CN112114215A - Transformer aging evaluation method and system based on error back propagation algorithm - Google Patents
Transformer aging evaluation method and system based on error back propagation algorithm Download PDFInfo
- Publication number
- CN112114215A CN112114215A CN202010977524.1A CN202010977524A CN112114215A CN 112114215 A CN112114215 A CN 112114215A CN 202010977524 A CN202010977524 A CN 202010977524A CN 112114215 A CN112114215 A CN 112114215A
- Authority
- CN
- China
- Prior art keywords
- transformer
- signal
- running time
- acoustic signal
- neural network
- 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.)
- Pending
Links
- 230000032683 aging Effects 0.000 title claims abstract description 58
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 27
- 238000011156 evaluation Methods 0.000 title claims abstract description 25
- 238000007781 pre-processing Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000003062 neural network model Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims description 33
- 210000002569 neuron Anatomy 0.000 claims description 26
- 238000013528 artificial neural network Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 22
- 238000000354 decomposition reaction Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000001914 filtration Methods 0.000 claims description 15
- 230000005236 sound signal Effects 0.000 claims description 12
- 238000010606 normalization Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 8
- 230000009467 reduction Effects 0.000 claims description 8
- 230000008030 elimination Effects 0.000 claims description 6
- 238000003379 elimination reaction Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 9
- 230000007547 defect Effects 0.000 abstract description 5
- 238000012423 maintenance Methods 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000012535 impurity Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000003878 thermal aging Methods 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- 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
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
-
- 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)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application relates to a transformer aging evaluation method and system based on an error back propagation algorithm, in particular to the field of equipment detection. The method comprises the steps of preprocessing an acquired acoustic signal; inputting the preprocessed acoustic signals into a preset neural network model, and predicting the running time of the transformer to be tested; and comparing the predicted running time with the actual running time to output the aging degree of the transformer. The noise signals transmitted by the transformer during the operation period contain rich internal mechanical state information, and the analysis of the noise signals of the transformer can accurately evaluate and diagnose the internal mechanical defects or faults of the transformer timely and effectively. The aging degree of the internal structure of the transformer is different, the characteristics of the transmitted noise signals are also different, and the operation state of the internal structure of the transformer is reflected on the basis of the characteristics without electrical connection and shutdown detection with the transformer, so that the monitoring and maintenance cost is reduced, and the operation completeness of power equipment is improved.
Description
Technical Field
The application relates to the field of equipment detection, in particular to a transformer aging evaluation method and system based on an error back propagation algorithm.
Background
Transformers are one of the most critical devices in an electrical power system. These devices develop defects due to mechanical and thermal aging during long-term operation, and seriously affect the stable operation of the entire power system.
In the prior art, mechanical defects and even serious faults can be generated in the internal part of many transformers before the design service life is reached. According to the long-term statistics of transformers, the main source of accidents is related to the core and the windings. Even if the transformer continues to operate, it may have certain inherent drawbacks that reduce the short-circuit resistance.
However, the detection method in the prior art rarely utilizes noise signals propagated around the transformer to perform aging evaluation, and the traditional method hardly adopts a neural network prediction method to predict the operation time of the transformer, and has low detection efficiency and insufficient safety, so that the internal aging degree of the transformer cannot be timely and effectively reflected.
Disclosure of Invention
The invention aims to provide a transformer aging evaluation method and system based on an error back propagation algorithm aiming at the defects in the prior art, so as to solve the problems that in the prior art, the aging evaluation is rarely carried out by using noise signals propagated to the periphery of a transformer, the operation time of the transformer is hardly predicted by adopting a neural network prediction method, the detection efficiency is low, the safety is insufficient, and the internal aging degree of the transformer cannot be timely and effectively reflected. In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:
in a first aspect, the present application provides a method for evaluating transformer aging based on an error back propagation algorithm, the method including:
acquiring an acoustic signal of a transformer to be tested;
preprocessing the acquired acoustic signals to obtain characteristic information of the acoustic signals, wherein the preprocessing comprises the following steps: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing;
inputting the characteristic information of the acoustic signal into a preset neural network model, and calculating to obtain the predicted operation time of the transformer to be tested;
and comparing the predicted running time with the actual running time to output the aging degree of the transformer.
