CN110174610B - Method for obtaining electric service life of alternating current contactor based on convolutional neural network - Google Patents
Method for obtaining electric service life of alternating current contactor based on convolutional neural network Download PDFInfo
- Publication number
- CN110174610B CN110174610B CN201910414443.8A CN201910414443A CN110174610B CN 110174610 B CN110174610 B CN 110174610B CN 201910414443 A CN201910414443 A CN 201910414443A CN 110174610 B CN110174610 B CN 110174610B
- Authority
- CN
- China
- Prior art keywords
- data
- alternating current
- arc
- current contactor
- 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.)
- Active
Links
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
- G01R31/327—Testing of circuit interrupters, switches or circuit-breakers
- G01R31/3277—Testing of circuit interrupters, switches or circuit-breakers of low voltage devices, e.g. domestic or industrial devices, such as motor protections, relays, rotation switches
- G01R31/3278—Testing of circuit interrupters, switches or circuit-breakers of low voltage devices, e.g. domestic or industrial devices, such as motor protections, relays, rotation switches of relays, solenoids or reed switches
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method for obtaining the electric service life of an alternating current contactor based on a convolutional neural network, which comprises the steps of obtaining the breaking arc experiment data of the alternating current contactor and the contact quality data before and after each breaking arc experiment; calculating mass loss data of the contact of each disjunction arc experiment, processing to obtain disjunction arc discrete samples, and further randomly dividing the mass loss data of the contact and the disjunction arc discrete samples into a training set and a testing set according to a certain proportion; constructing an alternating current contactor electric life prediction model based on convolution neural network regression; training and testing the prediction model, and comparing to obtain a trained prediction model; and acquiring current breaking arc data of the alternating current contactor, importing the current breaking arc data into a trained prediction model, and outputting the electric service life of the alternating current contactor. By implementing the invention, the accurate prediction of the electric service life of the alternating current contactor can be realized without the data of the breaking operation before the existing on-off operation method, and the reliability and the resource utilization rate are improved.
Description
Technical Field
The invention relates to the technical field of alternating current contactor detection, in particular to a method for acquiring the electric service life of an alternating current contactor based on a convolutional neural network.
Background
The ac contactor in the low voltage apparatus has the characteristic of frequent on-off operation, and is widely and largely used in the electrical control system. Electromagnetic ac contactors are the subject of study, and such contactors typically have a service life that is limited and the ac contactor cannot continue to operate.
The mechanical life and the electrical life are two major categories of the service life of the alternating current contactor, the electrical life is far shorter than the mechanical life, and if the electrical life can be prolonged, great economic benefits can be brought to a power system. In order to research the electric service life of the alternating current contactor, the electric service life can be predicted by a certain means, and a means is provided for prolonging the electric service life.
At present, the method of counting the number of on-off operations is adopted to predict the electric service life of the alternating current contactor, and in order to ensure safety, the maximum number of operations is conservatively arranged in a smaller area.
Disclosure of Invention
The technical problem to be solved by the embodiments of the present invention is to provide a method for obtaining the electrical life of an ac contactor based on a convolutional neural network, which can accurately predict the electrical life of the ac contactor without breaking operation data before a switching-on/off operation method in the prior art, thereby improving reliability and resource utilization rate.
In order to solve the technical problem, an embodiment of the present invention provides a method for obtaining an electrical life of an ac contactor based on a convolutional neural network, where the method includes the following steps:
acquiring disjunction arc experiment data of the alternating current contactor and contact quality data before and after each disjunction arc experiment;
calculating contact quality loss data of each arc breaking experiment according to the obtained contact quality data of the alternating current contactor before and after each arc breaking experiment, processing the obtained arc breaking experiment data of the alternating current contactor into discrete arc breaking samples, and further randomly dividing the calculated contact quality loss data and the processed discrete arc breaking samples into a training set and a testing set according to a certain proportion;
constructing an alternating current contactor electric service life prediction model based on convolution neural network regression by taking a disjunction electric arc discrete point sample as a model characteristic and taking contact quality loss data as a model label;
respectively training and testing the AC contactor electric life prediction model according to the training set and the testing set, and obtaining a trained AC contactor electric life prediction model by comparing the total mass loss label of the contact in the training and testing of the AC contactor electric life prediction model with the mean square error value output by the convolutional neural network;
and acquiring current breaking arc data of the alternating current contactor, and importing the acquired current breaking arc data of the alternating current contactor into the obtained trained alternating current contactor electric service life prediction model, wherein the result output by the trained alternating current contactor electric service life prediction model is the alternating current contactor electric service life.
