CN103294849B - Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural - Google Patents
Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural Download PDFInfo
- Publication number
- CN103294849B CN103294849B CN201310167662.3A CN201310167662A CN103294849B CN 103294849 B CN103294849 B CN 103294849B CN 201310167662 A CN201310167662 A CN 201310167662A CN 103294849 B CN103294849 B CN 103294849B
- Authority
- CN
- China
- Prior art keywords
- submodel
- fault diagnosis
- model
- output
- diagnosis model
- 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.)
- Expired - Fee Related
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/80—Technologies aiming to reduce greenhouse gasses emissions common to all road transportation technologies
- Y02T10/82—Elements for improving aerodynamics
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Tests Of Circuit Breakers, Generators, And Electric Motors (AREA)
Abstract
The invention discloses a kind of Fault Diagnosis of AC Electric Machine model construction method based on RBF neural, comprise the following steps: set up fault diagnosis model; Training fault diagnosis model; Test failure diagnostic model.Owing to present invention employs modular construction, form each module of this model, i.e. submodel, there is identical input number, the input end of each submodel trained only need be connected in parallel by the structure of fault diagnosis model, ensure that between each submodel separate.It is easy, flexible that this connected mode makes fault diagnosis model construct.Because fault diagnosis model training of the present invention only need be carried out submodel, therefore, greatly reducing the difficulty of fault diagnosis model training, providing condition for shortening the training time.Due to separate between each submodel of the present invention, thus through training after submodel more responsive to learnt specific fault conditions, which enhance the Fault Identification ability of fault diagnosis model.
Description
Technical field
The present invention relates to a kind of Fault Diagnosis of AC Electric Machine technology, particularly a kind of Fault Diagnosis of AC Electric Machine model construction method based on RBF neural.
Background technology
The fault type of induction motor is divided into mechanical fault and electric fault substantially, these faults can be divided into the polytypes such as bearing fault, stator failure, rotor fault and eccentric class fault again.Different faults has different fault signatures, and this is also the Main Basis carrying out fault diagnosis.But it should be noted that because induction motor is that collection is mechanical, electrical, the complex apparatus of magnetic one.Therefore, fault signature is also extremely complicated.At different operating conditions, fault of the same type, its feature also can change; Different faults has similar fault signature.These fault diagnosises being induction motor bring difficulty.
At present, most For Diagnosing Faults of Electrical model based on neural network, usually adopts multiple input/multiple output structure, namely builds with a Complex Neural Network model carrying out different fault diagnosis; The many employings of neural network three layers of BP (BackPropagation) neural network or three layers of radial basis (RBF, RadialBasisFunction) neural network.RBF neural and BP neural network are all Nonlinear Multi feedforward networks, and they all belong to general and approach device.Three-layer neural network is made up of input layer, hidden layer and output layer.
Transfer function due to RBF neural Hidden unit is about centrosymmetric radial basis function (as Gaussian function), the Hidden unit number of such three layers of static RBF feedforward neural network can according to the particular problem of research, adjust adaptively in the training stage, make the applicability of network better.Usually adopt nearest neighbor classifier mode training network to RBF neural, the distribution of such network Hidden unit is just only relevant with the distribution of training sample and the width of Hidden unit, has nothing to do with performing of task.On the basis that Hidden unit distributes, the mapping relations between constrained input, are realized by the weights between adjustment Hidden unit and output unit.
RBF neural compares with BP neural network, have that calculated amount is little, fast convergence rate, without features such as local minimums.Therefore, the present invention proposes the modularized motor fault diagnosis model based on RBF neural.All report is had with Publication about Document about the For Diagnosing Faults of Electrical Study on Problems based on neural network:
[1] yellow lead, Huang Cailun. based on the For Diagnosing Faults of Electrical expert system [J] of BP neural network model. instrument and meter for automation, 2003,24 (3): 15-17;
[2] Guo Xijin etc. based on asynchronous machine turn-to-turn short circuit degree research [J] of BP neural network. coal mine machinery, 2011,32 (10): 276-278;
[3] application of Zhang Jingzhai .RBF neural network in hydrogenerator fault diagnosis [J]. Computer Simulation, 2011,28 (12): 314-317;
[4] Wang Juan etc. based on the research [J] of the For Diagnosing Faults of Electrical of RBF neural. system simulation technology, 2009,5 (1): 36-39;
[5] application of .RBF neural network in asynchronous machine fault diagnosis [J] such as Mu Lijuan. system simulation technology, 2009,5 (1): 148-151;
[6] Bai Chengrong. the application [J] of RBF neural in For Diagnosing Faults of Electrical. Shanxi architecture, 201,37 (17): 213-214.