Optionally, the step of acquiring the acoustic signal of the transformer to be tested includes:
determining the number of neurons of an input layer and an output layer of a neural network prediction model, determining the number of hidden layers and the number of neuron nodes included in each hidden layer, initializing weights and training a neural network;
acquiring multi-channel acoustic signals of a plurality of groups of transformers with the same voltage grade in the same time period, and obtaining time domain characteristics and the running time of the transformers corresponding to the time domain characteristics through the acoustic signals and inputting the time domain characteristics into neural network training;
and obtaining a neural network model through multiple iterative training, wherein the trained neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
Optionally, the step of preprocessing the acquired acoustic signal further includes: normalizing the acquired acoustic signals, wherein the normalization adopts a formula as follows:
wherein x isiThe sound pressure value at the ith moment in a certain section of noise signal is in a form of a numerical value and has a unit of Pa; x is the number ofminRepresents the minimum sound pressure value in the section of signal, and the minimum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofmaxRepresents the maximum sound pressure value in the section of signal, and the maximum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofi' denotes a sound pressure value at the i-th time obtained by normalizing the sound signal, in the form of a numerical value in Pa.
Optionally, the step of preprocessing the acquired acoustic signal specifically includes:
performing low-pass filtering on the acquired acoustic signals when the transformer is in each state by using a Butterworth low-pass digital filter;
and performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
Optionally, the step of outputting the aging degree of the transformer according to the comparison between the predicted operation time and the actual operation time specifically includes:
the predicted running time is differed from the actual running time to obtain a running time difference value;
matching the running time difference with a preset running threshold value;
and outputting the aging degree of the transformer according to the matching result.
In a second aspect, the present application provides a system for transformer aging evaluation based on an error back propagation algorithm, the system comprising: the device comprises an acquisition module, a preprocessing module, a calculation module and an output module;
the acquisition module is used for acquiring acoustic signals of the transformer to be tested, eight microphones are arranged around each transformer, and noise signals transmitted when the transformers run are synchronously acquired;
the preprocessing module is used for preprocessing the acquired acoustic signals to obtain characteristic information of the acoustic signals, wherein the preprocessing comprises the following steps: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing;
the calculation module is used for inputting the characteristic information of the acoustic signal and the corresponding running time of the transformer into a preset neural network model, inputting the acoustic signal characteristic of the transformer to be researched into the trained network model after network training is finished, and calculating to obtain the predicted running time of the transformer to be researched;
and the output module is used for comparing the predicted running time with the actual running time and outputting the aging degree of the transformer according to the error between the predicted running time and the actual running time.
Optionally, the system further includes a modeling module, configured to determine numbers of neurons in the input layer and the output layer, determine numbers of hidden layers, numbers of neuron nodes included in each hidden layer, weight initialization, and neural network training;
acquiring acoustic signals of a plurality of groups of transformers with the same voltage grade in the same time period, and obtaining time domain characteristics and the running time of the transformers corresponding to the time domain characteristics through the acoustic signals and inputting the time domain characteristics into neural network training;
and obtaining a neural network model through multiple iterative training, wherein the neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
Optionally, the preprocessing module is further configured to perform normalization processing on the acquired acoustic signal, where the normalization processing adopts a formula:
wherein x isiThe sound pressure value at the ith moment in a certain section of noise signal is in a form of a numerical value and has a unit of Pa; x is the number ofminRepresents the minimum sound pressure value in the section of signal, and the minimum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofmaxRepresents the maximum sound pressure value in the section of signal, and the maximum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofi' denotes a sound pressure value at the i-th time obtained by normalizing the sound signal, in the form of a numerical value in Pa.
Optionally, the preprocessing module is specifically configured to perform low-pass filtering on the acoustic signals acquired when the transformer is in each state by using a butterworth low-pass digital filter;
and performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
Optionally, the output module is specifically configured to subtract the predicted running time from the actual running time to obtain a running time difference;
matching the running time difference with a preset running threshold value;
and outputting the aging degree of the transformer according to the matching result.