The contact mass loss data of each disjunction arc experiment is obtained by obtaining the contact mass loss before and after each disjunction arc experiment according to the contact mass data before and after each disjunction arc experiment, and further processing the contact mass loss before and after each disjunction arc experiment by adopting a linear interpolation method.
The method comprises the following specific steps of processing the acquired disjunction arc experiment data of the alternating current contactor into disjunction arc discrete samples:
determining sampling points in the acquired breaking arc experimental data of the alternating current contactor, amplifying the number of the sampling points by adopting a linear interpolation method, and further normalizing to limit the data range of the sampling points subjected to linear interpolation within [0,1 ];
extracting one-dimensional discretization data through preset data waveform software according to the sampling points subjected to linear interpolation and the corresponding data ranges thereof, and combining the extracted one-dimensional discretization data into a one-dimensional matrix; and automatically recombining the combined one-dimensional matrix into a two-dimensional matrix in a convolutional neural network in the AC contactor electric service life prediction model.
The specific steps of constructing the alternating current contactor electric service life prediction model based on convolution neural network regression by using the disjunction electric arc discrete point sample as a model characteristic and using the contact quality loss data as a model label include:
establishing a convolutional neural network by adopting a deep learning framework TensorFlow, establishing an electric service life prediction model of the alternating current contactor, respectively reading in the prediction model by taking a disjunction arc discrete point sample as a model characteristic and contact quality loss data as a model label, and defining a batch function module, a data reading module, a CNN structure module, a model evaluation index module and a training and testing module; wherein the convolutional neural network has two convolutional pooling alternating layers and two fully connected layers.
The embodiment of the invention has the following beneficial effects:
the method takes the current and voltage waveform of the disjunction electric arc as an input characteristic, predicts the mass loss of the contact through the convolution neural network model, overcomes the huge influence of the change of the arcing phase angle on the electric arc waveform, can predict the electric service life in real time aiming at each disjunction operation, does not need the data of the previous disjunction operation, greatly simplifies the complexity of the experimental process, and improves the research efficiency.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is within the scope of the present invention for a person skilled in the art to obtain other drawings based on the drawings without paying creative efforts.
Fig. 1 is a flowchart of a method for obtaining an electrical life of an ac contactor based on a convolutional neural network according to an embodiment of the present invention;
fig. 2 is a graph of a change trend of total mass loss of a contact in an application scenario of a method for obtaining an electrical life of an ac contactor based on a convolutional neural network according to an embodiment of the present invention;
fig. 3 is a data waveform analysis software interface in an application scenario of a method for acquiring an electrical life of an ac contactor based on a convolutional neural network according to an embodiment of the present invention;
FIG. 4 is an arc waveform diagram of one cycle in an application scenario of a method for obtaining an electrical life of an AC contactor based on a convolutional neural network according to an embodiment of the present invention;
fig. 5 is a linear interpolation arc waveform diagram in an application scenario of the method for acquiring the electrical life of the ac contactor based on the convolutional neural network according to the embodiment of the present invention;
FIG. 6 is a normalized arc waveform diagram in an application scenario of a method for obtaining an electrical life of an AC contactor based on a convolutional neural network according to an embodiment of the present invention;
fig. 7 is a graph of a change trend of contact quality loss after linear interpolation in an application scenario of a method for obtaining an electrical life of an ac contactor based on a convolutional neural network according to an embodiment of the present invention;
fig. 8 is a parameter diagram of an ac contactor electrical life prediction model in an application scenario of a method for obtaining an ac contactor electrical life based on a convolutional neural network according to an embodiment of the present invention;
fig. 9 is a diagram of an ac contactor electrical life prediction model established in an application scenario of a method for acquiring an ac contactor electrical life based on a convolutional neural network according to an embodiment of the present invention;
fig. 10 is a result of execution of an ac contactor electrical life prediction model in an application scenario of the method for acquiring the ac contactor electrical life based on the convolutional neural network according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.