Learn preferably to enable neural network and distinguish various faults feature, this network should have good complex nonlinear characteristic mapping ability.Certainly will require that the hidden layer of this network should have more node like this.Although existing document indicates that three-layer neural network is to approach any nonlinear characteristic in theory, but the network larger to a hidden node number, obviously can cause the difficult problem of the aspects such as convergence, training and retraining.The fault diagnosis model be made up of a multiple-input and multiple-output network is had at least to the deficiency of following several respects:
1) fault type that will identify is more, and the network of component model is more complicated;
2) complicated network structure, then network convergence and the corresponding increase of training difficulty;
3) once the newly-increased fault type that will identify or have additional sample data to certain fault signature, just need to carry out network retraining, increase training cost.
Summary of the invention
For solving the problems referred to above that prior art exists, the present invention will design a kind of Fault Diagnosis of AC Electric Machine model construction method based on RBF neural that can realize following object:
1, have conveniently, flexibly, the modularization model structure that can conveniently combine;
2, the convergence of model and training difficulty is reduced;
3, model modification is convenient and swift.
To achieve these goals, technical scheme of the present invention is as follows: a kind of Fault Diagnosis of AC Electric Machine model construction method based on RBF neural, comprises the following steps:
A, set up fault diagnosis model
A1, determine the structure of fault diagnosis model
First determine to form the submodel number of For Diagnosing Faults of Electrical model according to the number intending diagnosis alternating current generator failure mode; Each submodel is the model of a multi input-mono-output, and namely each submodel has multiple input end and single output terminal, and the input end number of each submodel is identical; Be connected in parallel by the input end of each submodel and namely form For Diagnosing Faults of Electrical model, namely the output terminal number of all submodels forms the output terminal number of For Diagnosing Faults of Electrical model, that is the number of submodel is exactly the output terminal number of For Diagnosing Faults of Electrical model;
Described submodel is made up of a multi input-mono-output three-layer neural network; The input end number of the corresponding submodel of the input layer number of described neural network, output layer node number is 1, the output terminal number of corresponding submodel, and each submodel is used for representing a kind of malfunction of motor;
A2, determine the input end number of fault diagnosis model
If submodel has m input end, that is fault diagnosis model has m input end, if the input signal of m input end forms an input vector x, is expressed as follows:
x＝(x
1x
2…x
m)
In formula, x
ifor i-th input signal of submodel, i=1,2 ..., m;
Described input signal is the radial vibration acceleration signal of alternating-current motor stator current signal or alternating current arbor or the axial vibration acceleration signal of alternating current arbor; The electrical fault energy Ratios vector that described input vector obtains through WAVELET PACKET DECOMPOSITION for input signal; Described energy Ratios vector is referred to as proper vector, and each component of proper vector correspond to the signal energy of a certain frequency band and the gross energy ratio of Whole frequency band signal; The element number of proper vector is determined by decomposed signal band number;
The element number of proper vector is the input end number m of fault diagnosis model, namely the input end number m of each submodel;
A3, determine the output terminal number of fault diagnosis model
If fault diagnosis model is by n sub-model-composing, then the output signal of all submodels that its output vector y is comprised by it is formed, that is:
y＝(y
1y
2…y
n)
In formula, y
jfor a jth output signal of For Diagnosing Faults of Electrical model, the namely output signal of a jth submodel, j=1,2 ..., n, the output terminal number of each submodel is fixed as 1, represents by a kind of malfunction of diagnosing motor;
A4, determine the Hidden nodes of submodel
First use default value 2 or 4 as the Hidden nodes of submodel, finally adjust the Hidden nodes of submodel according to training result;
B, training fault diagnosis model
The output state of B1, definition submodel
For each submodel defines its fault type that will represent, any one submodel so after training, only have when inputting its corresponding fault type energy Ratios vector, the output signal of this submodel is 1, otherwise output signal is 0;
B2, preparation submodel training sample set
To each fault type, the energy Ratios vector adopting the fault data being not less than 10 groups to generate forms the training sample set of each submodel;
B3, training submodel
Respectively each submodel is trained with the training sample set of obtained each submodel;
B4, synthesis fault diagnosis model
After all submodel training terminate, the input end of each submodel after training is in parallel, just constitute the Fault Diagnosis of AC Electric Machine model that will set up;
C, test failure diagnostic model
C1, preparation fault diagnosis model test sample book collection
Test sample book collection is formed with the energy Ratios vector comprising fault type corresponding to all submodels being different from training sample set, according to each fault energy of test sample book collection putting in order than vector, a corresponding motor status exports table, and the ideal referred to herein as fault diagnosis model exports table; The corresponding test sample book of every a line in table concentrates the malfunction represented by corresponding line to export, and namely the ideal of fault diagnosis model exports;
The performance of fault diagnosis of C2, test failure diagnostic model
Fault energy successively in continuous input test sample set is than vector, the output vector of record cast, obtain the actual output table of model, the ideal output table of model and actual output table are contrasted, test and evaluation is carried out to the performance of fault diagnosis of fault diagnosis model; If the matching degree of the actual output table of fault diagnosis model and desirable output table is greater than 80%, namely think and meet the demands, then this fault diagnosis model just can come into operation.