The invention has the beneficial effects that:
the method comprises the steps of obtaining an acoustic signal of a transformer to be tested; preprocessing the acquired acoustic signals, wherein the preprocessing comprises: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing; inputting the preprocessed acoustic signals into a preset neural network model, and calculating to obtain the predicted running time of the transformer to be tested; and comparing the predicted running time with the actual running time to output the aging degree of the transformer. Because the noise signals transmitted to the air during the operation of the transformer contain rich internal mechanical state information, the analysis of the noise signals outside the transformer can timely and effectively carry out accurate assessment and diagnosis on the internal mechanical defects or faults of the transformer. The transformer inner structure ageing degree is different, the noise signal characteristic that transmits out is also different, this application is through gathering the acoustic information of the transformer that awaits measuring, analysis and processing, data through with the neural network model of predetermineeing contrast, obtain the prediction operating time of this transformer, and contrast prediction operating time and actual operating time, obtain the ageing degree of this transformer that awaits measuring, not only need not detect that should await measuring the transformer shut down, and can get rid of some hidden faults, still reduce danger at the in-process that detects, then make this period can reflect transformer inner structure running state under the condition that does not carry out electrical connection and shut down the detection with the transformer, the cost of monitoring and maintenance has been reduced, the completeness of power equipment operation has been improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed 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 invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
Fig. 1 is a schematic view of an application scenario provided in the present application;
FIG. 2 is a schematic flow chart of a transformer aging evaluation method based on an error back propagation algorithm according to the present application;
FIG. 3 is a schematic flow chart of another method for evaluating transformer aging based on an error back propagation algorithm provided in the present application;
FIG. 4 is a block diagram of a system for transformer aging evaluation based on an error back propagation algorithm according to the present application;
fig. 5 is a block diagram of another system for transformer aging evaluation based on an error back propagation algorithm according to the present application.
Detailed Description
The technical solutions in the present application will be described clearly and completely with reference to the drawings in the present application, and it should be understood that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the present application, as generally described and illustrated in the figures herein, may 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. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
Fig. 1 is a schematic view of an application scenario provided by the present application, and as shown in fig. 1, the method in the present application may be applied to the electronic device 10 shown in fig. 1. As shown in fig. 1, the electronic device 10 may include: memory 11, processor 12, network module 13 and acoustic signal collector 14.
The memory 11, the processor 12, the network module 13 and the acoustic signal collector 14 are electrically connected to each other directly or indirectly to enable transmission or interaction of data. For example, the components may be electrically connected to each other via one or more communication buses or signal lines. The memory 11 stores at least one functional module which can be stored in the memory 11 in the form of software or firmware (firmware), and the processor 12 executes various functional applications and data processing by running the functional module stored in the memory 11 in the form of software or hardware, that is, implements the method executed by the electronic device 10 in the present application.
The Memory 11 may be, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Programmable Read Only Memory (PROM), an Erasable Read Only Memory (EPROM), an electrically Erasable Read Only Memory (EEPROM), a magnetic disk, a solid state disk, or the like. The memory 11 is used for storing a program, and the processor 12 executes the program after receiving an execution instruction.
The processor 12 may be an integrated circuit chip having data processing capabilities. The Processor 12 may be a general-purpose Processor including a Central Processing Unit (CPU), a Network Processor (NP), and the like. The methods, steps, and logic blocks of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The network module 13 is used for establishing a communication connection between the electronic device 10 and an external communication terminal through a network, and implementing transceiving operations of network signals and data. The network signal may include a wireless signal or a wired signal.
The acoustic signal collector 14 may be a microphone, and the specific setting position is determined according to actual needs, and no specific requirement is made here, generally, the acoustic signal collector 14 may be a microphone of PCB603C01 type, the microphone has a sensitivity of 40mV/Pa, a range of 16-134dB, a frequency band of 20-20kHz, and an applicable temperature of-30 ℃ to +80 ℃. A plurality of microphones can be symmetrically arranged on a transformer, the sampling frequency of each microphone is 10240Hz, the horizontal distance between each microphone and the transformer is 0.1m, and the height of each microphone is about 1/3 of the height of the transformer.
It will be appreciated that the configuration shown in FIG. 1 is merely illustrative and that electronic device 10 may include more or fewer components than shown in FIG. 1 or may have a different configuration than shown in FIG. 1. The components shown in fig. 1 may be implemented in hardware, software, or a combination thereof.
On the basis of the above, the present application further provides a computer-readable storage medium, which includes a computer program, and the computer program controls the electronic device 10 to execute the following method when running.
