure level max ；
S11, setting the maximum frequency band f of sound pressure level max Carrying out beam forming algorithm calculation on the corresponding vibration noise sound pressure signals to obtain the position of the maximum noise source of the vibration noise sound pressure of each sound source area on the surface of the transformer box body;
the computational formula of the beamforming algorithm is:
in the formula (1), B (t, theta) is the maximum frequency band f of sound pressure level max Corresponding to the position of the vibration noise sound pressure signal, P n Is the maximum frequency band f of sound pressure level max A corresponding vibration noise sound pressure signal; θ is the direction of focusing of the sound source; k (k) n Is the maximum frequency band f of sound pressure level max The weight vector of the corresponding vibration noise sound pressure signal, where k is taken n =1; n=1, 2. Once again, the total number of the components, N is the number of the microphones, τ is the compensation delay, τ=l n sinθ/c 0 ，l n Is the distance between the microphone with the number n and the reference microphone, c 0 The propagation speed of sound waves in a propagation medium is represented by t, which is the measured time;
s12, determining an area W where the maximum noise source is located according to the position where the maximum noise source is located.
S2, acquiring vibration signals P from an area W where the maximum noise source is located by adopting an acceleration sensor array through a vibration signal measuring module, and arranging the acceleration sensor array at a transformer shell closest to the vibration noise source where the maximum noise source is located, so that the vibration signals with the most abundant information can be acquired.
S3, performing transformer fault diagnosis on the vibration signal by adopting a BP neural network algorithm; which comprises the following steps:
s30, obtaining training sample data of a plurality of groups of vibration signals and transformer fault types corresponding to the training sample data of each group according to the vibration signals.
S31, performing FFT processing on each group of training sample data to obtain characteristic quantities of vibration signals, wherein the obtained characteristic quantities are used as neurons of an input layer of the BP neural network, and the number of the characteristic quantities is m; the characteristic quantity comprises fundamental frequency proportion, basic amplitude, dominant frequency proportion, dominant frequency amplitude and vibration entropy, namely, when the number m=5 of neurons of the input layer of the BP neural network, the calculation formula of the characteristic quantity is as follows:
fundamental frequency amplitude: the fundamental frequency f of the vibration signal is 2 times of the fundamental frequency of the current signal, so the fundamental frequency f of the vibration signal is 100Hz, and the fundamental frequency amplitude A is obtained after FFT processing 100 ；
Fundamental frequency specific gravity:
in the formula (2), P 100 The fundamental frequency f of the vibration signal is the fundamental frequency specific gravity of 100Hz, f is the fundamental frequency of the vibration signal, f max For maximum fundamental frequency of vibration signal, A f The amplitude corresponding to the frequency f in the vibration spectrum;
main frequency amplitude: selecting integer times of a fundamental frequency f of a vibration signal, wherein the fundamental frequency f is 100Hz, and the selected amplitude is as follows: 200Hz、300Hz、400Hz......f max Hz, the amplitude value with the largest amplitude value is selected as the main frequency amplitude value A main I.e. A main ＝max{A 200 ,A 300 ,A 400 ......A max }；
Main frequency specific gravity:
in the formula (3), P main The fundamental frequency f of the vibration signal is the main frequency specific gravity of 100;
vibration entropy:
in the formula (4), P f For the specific gravity of the fundamental frequency f of the vibration signal, TVE is the vibration entropy of the fundamental frequency f of the vibration signal.
In embodiment 1, the noise source identification module and the vibration signal measurement module detect the iron core and the winding of the transformer in the fault state and the normal state, so as to obtain the vibration signal of the iron core fault and the normal state and the vibration signal of the winding fault and the normal state, and the step S31 calculates the vibration signal of the iron core fault and the normal state and the vibration signal of the winding fault and the normal state, so as to obtain the characteristic quantity of the winding fault sample and the characteristic quantity of the iron core fault sample, as shown in table 1 and table 2.
Table 1: winding fault sample feature
| Amplitude of fundamental frequency | Specific gravity of fundamental frequency | Amplitude of dominant frequency | Major frequency specific gravity | Vibration entropy | |
| Normal state | 0.159 | 3％ | 0.698 | 64％ | 1.6091 |
| Failure of | 0.299 | 7％ | 0.957 | 75％ | 1.3177 |
Table 2: core fault sample feature
| Amplitude of fundamental frequency | Specific gravity of fundamental frequency | Amplitude of dominant frequency | Major frequency specific gravity | Vibration entropy | |
| Normal state | 0.60 | 3％ | 1.4322 | 57％ | 2.0457 |
| Failure of | 0.7895 | 3.1％ | 2.1659 | 75％ | 1.3537 |
S32, carrying out normalization processing on each group of characteristic quantities to obtain normalized data of each group of characteristic quantities; the formula of normalization processing is:
in the formula (5), x i Is the data after the normalization and is used for the data,
is a sample feature quantity, +.>
And->
The minimum data and the maximum data of the sample feature quantity, respectively.