Fault diagnosis model described in step C2 of the present invention is actual exports table and the desirable acquisition methods exporting the matching degree shown, and comprises the following steps:
Numerical value in the actual output table of the fault diagnosis model obtained in fault diagnosis model test process is the real number in one [0,1] interval; In the actual output table of fault diagnosis model, each value is got in close to ideal output table and is respectively worth, then represent that the performance of fault diagnosis of fault diagnosis model is better; In order to obtain, fault diagnosis model is actual to be exported and the desirable matching degree exported, first each numerical value in the actual output table of fault diagnosis model is carried out round process, obtain a corresponding new table formed by 0,1, be called the actual output of fault diagnosis model here and round table; When the actual output of fault diagnosis model rounds table and desirable output table corresponding row is respectively worth completely the same, claim corresponding row coupling, all mate with all row in desirable output table if actual output rounds table, then title reality exports to round and shows to mate with desirable output table 100%.
Compared with prior art, the present invention has following beneficial effect:
1, fault diagnosis model structure is easy, flexible.
The present invention proposes a kind of modularization Fault Diagnosis of AC Electric Machine model based on radial basis function (RBF) neural network.Because this For Diagnosing Faults of Electrical model have employed modular construction, form each module of this model, i.e. submodel, there is identical input number, the input end of each submodel trained only need be connected in parallel by the structure of fault diagnosis model, ensure that between each submodel separate.It is easy, flexible that this connected mode makes fault diagnosis model construct.Three layers of RBF neural of the corresponding multi input-mono-output of each submodel, are used for identifying a kind of motor status.This design ensure that the RBF neural structure forming submodel is relatively simple, for acceleration disturbance diagnostic model speed of convergence, shortening training time provide condition.In actual applications, required submodel number can be determined with condition according to specific needs.The input number of fault diagnosis model, is exactly the input number of submodel, is determined by practical situations, is exactly the number according to actual conditions determination decomposition frequency band specifically; Can be set as 2 or 4 under the hidden node number default condition of submodel, occurrence can according to the convergence situation increase and decrease in training process.
2, fault diagnosis model fast convergence rate, training time are short.
The present invention only need carry out submodel due to fault diagnosis model training, therefore, greatly reduces the difficulty of fault diagnosis model training, provides condition for shortening the training time.
The RBF radial basis function neural network that the present invention adopts is a kind of feed forward type neural network efficiently, and it has the best approximation capability and global optimum's characteristic that other feedforward networks do not have, and structure simply, and training speed is fast.RBF network is also called local acceptance domain network, namely only when input to drop in the input space in a very little appointed area time, hidden unit just makes significant non-zero response.The local of network accepts the concept that characteristic makes to imply during its decision-making distance, and namely only have when inputting the acceptance domain close to RBF network, network just can make response to it.Avoiding problems any division characteristic that BP network remote sensing brings.