As shown in fig. 1, in an embodiment of the present invention, a method for obtaining an electrical lifetime of an ac contactor based on a convolutional neural network is provided, where the method includes the following steps:
s1, acquiring disjuncted arc experiment data of the alternating current contactor and contact quality data before and after each disjuncted arc experiment;
specifically, under the AC-4 experiment condition, the electric life experiment system is used for collecting the disjointed arc data to obtain 6000 groups of disjointed arc experiment data, the mass of all contacts is weighed every 600 times of experiments, and the total weight is 11 times of experiments, including the mass data before the beginning of one experiment.
Step S2, calculating the contact mass loss data of each arc breaking experiment according to the obtained contact mass data of the alternating current contactor before and after each arc breaking experiment, processing the obtained arc breaking experiment data of the alternating current contactor into discrete arc breaking samples, and further randomly dividing the calculated contact mass loss data and the processed discrete arc breaking samples into training sets and testing sets according to a certain proportion;
specifically, the arc data and the contact quality data are preprocessed. Firstly, the contact mass loss data of each disjunction arc experiment is obtained by obtaining the contact mass loss before and after each disjunction arc experiment according to the contact mass data before and after each disjunction arc experiment, and further processing the contact mass loss before and after each disjunction arc experiment by adopting a linear interpolation method; secondly, determining sampling points in the acquired breaking arc experimental data of the alternating current contactor, amplifying the number of the sampling points by adopting a linear interpolation method, and further normalizing to limit the data range of the sampling points subjected to linear interpolation within [0,1 ]; extracting one-dimensional discretization data through preset data waveform software according to the sampling points subjected to linear interpolation and the corresponding data ranges thereof, and combining the extracted one-dimensional discretization data into a one-dimensional matrix; wherein the one-dimensional matrix is automatically reconstructed into a two-dimensional matrix in the convolutional neural network in the ac contactor electrical life prediction model in step S3.
S3, constructing an alternating current contactor electric service life prediction model based on convolution neural network regression by taking a disjunction arc discrete point sample as a model characteristic and taking contact quality loss data as a model label;
specifically, a deep learning framework TensorFlow is adopted to realize a convolutional neural network, an alternating current contactor electrical service life prediction model is established, a disjointed arc discrete point sample is taken as a model characteristic, contact quality loss data is taken as a model label and is respectively read into the prediction model, and a batch function module, a data reading module, a convolutional neural network CNN structure module, a model evaluation index module and a training and testing module are defined; the convolution neural network comprises two convolution pooling alternate layers and two full-connection layers.
In one example, a convolution kernel with the size of 3 × 3 is adopted in the ac contactor electrical life prediction model, the size of the pooling window is 2 × 2, the nonlinear activation function ReLU is used after each convolution operation, the number of channels of the convolutional neural network CNN model is changed from 6 to 32, from 32 to 64, from 64 to 1024, and the final output is only 1 channel. And (3) elongating the characteristic matrix output by the pooling layer 2 into a one-dimensional matrix, inputting the one-dimensional matrix into the fully-connected layer 1, and using a Dropout method at the output end of the fully-connected layer 1, wherein part of neurons in the network model are randomly discarded according to probability in a training stage.
The specific structure of the convolutional neural network is as follows:
input layer
< ═ 1 convolutional layer 1_1(3x3x64)
2 nonlinear response Relu layer
< ═ 3 convolutional layer 1_2(3x3x64)
4 nonlinear response Relu layer
< ═ 5 pooling layer (2x2/2)
2_1(3x3x128) convolutional layer 6 [ ]
< 7 nonlinear response Relu layer
2_2(3x3x128) convolutional layer 8 [ ]
9 nonlinear response Relu layer
< ═ 10 pooling layer (2x2/2)
< ═ 11 convolutional layer 3_1(3x3x256)
-12 nonlinear response Relu layer
< ═ 13 convolutional layer 3_2(3x3x256)
14 global average pooling layer
15 full connection layer (256x100) s
-16 nonlinear response Relu layer
< ═ 17 full connection layer (100x2)
< ═ 14 deconvolution layer D1(4x4x256)
< ═ 19 convolutional layer D1_1(3x3x256)
< ═ 20 convolutional layer D1_2(3x3x256)
< ═ 21 deconvolution layer D2(4x4x128)
<22convolutional layer D2_1(3x3x128)
< ═ 23 convolutional layer D2_2(3x3x128)
< ═ 24 convolutional layer D2_3(3x3x2)
Wherein, the number in front of the symbol is the current layer number, and the number behind the symbol is the input layer number; the inside of brackets behind the convolutional layer and the deconvolution layer are convolutional layer parameters, wherein the product of two multipliers in front of the convolutional layer parameters is the size of a convolutional kernel, and the multiplier behind the convolutional layer parameters is the number of channels; the bracketing layer parameter is arranged in brackets behind the pooling layer, wherein the product of two multipliers in front of the pooling layer parameter is the size of a pooling kernel, and the multiplier behind the pooling layer parameter is the step length; the parameters of the full connection layer are arranged in brackets behind the full connection layer, wherein the parameters behind the full connection layer are output categories, and whether the baby kicks the quilt or not is detected, so that the classification is a second classification; the nonlinear response layer is composed of a nonlinear activation function ReLU.
