CN106526400A - Grounding fault diagnosing method and apparatus of DC 600V train power supply system - Google Patents
Grounding fault diagnosing method and apparatus of DC 600V train power supply system Download PDFInfo
- Publication number
- CN106526400A CN106526400A CN201611064578.9A CN201611064578A CN106526400A CN 106526400 A CN106526400 A CN 106526400A CN 201611064578 A CN201611064578 A CN 201611064578A CN 106526400 A CN106526400 A CN 106526400A
- Authority
- CN
- China
- Prior art keywords
- fault
- earth fault
- waveform
- power supply
- supply system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
-
- 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
- G01R31/005—Testing of electric installations on transport means
- G01R31/008—Testing of electric installations on transport means on air- or spacecraft, railway rolling stock or sea-going vessels
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Short-Circuits, Discontinuities, Leakage, Or Incorrect Line Connections (AREA)
Abstract
The invention discloses a grounding fault diagnosing method and apparatus of a DC 600V train power supply system. The method includes the steps of obtaining fault waveform sets of the DC600V train power supply system when different types of ground faults occur, wherein each fault waveform sample set corresponds to one ground fault type, training each fault waveform set to obtain a fault classification model, and acquiring current fault waveform on a real-time basis when a grounding fault occurs to a target DC600V train power supply system and diagnosing the grounding fault type corresponding to the current fault waveform according to the fault classification model. The apparatus comprises a classification model training module and a fault diagnosing module. The method and apparatus have the advantages of simple realization operation, high diagnosis efficiency and precision, high universality, good expansion performance and easy execution and maintenance.
Description
Technical field
The present invention relates to electric locomotive technical field, more particularly to a kind of electric power system suitable for DC600V trains connects
Earth fault diagnostic method and device.
Background technology
DC600V power supply system of train is widely used in air-conditioned train, the earth-fault protection in its electric power system
It is to ensure train operation, the important component part of passenger's personal safety, is to ensure that electric locomotive train plays important work in operation
With.The key that earth-fault protection is realized be that the recognition detection to earth fault waveform and different faults pattern point
From i.e. grounding fault diagnosis.Grounding fault diagnosis are divided into two key steps of fault detect and fault reconstruction, wherein fault detect
To judge whether system occurs earth fault, quickly to respond action, life and property loss is reduced;Fault reconstruction is then
Failure to occurring is identified, positions, and even repairs automatically for removal of faults on system and provides important guiding.
DC600V electric power system earth fault types are various and are difficult to be accurately positioned, and first, electric power system is mainly divided
Rectification unit, DC power supply unit, inversion output unit etc., different parts ground connection is with different fault modes;Secondly, train
Middle electromagnetic environment is complicated, loadtype, load change all can the change of causing trouble waveform, i.e., fault waveform is in complicated work
It is likely to unstable on the vehicles such as condition, different batches, load change, and the electromagnetic environment on electric locomotive, motor-car is very
Complexity, in the presence of electromagnetic interference, minutia is also unreliable, and these all can be brought to the positioning of earth fault, mask work
Very big puzzlement.
Correlative study currently for the grounding fault diagnosis of DC600V electric power systems is less, does not also have effectively diagnosis real
Existing method, existing method for diagnosing faults are then generally in accordance with some local features of fault waveform and enter line algorithm design, real
Existing waveform recognition, fault diagnosis, and problems with is had based on the diagnostic method of fault waveform local feature:
1. identification feature is comprehensive, it is inaccurate to diagnose.Especially under new road conditions, new load environment, most probably occur
None- identified, the problems such as judge, fail to judge by accident so that yet suffer from potential safety hazard, it is impossible to effectively meet requirement of engineering;
2. poor universality.The above-mentioned algorithm for local feature is generally difficult to general, need to carry out for every kind of operating mode
Parameter optimization, and Feature Selection, algorithm design are carried out for every kind of failure, or even algorithm design is re-started, when failure mould
Formula is more, work condition environment is often difficult to when complicated;
3. maintainability is poor.When electric power system variation, external environment condition change, great effort is needed to adjust ginseng or even re-start
Algorithm is designed, and is difficult to safeguard when there is new fault mode, and is generally required for different application in practical engineering application
Locomotive carries out specific algorithm process so that the maintenance upgrade of said method is extremely difficult;
4. required workload is big.Need all to carry out feature for every kind of fault mode based on fault waveform local feature method
Analysis, algorithm design, inefficiency, required workload are big, time-consuming serious.
5. poor robustness.Recognizer generalisation properties based on local feature are poor, for complex environment is difficult to stable enter
Row Fault Identification, separation.