For RBF neural, the equal value of the weights between usual input layer to hidden layer is 1.Center and the radius of Hidden unit radial basis function also pre-determine usually, only have the weights between hidden layer to output layer adjustable.The hidden layer of RBF network performs a kind of changeless nonlinear transformation, and the input space is mapped to a new hidden layer space, output layer realizes linear combination in the space that this is new.Obviously due to the linear characteristic of output unit, its parameter regulates very simple, and there is not local minimum problem.In addition, there are some researches show, the nonlinear activation functional form that general RBF network utilizes, not most important on the impact of network performance, key factor is choosing of Basis Function Center.Therefore, belong to the RBF network of feedforward neural network together with BP network, overcome some problems existing for BP network to a certain extent.
For these reasons, the present invention has better convergence property and shorter training time.
3, there is better Fault Identification ability.
Due to separate between each submodel of the present invention, thus through training after submodel more responsive to learnt specific fault conditions, which enhance the Fault Identification ability of fault diagnosis model.
4, restructuring and model retraining can be carried out easily according to its application conditions.
The present invention is when the identification kind needing increase model to electrical fault, or have new characteristic to certain malfunction, when needing re-training correspondence submodel, only need to train or retraining corresponding submodel, be then recombined in original model.And for the For Diagnosing Faults of Electrical model be made up of a Multiple input-output neural network, then in said case, must to whole network re-training.Model structure is more complicated, then convergence is more difficult with training.
Accompanying drawing explanation
Drawings attached of the present invention 5, wherein:
Fig. 1 is structural representation of the present invention;
Fig. 2 is process flow diagram of the present invention;
Fig. 3 is training process flow diagram of the present invention;
Fig. 4 is test flow chart of the present invention;
Fig. 5 is one of the present invention and use-case structural representation is described.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described further.Based on a Fault Diagnosis of AC Electric Machine model construction method for RBF neural, comprise the following steps:
A, set up fault diagnosis model by shown in Fig. 2
A1, determine the structure of fault diagnosis model
First determine to form the submodel number of For Diagnosing Faults of Electrical model according to the number intending diagnosis alternating current generator failure mode; Each submodel is the model of a multi input-mono-output, and namely each submodel has multiple input end and single output terminal, and the input end number of each submodel is identical; The input of each submodel is connected in parallel the For Diagnosing Faults of Electrical model just constituted by n sub-model-composing, as shown in Figure 1.As shown in Figure 1, the input of model inputs parallel connection by each submodel and forms, and the output of model is made up of the output of all submodels, that is, and the number of submodel i.e. the output number of model.
Described submodel, as part in solid box in Fig. 1, is be made up of a multi input-mono-output three-layer neural network.The input number of the corresponding submodel of the input layer number of neural network, output layer node number is 1, the output number of corresponding submodel, and each submodel is used for representing a kind of malfunction of motor;
A2, determine the input end number of fault diagnosis model
If submodel has m input end, that is fault diagnosis model has m input end, if the input signal of m input end forms an input vector x, is expressed as follows:
x＝(x
1x
2…x
m)
In formula, x
ifor i-th input signal of submodel, i=1,2 ..., m;
Described input signal is the radial vibration acceleration signal of alternating-current motor stator current signal or alternating current arbor or the axial vibration acceleration signal of alternating current arbor; The electrical fault energy Ratios vector that described input vector obtains through WAVELET PACKET DECOMPOSITION for input signal; Described energy Ratios vector is referred to as proper vector, and each component of proper vector correspond to the signal energy of a certain frequency band and the gross energy ratio of Whole frequency band signal; The element number of proper vector is determined by decomposed signal band number; Set the signal band number that will decompose as 4 here, like this, the sample set form that prepare is the form shown in table 1.What the every a line in table 1 provided is an energy Ratios vector, because decomposed signal band number is 4, so this energy Ratios vector has four elements, also determines the input end number m=4 of fault diagnosis model simultaneously.
The element number of proper vector is the input number m of fault diagnosis model, namely the input number m of each submodel;
Table 1 is trained and test sample book sample table
Table 1 is submodel training and test sample book collection sample table.To illustrate the problem is simple, in table, only give the two states of motor, i.e. unfaulty conditions and malfunction.Each electrical fault proper vector is made up of four components (element), represent four different sub-band signal energy respectively with the ratio of Whole frequency band signal energy.