Step S4, according to the training set and the test set, respectively training and testing the AC contactor electrical life prediction model, and respectively obtaining a trained AC contactor electrical life prediction model by comparing the total mass loss label of the contact in the training and testing of the AC contactor electrical life prediction model with the mean square error value output by the convolutional neural network;
specifically, the AC contactor electrical life prediction model is trained by adopting samples of a training set, the AC contactor electrical life prediction model is tested by adopting samples of a testing set, and the error value of contact quality loss is evaluated by solving the mean square error value (MSE) output by an actual label and a CNN model, namely, the error value is trained by adopting a back propagation algorithm of the error until the convolutional neural network is converged, so that the trained AC contactor electrical life prediction model is obtained.
And step S5, acquiring current disjunction arc data of the alternating current contactor, and importing the acquired current disjunction arc data of the alternating current contactor into the obtained trained alternating current contactor electric life prediction model, wherein the result output by the trained alternating current contactor electric life prediction model is the alternating current contactor electric life.
Specifically, the actual to-be-detected broken arc data of the alternating current contactor is obtained and sent to a trained alternating current contactor electric life prediction model, and the result output by the trained alternating current contactor electric life prediction model at the moment is the alternating current contactor electric life.
As shown in fig. 2 to fig. 10, an application scenario of the method for obtaining the electrical life of the ac contactor based on the convolutional neural network in the embodiment of the present invention is further described:
the method collects 6000 groups of breaking arc experimental data and contact mass loss data under the AC-4 experimental condition. The experimental data is collected by an electric life control system (not shown in the figure), the system comprises a console system, a data collection system and a load system, wherein the console system controls the starting and stopping operations of the alternating current contactor, the load system is used for adjusting resistive and inductive loads and ensuring that the experimental current value is 6 times of the rated current value, and the data collection system is used for storing the collected disjunctive electric arc data.
Before the life test is started, the four ports of the test system are short-circuited, the actual current passed by the test is 6 times of the rated current 80A due to the fact that the power supply voltage is constant, namely 480A, the power factor is 0.35, the phase voltage when the load system is electrified is 386V by calculating and adjusting the impedance value and the inductive reactance value required by the test, and then the impedance value R and the inductive reactance value L of the load system are calculated. The calculation of R and L is as follows:
the following can be obtained:
and then, finely adjusting R and L to ensure that the actual current of each phase on the alternating current contactor is more than 480A and the error is within +5 percent. After the alternating current contactor is fixed, experiments are carried out according to a preset operation flow, and data acquired each time can be automatically stored under a specified folder.
The mass loss data of the contacts is acquired while the experimental data is acquired, wherein after each 600 times of experiments, the experiment is stopped, and the electronic scale with thousands of positions is used for weighing the masses of 6 static contacts and 3 moving contacts, so that 11 groups of masses need to be weighed, wherein the first group is the mass of the contacts weighed before the experiment, and the mass loss is the difference between the mass data before the experiment and the mass data after the experiment because the contact is more worn and more serious because the experiment times are more. In order to more intuitively show the variation trend of the mass loss, the form shown in fig. 2 is shown, and the points where the different symbols such as "+", "o" and the like are located in the figure are the mass loss data of the contact.