FIG. 2 is a schematic flow chart of a transformer aging evaluation method based on an error back propagation algorithm according to the present application; as shown in fig. 2, the present application provides a method for evaluating transformer aging based on an error back propagation algorithm, the method comprising:
s101, acquiring an acoustic signal of a transformer to be tested,
in order to comprehensively and synchronously acquire multi-channel noise data, the periphery of each transformer is selected as a noise acquisition area, and a signal acquisition instrument is 0.1m away from the surface of an oil tank and is one third of the height of the transformer. In practical application, the acoustic signal collectors 14 are disposed in a noise collection area of the transformer and configured to collect acoustic signals of the transformer, generally, the number of the acoustic signal collectors 14 is determined according to actual needs, and is not specifically limited herein, for convenience of description, in this embodiment, the number of the acoustic signal collectors 14 is four, for example, four acoustic signal collectors 14 are respectively disposed around the transformer, and each acoustic signal collector 14 is 0.1 meter away from a surface of a transformer tank, and the four acoustic signal collectors 14 collect acoustic signals of the transformer to be tested.
S102, preprocessing the acquired acoustic signals to obtain characteristic information of the acoustic signals.
The method comprises the steps of preprocessing an acquired acoustic signal, namely processing the acquired acoustic signal into data capable of obtaining characteristic information of the acoustic signal, and removing impurity information and interference information in the acoustic signal, wherein the method for removing the impurity information and the interference information comprises the steps of filtering, removing sample data with a larger threshold value or a smaller threshold value, then performing signal denoising, further reducing noise in the acoustic signal, then drawing the acoustic signal without the impurity, drawing a broken line graph or a column graph, performing characteristic extraction on the acoustic signal drawn into the graph, obtaining the characteristic information of the acoustic signal, and then performing dimension reduction on the acoustic signal, namely converting high-dimension data into low-dimension data, optionally directly inputting sound pressure information obtained after noise data normalization into a neural network.
S103, inputting the characteristic information of the acoustic signal into a preset neural network model, and calculating to obtain the predicted running time of the transformer to be tested.
Inputting the characteristic information of the acoustic signal and the running time of the transformer into a preset neural network model, wherein each iteration is subjected to calculation from an input layer to a hidden layer and then to an output layer, namely, an equation (2) and an equation (3), the calculation result is the actual output yk (i) of the running time of the transformer, an error is obtained through calculation of an equation (4), the error is compared with an error threshold, if the error is not less than the error threshold, the weighted value and the bias are updated, and then the next iteration is carried out; if the error is smaller than the error threshold value, the current weight value and the bias value are optimal, the network training is finished, the current output value is the final prediction result, and the current output value is the prediction running time.
Specifically, a training set including acoustic signals and commissioning time of a plurality of transformers is input into a neural network, the neural network initializes a weight value and a bias of each neuron before a first iteration, and a network model performs the following calculation for each iteration:
wherein, Wij (l)Is the weight of the connection between the jth neuron from the l-1 layer and the ith neuron from the l layer; bi (l)For biasing of the ith neuron at layer l, neti (l)And f is the input of the ith neuron of the l layers, the activation function of the neuron, the neuron of each layer becomes the input of the neuron of the next layer after being activated by the activation function, and finally the output of the preset neural network model is the predicted running time of the transformer. H in equation (2) in the calculation from the input layer to the hidden layerj (l-1)H in formula (3) in the calculation process from the last hidden layer to the output layer for inputting the noise signal of the neural networki (l)Is the output result of the neural network.
When the error between the final output result of the network and the ideal result is smaller than the error threshold, the network stops training, and the error function of the output result of the network and the ideal result is calculated as follows:
where dk (i) is the i-th expected output of the network, yk (i) represents the i-th actual output of the network, and i represents the numbers corresponding to different transformers.
Inputting the characteristic information of the acoustic signal and the running time of the transformer into a preset neural network model, wherein each iteration is subjected to calculation from an input layer to a hidden layer and then to an output layer, namely, an equation (2) and an equation (3), the calculation result is the actual output yk (i) of the running time of the transformer, an error is obtained through calculation of an equation (4), the error is compared with an error threshold, if the error is not less than the error threshold, the weighted value and the bias are updated, and then the next iteration is carried out; if the error is smaller than the error threshold value, the current weight value and the bias value are optimal, the network training is finished, the current output value is the final prediction result, the acoustic signal of the transformer to be researched is input into the trained neural network, and the prediction running time of the transformer can be obtained through the formulas (3) and (4) again.