A3, determine the output terminal number of fault diagnosis model
If fault diagnosis model is by n sub-model-composing, then the output signal of all submodels that its output vector y is comprised by it is formed, that is:
y＝(y
1y
2…y
n)
In formula, y
jfor a jth output signal of For Diagnosing Faults of Electrical model, the namely output signal of a jth submodel, j=1,2 ..., n, the output terminal number of each submodel is fixed as 1, represents by a kind of malfunction of diagnosing motor;
Simple for illustrating, if two kinds of electrical fault states will be identified, unfaulty conditions and malfunction; This two states indicates with a submodel respectively.That is, the For Diagnosing Faults of Electrical model that construct, by two sub-model-composings, might as well identify the unfaulty conditions of indication motor in other words here, identify the malfunction of motor with submodel 2 with submodel 1.Now, the output number n=2 of For Diagnosing Faults of Electrical model.
A4, determine the Hidden nodes of submodel
First use default value 2 or 4 as the Hidden nodes of submodel, finally adjust the Hidden nodes of submodel according to training result;
B, train fault diagnosis model by shown in Fig. 3
The output state of B1, definition submodel
For each submodel defines its fault type that will represent, any one submodel so after training, only have when inputting its corresponding fault type energy Ratios vector, the output signal of this submodel is 1, otherwise output signal is 0;
B2, preparation submodel training sample set
To each fault type, the energy Ratios vector adopting the fault data being not less than 10 groups to generate forms the training sample set of each submodel; When the electrical fault number of types that will identify is 2, when the sub-band number that decompose is 4, the submodel training sample set prepared is as shown in table 1.Table 1 training sample concentrates first 5 groups, and what provide is the energy Ratios that four sub-bands under motor unfaulty conditions are corresponding, and every a line forms an energy Ratios vector, as input vector during training submodel; These 5 energy Ratios vectors corresponding, the ideal of submodel 1 exports and is 1; Table 1 training sample concentrate after 5 groups what provide is the energy Ratios that four sub-bands under electrical fault state are corresponding, every a line forms an energy Ratios vector, as input vector during training submodel; These 5 energy Ratios vectors corresponding, the ideal of submodel 1 exports and is 0.In like manner, when training submodel 2 by these 10 groups of sample datas, the ideal of the submodel 2 corresponding to front 5 energy Ratios vectors exports and is 0; The ideal of the submodel 2 that rear 5 energy Ratios vector is corresponding exports and is 1.
B3, training submodel
Respectively each submodel is trained with the training sample set of obtained each submodel; At this, training sample set in submodel 1 and submodel 2 all employing table 1 is trained.
B4, synthesis fault diagnosis model
After all submodel training terminate, the input end of each submodel after training is in parallel, just constitute the Fault Diagnosis of AC Electric Machine model that will set up, as shown in Figure 5;
C, by the diagnostic model of test failure shown in Fig. 4
C1, preparation fault diagnosis model test sample book collection
Test sample book collection is formed with the energy Ratios vector comprising fault type corresponding to all submodels being different from training sample set, as shown in table 1.According to each fault energy of test sample book collection putting in order than vector, a corresponding motor status exports table, as shown in table 2, and the ideal referred to herein as fault diagnosis model exports table; The corresponding test sample book of every a line in table concentrates the malfunction represented by corresponding line to export, and namely the ideal of fault diagnosis model exports;
The performance of fault diagnosis of C2, test failure diagnostic model
Fault energy successively in continuous input test sample set is than vector, the output vector of record trouble diagnostic model, obtain the actual output table of fault diagnosis model, the ideal output table of fault diagnosis model and actual output table are contrasted, test and evaluation is carried out to the performance of fault diagnosis of fault diagnosis model; If the matching degree of the actual output table of fault diagnosis model and desirable output table is greater than 80%, namely think and meet the demands, then this fault diagnosis model just can come into operation.
Fault diagnosis model described in step C2 of the present invention is actual exports table and the desirable acquisition methods exporting the matching degree shown, and comprises the following steps:
Numerical value in the actual output table of the fault diagnosis model obtained in fault diagnosis model test process is the real number in one [0,1] interval; In the actual output table of fault diagnosis model, each value is got in close to ideal output table and is respectively worth, then represent that the performance of fault diagnosis of fault diagnosis model is better; In order to obtain, fault diagnosis model is actual to be exported and the desirable matching degree exported, first each numerical value in the actual output table of fault diagnosis model is carried out round process, obtain a corresponding new table formed by 0,1, be called the actual output of fault diagnosis model here and round table; When the actual output of fault diagnosis model rounds table and desirable output table corresponding row is respectively worth completely the same, claim corresponding row coupling, all mate with all row in desirable output table if actual output rounds table, then title reality exports to round and shows to mate with desirable output table 100%.