After data acquisition is finished, broken arc data are extracted by using waveform analysis software shown in fig. 3 and stored in a mat format, after original data are written into the right side of a software interface, voltage (blue line) and current (red line) waveforms are led into the left side of the interface, the waveforms are amplified to the positions where broken arcs are located, characteristic data in the middle of the interface are calculated, and finally the characteristic data are stored.
Fig. 4 is a waveform diagram of an interrupted arc in one period, each waveform period having 600 sampling points. In order to adapt to the processing mode of the CNN model, the number of sampling points needs to be expanded to 2n×2nThe number of sampling points is increased to 1024 (2) by adopting a linear interpolation method5×25) And normalizing the data range of the sampling point after linear interpolation to be limited to [0, 1%]And (4) the following steps.
Meanwhile, a 1024 × 1 one-dimensional matrix is automatically reconstructed into a 32 × 32 two-dimensional matrix when the model for predicting the electrical life of the ac contactor is input. The calculation formula for normalization and recombination is as follows:
where x (i) · 1., m2) represents the values of the arc sampling point waveforms, y (j, k) (j ═ 1.., m; k · 1., m) represents the regrouped two-dimensional matrix, and max (x) and min (x) refer to the maximum and minimum values of the sampling point data, respectively. Fig. 5 is a linear interpolated arc waveform and fig. 6 is a normalized arc waveform.
For the mass loss data, a linear interpolation method is also adopted, the number of the data of the total mass loss is increased to 6000, and the variation trend of the mass loss of the contact after interpolation is shown in figure 7. And randomly dividing the broken arc sample and the mass loss data of the contact into a training set and a testing set according to a proportion.
Deep learning modeling is realized on the Ubuntu system through a deep learning framework Tensorflow and a Convolutional Neural Network (CNN) model, wherein the graph comprises two convolution pooling alternate layers and two full-connection layers, and all parameters of the model layers are shown in figure 8.
The life prediction models all adopt convolution kernels with the size of 3 multiplied by 3, the size of a pooling window is 2 multiplied by 2, a nonlinear activation function Relu is used after each convolution operation, the number of channels of a CNN model is changed from 6 to 32, from 32 to 64 and from 64 to 1024 from input to output, and only 1 channel is finally output. Lengthening the characteristic matrix output by the pooling layer 2 into a one-dimensional matrix, inputting the one-dimensional matrix into the full connection layer 1, and using a Dropout method at the output end of the full connection layer 1, wherein the method randomly discards part of neurons in a network model according to probability in a training stage, and needs to keep all the neurons in a testing stage.
Wherein f (x)t) Watch (A)True value, here loss of quality tag data, ytRepresenting the estimated values, here the output of the CNN model, the resulting error values are then fed to an Adam optimizer with adaptive learning capabilities, and the global minimum is calculated, and the overall life prediction model is shown in fig. 9.
The experiment is carried out by adopting data with a ratio of 5:1, namely, the number of training samples is 5000, the number of test samples is 1000, the Dropout ratio in the training and testing stages is 0.5 and 1.0 respectively, the number of iterations is 106, the batch size is 256, a mean square error value (MSE) is output every 250 iterations, the training result and the testing result are represented in a form of a graph 10, the horizontal axis in the graph is the iteration number, the vertical axis is the MSE value, black and red curves respectively represent the training and testing errors, and the MSE values of the training and testing obtained after the training and testing stages are executed are marked at the lower right corner of the graph.
The processed breaking arc experiment data is imported into the trained alternating current contactor electric service life prediction model, and then the prediction model is executed to obtain a prediction result, so that the complexity of the experiment process is greatly simplified, and the research efficiency is improved.
The embodiment of the invention has the following beneficial effects:
the method takes the current and voltage waveform of the disjunction electric arc as an input characteristic, predicts the mass loss of the contact through the convolution neural network model, overcomes the huge influence of the change of the arcing phase angle on the electric arc waveform, can predict the electric service life in real time aiming at each disjunction operation, does not need the data of the previous disjunction operation, greatly simplifies the complexity of the experimental process, and improves the research efficiency.
It will be understood by those skilled in the art that all or part of the steps in the method for implementing the above embodiments may be implemented by instructing the relevant hardware through a program, and the program may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention, and it is therefore to be understood that the invention is not limited by the scope of the appended claims.