And S104, comparing the predicted running time with the actual running time, and outputting the aging degree of the transformer.
And obtaining the aging degree of the transformer according to the difference time, wherein generally, the larger the difference between the predicted running time and the actual running time is, the more serious the aging degree of the transformer is.
Optionally, the step of acquiring the acoustic signal of the transformer to be tested includes:
determining the number of neurons of an input layer and an output layer of a neural network prediction model, determining the number of hidden layers and the number of neuron nodes included in each hidden layer, initializing weights and training a neural network;
acquiring multi-channel acoustic signals of a plurality of groups of transformers with the same voltage grade in the same time period, and obtaining time domain characteristics and the running time of the transformers corresponding to the time domain characteristics through the acoustic signals and inputting the time domain characteristics into neural network training;
and obtaining a neural network model through multiple iterative training, wherein the trained neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
For the convenience of description, the process of establishing the neural network model is illustrated with the number of input layer neurons being 103, the number of output layer neurons being 1, the number of hidden layer neurons being 3, and the numbers of hidden layer neurons being 30, 20, and 10, respectively. The neuron of the input layer is combined with the corresponding weight value and input into the hidden layer, and the neuron of the hidden layer is combined with the corresponding weight value and input into the output layer, so that an output result is finally obtained.
Optionally, the step of preprocessing the acquired acoustic signal further includes: normalizing the acquired acoustic signals, wherein the normalization adopts a formula as follows:
wherein x isiThe sound pressure value at the ith moment in a certain section of noise signal is in a form of a numerical value and has a unit of Pa; x is the number ofminRepresents the minimum sound pressure value in the section of signal, and the minimum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofmaxRepresents the maximum sound pressure value in the section of signal, and the maximum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofi' denotes a sound pressure value at the i-th time obtained by normalizing the sound signal, in the form of a numerical value in Pa.
Optionally, the step of preprocessing the acquired acoustic signal specifically includes:
and performing low-pass filtering on the acoustic signals of the acquisition transformer in each state by using a Butterworth low-pass digital filter.
The order N of the butterworth low-pass digital filter used for low-pass filtering is 8, the passband cutoff frequency f p is 1000Hz, the stopband cutoff start frequency f s is 1200Hz, the minimum attenuation of fluctuation in the passband R p is 1dB, and the minimum attenuation in the stopband R s is 50 dB.
And performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
The first step, wavelet decomposition. Setting the sound of the transformer processed in the step(s) as S (n), and performing 5-layer wavelet decomposition on the S (n) by using db4 wavelet; and a second step of wavelet coefficient threshold processing, wherein the wavelet coefficient obtained by performing wavelet decomposition on the sound signal in the first step of wavelet decomposition is set as a set threshold value after the wavelet coefficient subjected to the wavelet coefficient threshold processing in the second step of wavelet decomposition.
FIG. 3 is a schematic flow chart of another method for evaluating transformer aging based on an error back propagation algorithm provided in the present application; as shown in fig. 3, optionally, the step of outputting the aging degree of the transformer specifically includes:
s201, the predicted running time is different from the actual running time to obtain a running time difference value.
And (4) subtracting the predicted running time output by the preset neural network model from the time running time to obtain a running time difference value.
And S202, matching the running time difference value with a preset running threshold value.
The running time difference value is matched with a preset running threshold value, the threshold value may be one value or a range, the threshold value may be one or multiple, for example, multiple threshold values are respectively 10, 12 and 15, the multiple threshold values respectively correspond to different aging degrees, wherein the aging degrees of 10, 12 and 15 are sequentially increased.
And S203, outputting the aging degree of the transformer according to the matching result.
And outputting the aging degree of the transformer according to the matching result of the calculated running time difference value and a preset running threshold value, and if the transformer degree is larger, alarming the transformer.
The method comprises the steps of obtaining an acoustic signal of a transformer to be tested; preprocessing the acquired acoustic signals, wherein the preprocessing comprises: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing; inputting the preprocessed acoustic signals into a preset neural network model, and calculating to obtain the predicted running time of the transformer to be tested; the method comprises the steps of comparing the predicted running time with the actual running time, outputting the aging degree of the transformer, acquiring, analyzing and processing an acoustic signal of the transformer to be detected, comparing the acoustic signal with data in a preset neural network model to obtain the predicted running time of the transformer, comparing the predicted running time with the actual running time to obtain the aging degree of the transformer to be detected, not only avoiding shutdown detection of the transformer to be detected, but also eliminating some hidden faults and reducing dangerousness in the detection process.