In the present embodiment, the real output value in table 3 is carried out the calculating that rounds up, obtains table 4, then contrast with table 2 line by line.Only have the third line of the third line and table 2 inconsistent in this routine table 4, so its matching degree is (1-1/12) * 100%=92%.Illustrate that this fault diagnosis model meets the demands, can come into operation.
The model ideal of table 2 corresponding table 1 test sample book collection exports table
Table 2 is the desirable output tables of fault diagnosis model.Table 2 is for test sample book in table 1, exports y at hypothesis submodel 1
1instruction unfaulty conditions, submodel 2 exports y
2fault diagnosis model ideal in indicating fault status situation exports table.
The actual output table of model of table 3 corresponding table 1 test sample book collection
Table 3 is the actual output tables of fault diagnosis model.Table 3 is that test sample book successively order is input to the actual output table of model generated after in fault diagnosis model.
The actual output of table 4 model rounds table
Table 4 be fault diagnosis model actual export round table.This table, from table 3, obtains after being rounded by element each in table 3 according to the principle rounded up.
For the fault diagnosis model that can come into operation, if some components are y in the actual output vector y of fault diagnosis model
j=1, then show that motor is in the malfunction of a jth sub-model instruction; If y
j=0, then show that this fault does not appear in motor; If 0<y
j<1, then show that motor is with y
jdegree of confidence may there is this fault.
Claims (2)
1., based on a Fault Diagnosis of AC Electric Machine model construction method for RBF neural, it is characterized in that: comprise the following steps:
A, set up fault diagnosis model
A1, determine the structure of fault diagnosis model
First determine to form the submodel number of For Diagnosing Faults of Electrical model according to the number intending diagnosis alternating current generator failure mode; Each submodel is the model of a multi input-mono-output, and namely each submodel has multiple input end and single output terminal, and the input end number of each submodel is identical; Be connected in parallel by the input end of each submodel and namely form For Diagnosing Faults of Electrical model, namely the output terminal number of all submodels forms the output terminal number of For Diagnosing Faults of Electrical model, that is the number of submodel is exactly the output terminal number of For Diagnosing Faults of Electrical model;
Described submodel is made up of a multi input-mono-output three-layer neural network; The input end number of the corresponding submodel of the input layer number of described neural network, output layer node number is 1, the output terminal number of corresponding submodel, and each submodel is used for representing a kind of malfunction of motor;
A2, determine the input end number of fault diagnosis model
If submodel has m input end, that is fault diagnosis model has m input end, if the input signal of m input end forms an input vector x, is expressed as follows:
x＝(x
1x
2…x
m)
In formula, x
ifor i-th input signal of submodel, i=1,2 ..., m;
Described input signal is the radial vibration acceleration signal of alternating-current motor stator current signal or alternating current arbor or the axial vibration acceleration signal of alternating current arbor; The electrical fault energy Ratios vector that described input vector obtains through WAVELET PACKET DECOMPOSITION for input signal; Described energy Ratios vector is referred to as proper vector, and each component of proper vector correspond to the signal energy of a certain frequency band and the gross energy ratio of Whole frequency band signal; The element number of proper vector is determined by decomposed signal band number;
The element number of proper vector is the input end number m of fault diagnosis model, namely the input end number m of each submodel;
A3, determine the output terminal number of fault diagnosis model
If fault diagnosis model is by n sub-model-composing, then the output signal of all submodels that its output vector y is comprised by it is formed, that is:
y＝(y
1y
2…y
n)
In formula, y
jfor a jth output signal of For Diagnosing Faults of Electrical model, the namely output signal of a jth submodel, j=1,2 ..., n, the output terminal number of each submodel is fixed as 1, represents by a kind of malfunction of diagnosing motor;
A4, determine the Hidden nodes of submodel
First use default value 2 or 4 as the Hidden nodes of submodel, finally adjust the Hidden nodes of submodel according to training result;
B, training fault diagnosis model
The output state of B1, definition submodel
For each submodel defines its fault type that will represent, any one submodel so after training, only have when inputting its corresponding fault type energy Ratios vector, the output signal of this submodel is 1, otherwise output signal is 0;
B2, preparation submodel training sample set
To each fault type, the energy Ratios vector adopting the fault data being not less than 10 groups to generate forms the training sample set of each submodel;
B3, training submodel
Respectively each submodel is trained with the training sample set of obtained each submodel;
B4, synthesis fault diagnosis model
After all submodel training terminate, the input end of each submodel after training is in parallel, just constitute the Fault Diagnosis of AC Electric Machine model that will set up;
C, test failure diagnostic model
C1, preparation fault diagnosis model test sample book collection
Test sample book collection is formed with the energy Ratios vector comprising fault type corresponding to all submodels being different from training sample set, according to each fault energy of test sample book collection putting in order than vector, a corresponding motor status exports table, and the ideal referred to herein as fault diagnosis model exports table; The corresponding test sample book of every a line in table concentrates the malfunction represented by corresponding line to export, and namely the ideal of fault diagnosis model exports;
The performance of fault diagnosis of C2, test failure diagnostic model
Fault energy successively in continuous input test sample set is than vector, the output vector of record cast, obtain the actual output table of model, the ideal output table of model and actual output table are contrasted, test and evaluation is carried out to the performance of fault diagnosis of fault diagnosis model; If the matching degree of the actual output table of fault diagnosis model and desirable output table is greater than 80%, namely think and meet the demands, then this fault diagnosis model just can come into operation.