Claims (3)
1. A method for obtaining ac contactor electrical life based on convolutional neural network, the method comprising the steps of:
acquiring disjunction arc experiment data of the alternating current contactor and contact quality data before and after each disjunction arc experiment;
calculating contact quality loss data of each arc breaking experiment according to the obtained contact quality data of the alternating current contactor before and after each arc breaking experiment, processing the obtained arc breaking experiment data of the alternating current contactor into discrete arc breaking samples, and further randomly dividing the calculated contact quality loss data and the processed discrete arc breaking samples into a training set and a testing set according to a certain proportion;
constructing an alternating current contactor electric service life prediction model based on convolution neural network regression by taking a disjointed arc discrete sample as a model characteristic and taking contact quality loss data as a model label;
respectively training and testing the electric life prediction model of the alternating current contactor according to the training set and the testing set, and obtaining the trained electric life prediction model of the alternating current contactor by comparing contact quality loss data of the electric life prediction model of the alternating current contactor in the training and the testing with a mean square error value output by the convolutional neural network;
acquiring current disjunction arc data of the alternating current contactor, and importing the acquired current disjunction arc data of the alternating current contactor into an obtained trained alternating current contactor electric service life prediction model, wherein the result output by the trained alternating current contactor electric service life prediction model is the alternating current contactor electric service life;
the specific steps of processing the acquired disjunction arc experiment data of the alternating current contactor into disjunction arc discrete samples comprise:
determining sampling points in the acquired breaking arc experimental data of the alternating current contactor, amplifying the number of the sampling points by adopting a linear interpolation method, and further normalizing to limit the data range of the sampling points subjected to linear interpolation within [0,1 ];
extracting one-dimensional discretization data through preset data waveform software according to the sampling points subjected to linear interpolation and the corresponding data ranges thereof, and combining the extracted one-dimensional discretization data into a one-dimensional matrix; and automatically recombining the combined one-dimensional matrix into a two-dimensional matrix in a convolutional neural network in the AC contactor electric service life prediction model.
2. The method for obtaining the electric life of the alternating current contactor based on the convolutional neural network as claimed in claim 1, wherein the contact mass loss data of each arc breaking experiment is obtained by obtaining the contact mass loss before and after each arc breaking experiment according to the contact mass data before and after each arc breaking experiment, and further processing the contact mass loss before and after each arc breaking experiment by using a linear interpolation method.
3. The method for obtaining the electric life of the alternating current contactor based on the convolutional neural network as claimed in claim 1, wherein the specific step of constructing the alternating current contactor electric life prediction model based on the convolutional neural network regression by using the discrete broken arc sample as the model feature and the contact quality loss data as the model label comprises:
adopting a deep learning framework TensorFlow to realize convolution neural network regression, establishing an alternating current contactor electrical service life prediction model, taking a disjunction arc discrete sample as a model characteristic, taking contact quality loss data as a model label, respectively reading into the prediction model, and defining a batch function module, a data reading module, a CNN structure module, a model evaluation index module and a training and testing module; wherein the convolutional neural network has two convolutional pooling alternating layers and two fully connected layers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414443.8A CN110174610B (en) | 2019-05-17 | 2019-05-17 | Method for obtaining electric service life of alternating current contactor based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910414443.8A CN110174610B (en) | 2019-05-17 | 2019-05-17 | Method for obtaining electric service life of alternating current contactor based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110174610A CN110174610A (en) | 2019-08-27 |
| CN110174610B true CN110174610B (en) | 2021-10-22 |
Family