FIG. 4 is a block diagram of a system for transformer aging evaluation based on an error back propagation algorithm according to the present application; as shown in fig. 4, the present application provides a system for transformer aging evaluation based on an error back propagation algorithm, the system comprising: an acquisition module 31, a preprocessing module 32, a calculation module 33 and an output module 34;
the acquiring module 31 is configured to acquire an acoustic signal of the transformer to be tested;
a preprocessing module 32, configured to perform preprocessing on the acquired acoustic signal, where the preprocessing includes: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing;
the calculation module 33 is configured to input the preprocessed acoustic signals into a preset neural network model, and calculate to obtain the predicted operation time of the transformer to be tested;
and the output module 34 is used for comparing the predicted running time with the actual running time and outputting the aging degree of the transformer.
Fig. 5 is a schematic block diagram of another system for transformer aging evaluation based on an error back propagation algorithm provided in the present application, as shown in fig. 5, optionally, the system further includes a modeling module 35, which is configured to determine the numbers of neurons in the input layer and the output layer, determine the number of hidden layers and the number of neuron nodes included in each hidden layer, initialize weights, and train a neural network;
acquiring acoustic signals of a plurality of groups of transformers with the same grade in different time periods, and obtaining time domain characteristics and data of the state of the transformer corresponding to the time domain characteristics through the acoustic signals and inputting the data into neural network training;
and obtaining a neural network model through multiple iterative training, wherein the neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
Optionally, the preprocessing module 32 is further configured to perform normalization processing on the acquired acoustic signal, where the normalization processing adopts a formula:
wherein x isiThe sound pressure value at a certain moment in a certain section of noise signal is in a form of a numerical value, and the unit is Pa; x is the number ofminRepresenting a minimum sound pressure value in a segment of the signal, in the form of a numerical value, in Pa; x is the number ofmaxRepresenting a minimum sound pressure value in a segment of the signal, in the form of a numerical value, in Pa; x is the number ofi' denotes a value obtained by normalizing the acoustic signal by a calculation, in the form of a numerical value, in Pa.
Optionally, the preprocessing module 32 is specifically configured to perform low-pass filtering on the acoustic signal when the acquisition transformer is in each state by using a butterworth low-pass digital filter;
and performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
Optionally, the output module 34 is specifically configured to subtract the actual running time of the predicted running time to obtain a running time difference;
matching the running time difference with a preset running threshold value;
and outputting the aging degree of the transformer according to the matching result.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for transformer aging evaluation based on an error back propagation algorithm is characterized in that the method comprises the following steps:
acquiring an acoustic signal of a transformer to be tested;
preprocessing the acquired acoustic signal to obtain characteristic information of the acoustic signal, wherein the preprocessing comprises: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing;
inputting the characteristic information of the acoustic signal into a preset neural network model, and calculating to obtain the predicted operation time of the transformer to be tested;
and comparing the predicted running time with the actual running time, and outputting the aging degree of the transformer.
2. The method for transformer aging evaluation based on error back propagation algorithm according to claim 1, wherein the step of obtaining the acoustic signal of the transformer under test is preceded by:
determining the number of neurons of an input layer and an output layer of a neural network prediction model, determining the number of hidden layers and the number of neuron nodes included in each hidden layer, initializing weights and training a neural network;
acquiring multi-channel acoustic signals of a plurality of groups of transformers with the same voltage grade in the same time period, and acquiring time domain characteristics and inputting the running time of the transformers corresponding to the time domain characteristics into the neural network training through the acoustic signals;
and obtaining a neural network model through multiple iterative training, wherein the trained neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
3. The method for transformer aging evaluation based on error back propagation algorithm according to claim 1, wherein the step of preprocessing the acquired acoustic signals further comprises: normalizing the acquired acoustic signals, wherein the normalization adopts a formula as follows:
wherein x isiThe sound pressure value at the ith moment in a certain section of noise signal is in a form of a numerical value and has a unit of Pa; x is the number ofminRepresents the minimum sound pressure value in the section of signal, and the minimum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofmaxRepresents the maximum sound pressure value in the section of signal, and the maximum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofi' denotes a sound pressure value at the i-th time obtained by normalizing the sound signal, in the form of a numerical value in Pa.