2. a kind of Fault Diagnosis of AC Electric Machine model construction method based on RBF neural according to claim 1, it is characterized in that: the fault diagnosis model described in step C2 is actual exports table and the desirable acquisition methods exporting the matching degree shown, and comprises the following steps:
Numerical value in the actual output table of the fault diagnosis model obtained in fault diagnosis model test process is the real number in one [0,1] interval; In the actual output table of fault diagnosis model, each value is got in close to ideal output table and is respectively worth, then represent that the performance of fault diagnosis of fault diagnosis model is better; In order to obtain, fault diagnosis model is actual to be exported and the desirable matching degree exported, first each numerical value in the actual output table of fault diagnosis model is carried out round process, obtain a corresponding new table formed by 0,1, be called the actual output of fault diagnosis model here and round table; When the actual output of fault diagnosis model rounds table and desirable output table corresponding row is respectively worth completely the same, claim corresponding row coupling, all mate with all row in desirable output table if actual output rounds table, then title reality exports to round and shows to mate with desirable output table 100%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310167662.3A CN103294849B (en) | 2013-05-08 | 2013-05-08 | Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310167662.3A CN103294849B (en) | 2013-05-08 | 2013-05-08 | Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103294849A CN103294849A (en) | 2013-09-11 |
| CN103294849B true CN103294849B (en) | 2015-11-18 |
Family
ID=49095706
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310167662.3A Expired - Fee Related CN103294849B (en) | 2013-05-08 | 2013-05-08 | Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103294849B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ES2549652B1 (en) * | 2014-04-24 | 2016-10-07 | Alstom Transporte, S.A. | Method and system to automatically detect faults in a rotary axis |
| CN105116323B (en) * | 2015-08-14 | 2017-10-17 | 江苏科技大学 | A kind of electrical fault detection method based on RBF |
| CN105487009A (en) * | 2015-11-19 | 2016-04-13 | 上海电机学院 | Motor fault diagnosis method based on k-means RBF neural network algorithm |
| CN106872894B (en) * | 2017-03-03 | 2020-01-17 | 南方科技大学 | Fault detection method and device for three-phase motor |
| CN108106846B (en) * | 2017-12-21 | 2019-09-27 | 大连交通大学 | A kind of rolling bearing fault damage extent identification method |
| CN108931724A (en) * | 2018-07-30 | 2018-12-04 | 袁小芳 | A kind of servo motor method for diagnosing faults |
| CN110135021B (en) * | 2019-04-24 | 2023-07-14 | 南京航空航天大学 | ATRU system fault grading diagnosis method based on multi-source signals and RBF neural network |
| CN112346941A (en) * | 2019-08-08 | 2021-02-09 | 北京国双科技有限公司 | Fault diagnosis method and device |
| CN110954827B (en) * | 2019-12-17 | 2022-04-19 | 北京昊鹏智能技术有限公司 | Fault diagnosis method and device, electronic equipment and system |
| CN111002832B (en) * | 2019-12-24 | 2022-12-23 | 东风电子科技股份有限公司 | Method for realizing configurable processing of motor controller fault diagnosis system in electric automobile based on computer software |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101661075A (en) * | 2009-06-08 | 2010-03-03 | 浙江大学 | Power system failure diagnostic method based on neural network and fuzzy integral |
| CN101872165A (en) * | 2010-06-13 | 2010-10-27 | 西安交通大学 | Method for fault diagnosis of wind turbines on basis of genetic neural network |
| TW201144840A (en) * | 2010-06-07 | 2011-12-16 | Nat Univ Chin Yi Technology | Diagnosis method and device for dysfunctional motor using RBF information integration technique |
| CN102707228A (en) * | 2011-11-12 | 2012-10-03 | 江苏兴航智控科技股份有限公司 | Neural network expert system-based electric machine fault intelligent diagnosis system |
-
2013