S33, encoding the fault types of the transformers to obtain the numbers of the fault types of the transformersAn amount of; the fault type of the transformer is the neuron of the BP neural network output layer, and the number of the neurons of the output layer is n. The transformer fault types include: fault-free y k1 Failure of iron core y k2 Winding fault y k3 The BP neuron network outputs a layer neuron number n=3. The method for coding the fault type of the transformer is as follows: y when the transformer is in the ith state ki ＝1,y kj =0 (j+.i), network output is Y k ＝[y k1 ,y k2 ,y k3 ]. In order to make BP neural network have better generalization capability, the output y of network training sample in actual operation ki ＝0.9,y kj ＝0.1(j≠i)；
The sample feature values of table 1 and table 2 obtained in example 1 are sequentially input into the BP neural network, so as to obtain a BP neural network output fault code of the winding sample and a BP neural network output fault code of the core fault sample, respectively, as shown in table 3 and table 4.
Table 3: BP neural network output fault code of winding sample
Table 4: BP neural network output fault code of iron core fault sample
S34, obtaining the hidden layer neuron number h of the BP neural network through the input layer neuron number m and the output layer neuron number n, wherein the hidden layer neuron number h comprises the following steps:
s3400, calculating the number of hidden layer neurons by adopting the following steps:
in the formula (6), h is the number of neurons of an hidden layer, a is an adjustment constant between 1 and 10, m is the number of neurons of an input layer, and n is the number of neurons of an output layer;
s3401, from h min Initially, the number of neurons is increased one by one until h max Will [ h ] min ,h max ]Respectively performing training verification;
s3402, selecting the number of neurons corresponding to the optimal verification result in the training verification result, wherein the number of neurons is the number h of neurons of an hidden layer of the BP neural network; wherein, the calculation of the optimal verification result is as follows, for example:
1. firstly, training a sample library (known in the fault type of a transformer);
2. each sample in the sample library has a known, fixed fault type code, i.e., a standard library;
important parameter h of BP neural network, range is [ h ] min ,h max ]The training process is to carry h into the calculation, namely step S3401;
4. step S3401 is carried out by h1, a sample library is calculated by BP neural network to obtain codes corresponding to each sample, namely an output library 1, and a deviation E1 exists between the output library 1 and a standard library; similarly, when h2 is brought into iteration, an output library 2 is obtained, and the deviation between the output library 2 and the standard library is E2; and similarly, each h corresponds to a deviation E, the minimum deviation E is selected, and the h corresponding to the minimum deviation E is the optimal verification result.
In embodiment 2, the vibration signal measurement module measures 100 groups of original vibration signals of the transformer by using an acceleration sensor, wherein 30 groups are iron core fault vibration signals, 30 groups are winding fault vibration signals, and 40 groups are fault-free vibration signals. The construction of the BP neural network is 5-8-3, namely 5 neurons are arranged on the input layer, 3 neurons are arranged on the output layer, and 8 neurons are arranged on the hidden layer through a transformer fault diagnosis method based on vibration noise and the BP neural network.
S35, according to the number of the neurons of the input layer being m, the number of the neurons of the output layer being n, the number of the neurons of the hidden layer of the BP neural network being h, the weight omega between the neurons of the input layer i and the neurons of the hidden layer j ij And weights ω between hidden layer neuron j and output layer neuron k jk Construction of the first stageAn initial BP neural network;
wherein, the activation function of the BP neural network adopts an S-shaped function f (x):
the output of hidden layer neuron j is s j ：
In the formula (8), ω ij For inputting weights between layer neuron i and hidden layer neuron j, θ 1 ，θ 2 For the offset, x i The i-th input quantity of the output layer;
the output of the output layer neuron k is y k ：
In the formula (9), ω jk For weights between hidden layer neuron j and output layer neuron k, θ 1 ，θ 2 Is the offset.