4. The method for transformer aging evaluation based on the error back propagation algorithm according to claim 1, wherein the step of preprocessing the acquired acoustic signal specifically comprises:
performing low-pass filtering on the acquired acoustic signals when the transformer is in each state by using a Butterworth low-pass digital filter;
and performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
5. The method for transformer aging evaluation based on error back propagation algorithm according to claim 1, wherein the step of outputting the aging degree of the transformer according to the comparison between the predicted operation time and the actual operation time specifically comprises:
the predicted running time is differed from the actual running time to obtain a running time difference value;
matching the running time difference value with a preset running threshold value;
and outputting the aging degree of the transformer according to the matching result.
6. A system for transformer aging evaluation based on an error back propagation algorithm, the system comprising: the device comprises an acquisition module, a preprocessing module, a calculation module and an output module;
the acquisition module is used for acquiring an acoustic signal of the transformer to be tested;
the preprocessing module is configured to preprocess the acquired acoustic signal to obtain characteristic information of the acoustic signal, where the preprocessing includes: at least one of filtering, signal denoising, feature extraction and data dimension reduction processing;
the calculation module is used for inputting the characteristic information of the acoustic signal into a preset neural network model and calculating to obtain the predicted operation time of the transformer to be tested;
and the output module is used for comparing the predicted running time with the actual running time and outputting the aging degree of the transformer.
7. The system for transformer aging evaluation based on error back propagation algorithm according to claim 6, further comprising a modeling module for determining the number of neurons in the input and output layers, determining the number of hidden layers and the number of neuron nodes each hidden layer comprises, weight initialization and neural network training;
acquiring acoustic signals of a plurality of groups of transformers with the same voltage grade in the same time period, and acquiring time domain characteristics and the running time of the transformers corresponding to the time domain characteristics through the acoustic signals and inputting the time domain characteristics into the neural network training;
and obtaining a neural network model through multiple iterative training, wherein the neural network model comprises the corresponding relation between the acoustic signal and the predicted operation time of the transformer to be tested.
8. The system for transformer aging evaluation based on error back propagation algorithm according to claim 6, wherein the preprocessing module is further configured to perform normalization processing on the acquired acoustic signal, wherein the normalization processing is performed by using the formula:
wherein x isiThe sound pressure value at the ith moment in a certain section of noise signal is in a form of a numerical value and has a unit of Pa; x is the number ofminRepresents the minimum sound pressure value in the section of signal, and the minimum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofmaxRepresents the maximum sound pressure value in the section of signal, and the maximum sound pressure value is in the form of a numerical value and has the unit of Pa; x is the number ofi' denotes a sound pressure value at the i-th time obtained by normalizing the sound signal, in the form of a numerical value in Pa.
9. The system for transformer aging evaluation based on error back propagation algorithm according to claim 6, wherein the preprocessing module is specifically configured to perform low-pass filtering on the acoustic signals acquired when the transformer is in each state by using a Butterworth low-pass digital filter;
and performing wavelet decomposition on the acoustic signal of the transformer, performing threshold processing on the wavelet coefficient after the wavelet decomposition, and reconstructing by using the wavelet coefficient after the threshold processing to obtain the sound signal of the power transformer after the signal noise elimination.