ID=67691497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910414443.8A Active CN110174610B (en) | 2019-05-17 | 2019-05-17 | Method for obtaining electric service life of alternating current contactor based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110174610B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111505490A (en) * | 2020-03-23 | 2020-08-07 | 温州大学乐清工业研究院 | AC contactor ablation condition evaluation method based on convolutional neural network regression |
| CN112098833B (en) * | 2020-09-18 | 2023-10-13 | 株洲国创轨道科技有限公司 | Relay service life prediction method, system, medium and equipment |
| CN112231910A (en) * | 2020-10-15 | 2021-01-15 | 华人运通(江苏)技术有限公司 | System and method for determining service life of high-voltage direct-current contactor |
| CN112464556A (en) * | 2020-11-17 | 2021-03-09 | 沈阳工业大学 | AC contactor electric service life prediction method based on long-short term memory neural network |
| CN113466681B (en) * | 2021-05-31 | 2024-05-10 | 国网浙江省电力有限公司营销服务中心 | Breaker service life prediction method based on small sample learning |
| CN113255579B (en) * | 2021-06-18 | 2021-09-24 | 上海建工集团股份有限公司 | Method for automatically identifying and processing construction monitoring abnormal acquisition data |
| CN113420063A (en) * | 2021-06-18 | 2021-09-21 | 上海建工集团股份有限公司 | System for construction monitoring abnormal acquisition data automatic identification and processing |
| CN114137403B (en) * | 2021-11-22 | 2023-04-07 | 重庆大学 | On-load tap-changer electrical life evaluation system and method based on radiation electromagnetic waves |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106291339A (en) * | 2015-05-20 | 2017-01-04 | 国网上海市电力公司 | A kind of circuit breaker failure diagnostic expert system based on artificial neural network |
| CN106291351B (en) * | 2016-09-20 | 2019-01-18 | 西安工程大学 | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm |
| US10332698B2 (en) * | 2016-12-21 | 2019-06-25 | Eaton Intelligent Power Limited | System and method for monitoring contact life of a circuit interrupter |
| CN108734332B (en) * | 2018-03-29 | 2021-11-26 | 浙江长兴笛卡尔科技有限公司 | Method and device for predicting electrical parameters by machine learning |
-
2019
- 2019-05-17 CN CN201910414443.8A patent/CN110174610B/en active Active
Non-Patent Citations (1)
| Title |
|---|
| A Novel Image Feature for the Remaining Useful Lifetime Prediction of Bearings Based on Continuous Wavelet Transform and Convolutional Neural Network;Youngji Yoo等;《Applied Sciences》;20180708;第8卷(第7期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110174610A (en) | 2019-08-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110174610B (en) | Method for obtaining electric service life of alternating current contactor based on convolutional neural network | |
| CN106909933B (en) | A kind of stealing classification Forecasting Methodology of three stages various visual angles Fusion Features | |
| CN111833583B (en) | Training method, device, equipment and medium for power data anomaly detection model | |
| CN109033513B (en) | Power transformer fault diagnosis method and power transformer fault diagnosis device | |
| CN112114986A (en) | Data anomaly identification method and device, server and storage medium | |
| CN111523683B (en) | Method and system for predicting technological parameters in tobacco processing | |
| JP2020042519A (en) | Abnormality detection device, abnormality detection method, and abnormality detection program | |
| US20200380409A1 (en) | Apparatus and method for analyzing time-series data based on machine learning | |
| CN117375237B (en) | Substation operation and maintenance method and system based on digital twin technology | |
| EP3745317A1 (en) | Apparatus and method for analyzing time series data based on machine learning | |
| CN116739829B (en) | Big data-based power data analysis method, system and medium | |
| CN117156442A (en) | Cloud data security protection method and system based on 5G network | |
| CN111738348A (en) | Power data anomaly detection method and device | |
| CN113543117A (en) | Prediction method and device for number portability user and computing equipment | |
| CN114004162A (en) | Modeling method for smelting load harmonic emission level under multi-working-condition scene | |
| CN113945862A (en) | Method, device and equipment for identifying high-resistance grounding fault of power distribution network | |
| CN117589233A (en) | Intelligent detection method and system for running state of relay protection device | |
| CN112595918A (en) | Low-voltage meter reading fault detection method and device | |
| CN111898446A (en) | Single-phase earth fault studying and judging method based on multi-algorithm normalization analysis | |
| WO2014173271A1 (en) | Optimization method and system for the number of monitoring units of digital man-machine interface | |
| CN116298725A (en) | Fault arc detection method, system and storage medium | |
| CN116757870A (en) | Intelligent energy monitoring data processing method and system for energy Internet of things | |
| CN106817328B (en) | Digital predistortion table generation method and device | |
| CN114821309A (en) | Indoor transformer substation switch and indicator lamp state identification method and system | |
| CN106600034A (en) | Three-phase unbalance governance device having in-situ temperature alarm function and realization method thereof |
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 |