S36, training the BP neural network according to the characteristic quantity of each group and the fault type of the transformer corresponding to the characteristic quantity of each group to obtain a trained BP neural network, wherein the BP neural network training comprises the following steps:
s3600, inputting the feature values of the training sample data obtained in the step S31 into an initial BP neural network to obtain an output result, and obtaining an error of the output result through the output result; the calculation formula of the total error E is as follows:
in the formula (10), d ij The j-th encoded output value for the i-th sample, y ij The ith sample, jthAn encoded expected value;
s3601, judging whether the error is smaller than a preset threshold value, and if the error is larger than the preset threshold value, entering step S3602; if the error is smaller than the preset threshold, step S3603 is performed;
s3602, reversely transferring the initial BP neural network by adopting a gradient descent method, continuously adjusting the weight and the threshold value of the network along the steepest descent direction of the sum of squares of relative errors, reversely calculating by an output layer, an implicit layer and an input layer, outputting the result of each layer, and returning to the step S35;
s3603, constructing the trained BP neural network by adopting the output result.
And S37, extracting the characteristic quantity of the transformer vibration signal to be diagnosed, and diagnosing by adopting the BP neural network obtained after training to obtain the fault type of the transformer vibration signal.
The foregoing disclosure is merely illustrative of specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art will readily recognize that changes and modifications are possible within the scope of the present invention.
Claims (4)
1. A transformer fault diagnosis method based on vibration noise and BP neural network is characterized in that: the method comprises the following steps:
s1, collecting vibration noise sound pressure signals of all areas of a transformer through a noise source identification module, and obtaining an area where a maximum noise source is located according to the vibration noise sound pressure signals;
the S1 comprises the following steps:
s10, carrying out Fourier change on the vibration noise sound pressure signal to obtain a sound pressure level maximum frequency band;
s11, carrying out beam forming algorithm calculation on the vibration noise sound pressure signal corresponding to the maximum frequency band of the sound pressure level to obtain the position of the maximum noise source of the vibration noise sound pressure of each sound source area on the surface of the transformer box body;
s12, determining an area where the maximum noise source is located according to the position where the maximum noise source is located;
s2, collecting vibration signals from the area where the maximum noise source is located through a vibration signal measurement module;
s3, performing transformer fault diagnosis on the vibration signal by adopting a BP neural network algorithm;
the step S3 comprises the following steps:
s30, obtaining training sample data of a plurality of groups of vibration signals and transformer fault types corresponding to the training sample data according to the vibration signals;
s31, performing FFT processing on each group of training sample data to obtain characteristic quantities of vibration signals, wherein the characteristic quantities are used as neurons of an input layer of a BP neural network, and the number of the characteristic quantities is the number m of the neurons of the input layer;
s32, carrying out normalization processing on the characteristic quantities of each group to obtain normalized data of the characteristic quantities of each group;
s33, encoding the transformer fault types to obtain the number of the transformer fault types, wherein the transformer fault types are neurons of an BP neural network output layer, and the number of the transformer fault types is the number n of the neurons of the output layer;
s34, obtaining the hidden layer neuron number h of the BP neural network through the input layer neuron number m and the output layer neuron number n;
s35, constructing a BP neural network according to the number m of neurons of the input layer, the number n of neurons of the output layer and the number h of neurons of an hidden layer of the BP neural network;
s36, training the BP neural network according to the characteristic quantity of each group and the fault type of the transformer corresponding to the characteristic quantity of each group to obtain a trained BP neural network;
s37, extracting characteristic quantity of a transformer vibration signal to be diagnosed, and diagnosing by adopting a trained BP neural network to obtain a fault type of the transformer vibration signal;
the characteristic quantity of the vibration signal includes: fundamental frequency specific gravity, basic amplitude, dominant frequency specific gravity, dominant frequency amplitude and vibration entropy;
the transformer fault types include: no faults, core faults and winding faults.
2. The transformer fault diagnosis method according to claim 1, wherein: the vibration signal measuring module collects the vibration signals by adopting an acceleration sensor array.
3. The transformer fault diagnosis method according to claim 1, wherein: the S34 includes:
s3400, calculating the number of hidden layer neurons by adopting the following steps:
wherein h is the number of neurons of an hidden layer, a is an adjustment constant between 1 and 10, m is the number of neurons of an input layer, and n is the number of neurons of an output layer;
s3401, from h min Initially, the number of neurons is increased one by one until h max Will [ h ] min ,h max ]Respectively performing training verification;
s3402, selecting the number of neurons corresponding to the optimal verification result in the training verification result, wherein the number of neurons is the number h of neurons in an hidden layer of the BP neural network.