10. The system for transformer aging evaluation based on error back propagation algorithm according to claim 6, wherein the output module is specifically configured to subtract the predicted running time from the actual running time to obtain a running time difference value;
matching the running time difference value with a preset running threshold value;
and outputting the aging degree of the transformer according to the matching result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010977524.1A CN112114215A (en) | 2020-09-17 | 2020-09-17 | Transformer aging evaluation method and system based on error back propagation algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010977524.1A CN112114215A (en) | 2020-09-17 | 2020-09-17 | Transformer aging evaluation method and system based on error back propagation algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112114215A true CN112114215A (en) | 2020-12-22 |
Family
ID=73803408
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010977524.1A Pending CN112114215A (en) | 2020-09-17 | 2020-09-17 | Transformer aging evaluation method and system based on error back propagation algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112114215A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114252731A (en) * | 2021-12-13 | 2022-03-29 | 广西电网有限责任公司桂林供电局 | Relay action characteristic evaluation method and device based on multiple parameters |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002175107A (en) * | 2000-12-05 | 2002-06-21 | Sharp Corp | Process controlling method, its device, and program storage medium therefor |
CN106921158A (en) * | 2017-02-09 | 2017-07-04 | 国网福建省电力有限公司 | Coefficient Analysis method the need for a kind of history gathered data based on distribution transformer time series |
CN109740523A (en) * | 2018-12-29 | 2019-05-10 | 国网陕西省电力公司电力科学研究院 | A kind of method for diagnosing fault of power transformer based on acoustic feature and neural network |
CN110807550A (en) * | 2019-10-30 | 2020-02-18 | 国网上海市电力公司 | Distribution transformer overload identification early warning method based on neural network and terminal equipment |
CN111638028A (en) * | 2020-05-20 | 2020-09-08 | 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 |
-
2020
- 2020-09-17 CN CN202010977524.1A patent/CN112114215A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002175107A (en) * | 2000-12-05 | 2002-06-21 | Sharp Corp | Process controlling method, its device, and program storage medium therefor |
CN106921158A (en) * | 2017-02-09 | 2017-07-04 | 国网福建省电力有限公司 | Coefficient Analysis method the need for a kind of history gathered data based on distribution transformer time series |
CN109740523A (en) * | 2018-12-29 | 2019-05-10 | 国网陕西省电力公司电力科学研究院 | A kind of method for diagnosing fault of power transformer based on acoustic feature and neural network |
CN110807550A (en) * | 2019-10-30 | 2020-02-18 | 国网上海市电力公司 | Distribution transformer overload identification early warning method based on neural network and terminal equipment |
CN111638028A (en) * | 2020-05-20 | 2020-09-08 | 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 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114252731A (en) * | 2021-12-13 | 2022-03-29 | 广西电网有限责任公司桂林供电局 | Relay action characteristic evaluation method and device based on multiple parameters |
CN114252731B (en) * | 2021-12-13 | 2024-04-05 | 广西电网有限责任公司桂林供电局 | Relay action characteristic evaluation method and device based on multiple parameters |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111325095B (en) | Intelligent detection method and system for equipment health state based on acoustic wave signals | |
CN111238814B (en) | Rolling bearing fault diagnosis method based on short-time Hilbert transform | |
CN110398647B (en) | Transformer state monitoring method | |
CN112599134A (en) | Transformer sound event detection method based on voiceprint recognition | |
CN111912519B (en) | Transformer fault diagnosis method and device based on voiceprint frequency spectrum separation | |
CN113740635A (en) | Electrical equipment fault diagnosis method, terminal and multi-probe sensing device | |
CN112182490B (en) | Reactor state diagnosis method and system | |
CN112052712B (en) | Power equipment state monitoring and fault identification method and system | |
CN117909668B (en) | Bearing fault diagnosis method and system based on convolutional neural network | |
CN114372344B (en) | Quantitative identification method for subsynchronous oscillation damping characteristic influence factors | |
CN115901265A (en) | Rolling bearing fault diagnosis method based on MFCC-FcaNet | |
CN112114215A (en) | Transformer aging evaluation method and system based on error back propagation algorithm | |
CN114157023A (en) | Distribution transformer early warning information acquisition method | |
CN117686774A (en) | Broadband voltage signal monitoring method, system, equipment and storage medium | |
CN112098066A (en) | High-voltage shunt reactor fault diagnosis method and system based on gate control circulation unit | |
CN116482526A (en) | Analysis system for non-fault phase impedance relay | |
CN115754821A (en) | Method and device for diagnosing mechanical instability of converter transformer based on vibration voiceprint correlation | |
CN115902557A (en) | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium | |
CN113283157A (en) | System, method, terminal and medium for predicting life cycle of intelligent stamping press part | |
CN112560674A (en) | Method and system for detecting quality of sound signal | |
CN113933658B (en) | Dry-type transformer discharge detection method and system based on audible sound analysis | |
CN116125347B (en) | Method and system for detecting windings of oil immersed transformer based on optical fiber sensor | |
CN117783792B (en) | Valve side sleeve insulation state detection method and system based on multiparameter real-time monitoring | |
CN114878674B (en) | Transformer winding defect diagnosis method based on comprehensive characteristics of winding stress and magnetic leakage parameter of fusion algorithm | |
CN117894317B (en) | Box-type transformer on-line monitoring method and system based on voiceprint analysis |
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 |