- 2013-05-08 CN CN201310167662.3A patent/CN103294849B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101661075A (en) * | 2009-06-08 | 2010-03-03 | 浙江大学 | Power system failure diagnostic method based on neural network and fuzzy integral |
| TW201144840A (en) * | 2010-06-07 | 2011-12-16 | Nat Univ Chin Yi Technology | Diagnosis method and device for dysfunctional motor using RBF information integration technique |
| CN101872165A (en) * | 2010-06-13 | 2010-10-27 | 西安交通大学 | Method for fault diagnosis of wind turbines on basis of genetic neural network |
| CN102707228A (en) * | 2011-11-12 | 2012-10-03 | 江苏兴航智控科技股份有限公司 | Neural network expert system-based electric machine fault intelligent diagnosis system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103294849A (en) | 2013-09-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103294849B (en) | Based on the Fault Diagnosis of AC Electric Machine model construction method of RBF neural | |
| Li et al. | Motor current signal analysis using deep neural networks for planetary gear fault diagnosis | |
| Seshadrinath et al. | Investigation of vibration signatures for multiple fault diagnosis in variable frequency drives using complex wavelets | |
| Zhao et al. | Intelligent fault diagnosis of multichannel motor–rotor system based on multimanifold deep extreme learning machine | |
| CN106184215B (en) | The active damping controls device and method of hybrid vehicle | |
| Ayhan et al. | Multiple discriminant analysis and neural-network-based monolith and partition fault-detection schemes for broken rotor bar in induction motors | |
| Merainani et al. | A novel gearbox fault feature extraction and classification using Hilbert empirical wavelet transform, singular value decomposition, and SOM neural network | |
| Jiang et al. | An efficient fault diagnostic method for three-phase induction motors based on incremental broad learning and non-negative matrix factorization | |
| CN112041693A (en) | Power distribution network fault positioning system based on mixed wave recording | |
| Duan et al. | Method of inter‐turn fault detection for next‐generation smart transformers based on deep learning algorithm | |
| Gao et al. | ASM1D-GAN: An intelligent fault diagnosis method based on assembled 1D convolutional neural network and generative adversarial networks | |
| CN103136587A (en) | Power distribution network operating state classification recognition method based on support vector machine | |
| CN103207950A (en) | Intelligent transformer fault diagnostic method based on RBF (radial basis function) neural network | |
| Guolian et al. | Research on fault diagnosis of wind turbine control system based on artificial neural network | |
| Rahnama et al. | Machine‐learning approach for fault detection in brushless synchronous generator using vibration signals | |
| CN104502754A (en) | Fault diagnosis method for pure electric vehicle power system | |
| Hou et al. | Multiple sensors fault diagnosis for rolling bearing based on variational mode decomposition and convolutional neural networks | |
| CN113189513B (en) | Ripple-based redundant power supply current sharing state identification method | |
| CN114118149A (en) | Induction motor fault diagnosis system based on finite element simulation and symmetric feature migration | |
| CN116992326A (en) | Transformer fault identification method based on wavelet packet-mapping network algorithm | |
| CN105116323B (en) | A kind of electrical fault detection method based on RBF | |
| CN105976127A (en) | Risk evaluation method and apparatus for integrated electric propulsion system | |
| Sun et al. | An improved data privacy diagnostic framework for multiple machinery components data based on swarm learning algorithm | |
| CN114814391A (en) | Charging pile fault identification method and storage medium | |
| CN113419172A (en) | New energy automobile three-phase asynchronous motor fault identification based on GCN and VMD-ED |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20151118 Termination date: 20160508 |