The content of the invention
The technical problem to be solved in the present invention is that:For the technical problem that prior art is present, the present invention provides one
Kind realize easy to operate, diagnosis efficiency and high precision, highly versatile, scalability be good and the DC600V trains of maintenance easy to carry out
The earth fault diagnostic method and device of electric power system.
To solve above-mentioned technical problem, technical scheme proposed by the present invention is:
A kind of earth fault diagnostic method of DC600V power supply system of train, step include:
1) fault waveform sample set of the DC600V power supply system of train in different earth fault types, each institute are obtained
State a kind of earth fault type of fault waveform sample set correspondence;Each fault waveform sample set is trained, failure is obtained
Disaggregated model;
2) when target DC600V power supply system of train occurs earth fault, Real-time Collection current failure waveform, and according to
Earth fault type corresponding to the failure modes Model Diagnosis current failure waveform.
As the further improvement of the inventive method:The fault waveform is that DC600V power supply system of train occurs ground connection event
The waveform of half voltage during barrier.
As the further improvement of the inventive method:The step 1) in it is concrete respectively each fault waveform is concentrated therefore
Barrier waveform carries out spectrum analysis, and extraction includes that the characteristic vector of spectrum information is trained, and obtains failure modes model;It is described
Step 2) in specifically carry out frequency analysis to current failure waveform, and extract the characteristic vector for including spectrum information, by institute
State failure modes model to be identified the characteristic vector extracted, obtain corresponding earth fault type.
As the further improvement of the inventive method:The characteristic vector include the Amplitude Ration of front nth harmonic and fundamental wave with
And DC component information, wherein n is the harmonic number specified;The characteristic vector also includes phase information.
As the further improvement of the inventive method:It is described when carrying out spectrum analysis, if periodic signal, then to be analyzed
Waveform carries out periodic signal positioning;If multicycle and the signal of mechanical periodicity, then treat analysis waveform and classified.
As the further improvement of the inventive method:The step 1) also include model optimization step, concretely comprise the following steps:Institute
State step 1) also include model optimization step, concretely comprise the following steps:Test sample is given by the failure modes model to connect in difference
The prediction probability of earth fault type, is carried out to the failure modes model according to the prediction probability of each earth fault type excellent
Change;And increase required earth fault type, fault waveform in training set newly, after checking correction is repeated, obtain final
Failure modes model.
As the further improvement of the inventive method:The step 1) in be specifically based on random forest grader training failure
Disaggregated model.
As the further improvement of the inventive method:The earth fault type include rectification side anode, rectification side negative terminal,
The ground connection of reactance prime, DC600V positive output ends, DC600V negative output terminals, inverter side A phase, inverter side B phase and inverter side C phase
Failure.
As the further improvement of the inventive method:The step 2) in connect when target DC600V power supply system of train
It is during earth fault, concrete that DC ground fault is isolated according to the half voltage of target DC600V power supply system of train first, if not
DC ground fault, then Real-time Collection current failure waveform, and according to the failure modes Model Diagnosis current failure waveform institute
Corresponding earth fault type.
It is as the further improvement of the inventive method, described to isolate concretely comprising the following steps for DC ground fault:If current
Half voltage is DC voltage, then be judged to DC ground fault, and if the value of current half voltage half when being deflected downwardly nominal situation
Voltage, is diagnosed as DC600V positive output ends and earth fault occurs, and if the value of current half voltage when being upwardly deviated from nominal situation
Half voltage, is diagnosed as DC600V negative output terminals and earth fault occurs.
As the further improvement of the inventive method, the step 1) afterwards, step 2) before also include failure detection steps, tool
Body step is:The half voltage and leakage current of real-time detection target DC600V power supply system of train, the half voltage that comprehensive detection is arrived
And leakage current, determine whether earth fault, if it is determined that to there is earth fault, proceeding to execution step 2).
A kind of earth fault diagnostic equipment of DC600V power supply system of train, including:
Disaggregated model training module, for obtaining DC600V power supply system of train when there is different earth fault type
Fault waveform sample set, a kind of corresponding earth fault type of each described fault waveform sample set;By each fault waveform
Sample set is trained, and obtains failure modes model;
Fault diagnosis module, for when target DC600V power supply system of train occurs earth fault, Real-time Collection is current
Fault waveform, and the earth fault type according to corresponding to the failure modes Model Diagnosis current failure waveform.