4. The transformer fault diagnosis method according to claim 1, wherein: the BP neural network training in S36 includes the following steps:
s3600, inputting the characteristic quantity of each group of training sample data obtained in the step S31 into the BP neural network to obtain an output result and an error of the output result;
s3601, judging whether the error is smaller than a preset threshold value, and if the error is larger than the preset threshold value, entering step S3602; if the error is smaller than the preset threshold, step S3603 is performed;
s3602, the BP neural network adopts a gradient descent method to reversely transfer, continuously adjusts the weight and the threshold value of the network along the steepest descent direction of the sum of squares of relative errors, reversely calculates through an output layer, an hidden layer and an input layer, outputs the result of each layer, and returns to the step S35;
s3603, constructing a trained BP neural network by adopting the output result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810805455.9A CN109033612B (en) | 2018-07-20 | 2018-07-20 | Transformer fault diagnosis method based on vibration noise and BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810805455.9A CN109033612B (en) | 2018-07-20 | 2018-07-20 | Transformer fault diagnosis method based on vibration noise and BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109033612A CN109033612A (en) | 2018-12-18 |
| CN109033612B true CN109033612B (en) | 2023-05-05 |
Family
ID=64644882
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810805455.9A Active CN109033612B (en) | 2018-07-20 | 2018-07-20 | Transformer fault diagnosis method based on vibration noise and BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109033612B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110174255B (en) * | 2019-06-03 | 2021-04-27 | 国网上海市电力公司 | Transformer vibration signal separation method based on radial basis function neural network |
| CN110689069A (en) * | 2019-09-25 | 2020-01-14 | 贵州电网有限责任公司 | Transformer fault type diagnosis method based on semi-supervised BP network |
| CN111026095B (en) * | 2019-12-30 | 2020-12-04 | 太原科技大学 | Fault diagnosis method with noise label based on recurrent neural network |
| CN111175698B (en) * | 2020-01-18 | 2022-12-20 | 国网山东省电力公司菏泽供电公司 | Transformer noise source positioning method, system and device based on sound and vibration combination |
| CN111638028B (en) * | 2020-05-20 | 2022-05-10 | 国网河北省电力有限公司电力科学研究院 | High-voltage parallel reactor mechanical state evaluation method based on vibration characteristics |
| CN112098066A (en) * | 2020-09-21 | 2020-12-18 | 陕西师范大学 | High-voltage shunt reactor fault diagnosis method and system based on gate control circulation unit |
| CN112150744A (en) * | 2020-09-24 | 2020-12-29 | 国网山东省电力公司临沂供电公司 | Transformer vibration noise abnormity analysis early warning system |
| CN112307918A (en) * | 2020-10-21 | 2021-02-02 | 华北电力大学 | Diagnosis method for transformer direct-current magnetic biasing based on fuzzy neural network |
| CN112665707B (en) * | 2020-12-15 | 2023-03-03 | 国网天津市电力公司电力科学研究院 | Cumulative effect after short circuit impact of transformer and diagnosis method |
| CN112879278B (en) * | 2021-01-11 | 2022-09-30 | 苏州欣皓信息技术有限公司 | Pump station unit fault diagnosis method based on noise signal A weighting analysis |
| CN113283310A (en) * | 2021-05-07 | 2021-08-20 | 国网浙江省电力有限公司武义县供电公司 | System and method for detecting health state of power equipment based on voiceprint features |
| CN113392511B (en) * | 2021-05-28 | 2022-11-22 | 广西电网有限责任公司电力科学研究院 | On-load tap-changer mechanical state monitoring method based on frequency spectrum envelope symbol entropy |
| CN113963714A (en) * | 2021-10-28 | 2022-01-21 | 广东电网有限责任公司佛山供电局 | Method and system for separating noise of transformer body from noise of cooling device |
| CN114235366A (en) * | 2021-12-14 | 2022-03-25 | 国网天津市电力公司电力科学研究院 | Transformer fault identification method based on transformer noise vibration characteristics |
| CN115358110A (en) * | 2022-07-25 | 2022-11-18 | 国网江苏省电力有限公司淮安供电分公司 | Transformer fault detection system based on acoustic sensor array |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246043A (en) * | 2008-03-28 | 2008-08-20 | 清华大学 | On-line monitoring method for vibration and noise of AC power transformer influenced by DC magnetic biasing |
| JP2016051850A (en) * | 2014-09-01 | 2016-04-11 | 株式会社ダイヘン | Stationary induction apparatus |
| CN106066468A (en) * | 2016-05-25 | 2016-11-02 | 哈尔滨工程大学 | A kind of based on acoustic pressure, the vector array port/starboard discrimination method of vibration velocity Mutual spectrum |
| CN107271022A (en) * | 2017-05-31 | 2017-10-20 | 国网山西省电力公司电力科学研究院 | A kind of transformer vibration noise Integrated Measurement System |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103218662A (en) * | 2013-04-16 | 2013-07-24 | 郑州航空工业管理学院 | Transformer fault diagnosis method based on back propagation (BP) neural network |
| CN103245524A (en) * | 2013-05-24 | 2013-08-14 | 南京大学 | Acoustic fault diagnosis method based on neural network |
| CN105353255B (en) * | 2015-11-27 | 2018-07-31 | 南京邮电大学 | A kind of Diagnosis Method of Transformer Faults based on neural network |
| CN107748314A (en) * | 2017-10-18 | 2018-03-02 | 国网重庆市电力公司北碚供电分公司 | Transformer Faults Analysis system based on sound wave shock detection |
-
2018
- 2018-07-20 CN CN201810805455.9A patent/CN109033612B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246043A (en) * | 2008-03-28 | 2008-08-20 | 清华大学 | On-line monitoring method for vibration and noise of AC power transformer influenced by DC magnetic biasing |
| JP2016051850A (en) * | 2014-09-01 | 2016-04-11 | 株式会社ダイヘン | Stationary induction apparatus |
| CN106066468A (en) * | 2016-05-25 | 2016-11-02 | 哈尔滨工程大学 | A kind of based on acoustic pressure, the vector array port/starboard discrimination method of vibration velocity Mutual spectrum |
| CN107271022A (en) * | 2017-05-31 | 2017-10-20 | 国网山西省电力公司电力科学研究院 | A kind of transformer vibration noise Integrated Measurement System |
Non-Patent Citations (1)
| Title |
|---|
| "220kV油浸式变压器振动与噪声试验研究";马裕超等;《变压器》;20170725;第54卷(第07期);第49-55页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109033612A (en) | 2018-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109033612B (en) | Transformer fault diagnosis method based on vibration noise and BP neural network | |
| CN108426713B (en) | Rolling bearing weak fault diagnosis method based on wavelet transformation and deep learning | |
| CN112906473B (en) | Fault diagnosis method for rotary equipment | |
| WO2022116570A1 (en) | Microphone array-based method for locating and identifying fault signal in industrial equipment | |
| CN107748314A (en) | Transformer Faults Analysis system based on sound wave shock detection | |
| CN105196114B (en) | Tool wear realtime on-line monitoring method based on wavelet analysis and neutral net | |
| Oliveira et al. | Ultrasound-based identification of damage in wind turbine blades using novelty detection | |
| CN105678343B (en) | Hydropower Unit noise abnormality diagnostic method based on adaptive weighted group of sparse expression | |
| CN102122133A (en) | Self-adaption wavelet neural network abnormity detection and fault diagnosis classification system and method | |
| CN108256556A (en) | Wind-driven generator group wheel box method for diagnosing faults based on depth belief network | |
| CN110969096A (en) | Motor fault mode diagnosis method based on particle swarm optimization support vector machine | |
| KR20200014129A (en) | Diagnosis method of electric transformer using Deep Learning | |
| CN109323754A (en) | A kind of train wheel polygon fault diagnosis detection method | |
| CN103033567A (en) | Pipeline defect signal identification method based on guided wave | |
| CN111239549A (en) | Power distribution fault rapid positioning method based on discrete wavelet transform | |
| CN111695452B (en) | RBF neural network-based parallel reactor internal aging degree assessment method | |
| CN113283310A (en) | System and method for detecting health state of power equipment based on voiceprint features | |
| CN202119467U (en) | Self-adaptive wavelet neural network categorizing system of anomaly detection and fault diagnosis | |
| CN110398647A (en) | Transformer state monitoring method | |
| CN110488675A (en) | A kind of substation's Abstraction of Sound Signal Characteristics based on dynamic time warpping algorithm | |
| CN115358110A (en) | Transformer fault detection system based on acoustic sensor array | |
| CN117076955A (en) | Fault detection method and system for high-voltage frequency converter | |
| CN112348052A (en) | Power transmission and transformation equipment abnormal sound source positioning method based on improved EfficientNet | |
| CN115542099A (en) | Online GIS partial discharge detection method and device | |
| CN115600088A (en) | Distribution transformer fault diagnosis method based on vibration signals |
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 |