Used as the further improvement of apparatus of the present invention, the disaggregated model training module includes:
Fisrt feature amount extraction unit, for concentrating fault waveform to carry out spectrum analysis each fault waveform respectively,
And extract the characteristic vector composition failure modes model training collection for including spectrum information;
Training unit, for being trained to the failure modes model training collection, obtains failure modes model;
The fault diagnosis module includes:
Data acquisition unit, for when target DC600V power supply system of train occurs earth fault, Real-time Collection is current
Fault waveform;
Second feature amount extraction unit, for frequency analysis is carried out to current failure waveform, and extracts described including frequency spectrum
The characteristic vector of information, obtains characteristic vector to be sorted;
Diagnosis recognition unit, for diagnosis identification being carried out to the characteristic vector to be sorted by the failure modes model,
Obtain corresponding earth fault type.
Compared with prior art, it is an advantage of the current invention that:
1) it is of the invention based on DC600V power supply system of train earth fault types are more, be difficult the features such as positioning, by obtaining
Fault waveform sample set training during different earth fault types obtains failure modes model, recycles the failure point that training is obtained
Class model carries out diagnosis identification to the real-time earth fault type for occurring, and it is convenient that operation is realized, can be common to DC600V trains
The various grounding fault diagnosis of electric power system, without the need for carrying out special algorithm design for different faults type, greatly reduce institute
The workload for needing, and the failure modes model obtained based on a large amount of fault waveform samples, can quickly, accurately when failure occurs
Diagnosis identify earth fault type, while can easily carry out upgrade maintenance by disaggregated model;
2) present invention is further used as characteristic vector by obtaining spectrum information, is capable of the entirety of Efficient Characterization fault waveform
Feature, it is ensured that the accuracy of identification of failure modes model;Characteristic vector specifically take including front nth harmonic and fundamental voltage amplitude than Mag and
, there is no two dimensional image in the one-dimensional characteristic vector of DC component information, characteristic vector chooses reasonable, can be very
The global feature of good sign waveform, minutia is filtered, while meet the characteristics such as constant, translation invariant of stretching, so as to obtain
High-precision failure modes model;Characteristic vector further comprises phase place Phase, and characterization failure waveform that can be complete is special
Levy;
3) present invention disclosure satisfy that the spirit of different faults type further by verifying repeatedly and optimizing failure modes model
Quick identification, realizes the effective classification to various faults pattern, while new fault type or right can be supported according to the change of sample
In same fault type, under new operating mode, the diagnosis of waveform is recognized, be with good expansibility energy and maintainability performance;
4) present invention is based further on random forest grader training failure modes model, the random forest obtained by training
Disaggregated model has very high stability to complex work environment, carries out constantly expanding modification by the training set to model, i.e.,
Required nicety of grading can be realized, while by increasing new samples or deleting old sample re -training model, in that context it may be convenient to real
The capacity upgrade of existing random forest grader;
5) it is further contemplated that in actual environment half voltage waveform disturbances etc. caused by load effect, electromagnetic interference
Situation, comprehensive half voltage, leakage current judge whether system occurs earth fault, can effectively improve fault detect accuracy,
Robustness, so as to action can be quickly responded when failure occurs.
Description of the drawings
Fig. 1 is that the earth fault diagnostic method of the present embodiment DC600V power supply system of train realizes schematic flow sheet.
Fig. 2 is the principle schematic diagram. of DC600V electric power systems.
Fig. 3 is 1 point of earth fault waveform diagram of DC600V electric power systems in the specific embodiment of the invention.
Fig. 4 is the waveform result schematic diagram obtained by low-order harmonic fitting in the specific embodiment of the invention.
Fig. 5 is the present embodiment realizes principle schematic based on what random forest was grounded fault diagnosis.
Specific embodiment
Below in conjunction with Figure of description and concrete preferred embodiment, the invention will be further described, but not therefore and
Limit the scope of the invention.
As shown in figure 1, the earth fault diagnostic method of the present embodiment DC600V power supply system of train, step includes:
1) fault waveform sample set of the DC600V power supply system of train in different earth fault types is obtained, each event
A kind of earth fault type of barrier waveform sample collection correspondence;Each fault waveform sample set is trained, failure modes model is obtained;
2) when target DC600V power supply system of train occurs earth fault, Real-time Collection current failure waveform, and according to
Earth fault type corresponding to failure modes Model Diagnosis current failure waveform.
The present embodiment based on DC600V power supply system of train earth fault types are more, the features such as be difficult positioning, by acquisition
Fault waveform sample set training during different earth fault types obtains failure modes model, recycles the failure point that training is obtained
Class model carries out diagnosis identification to the real-time earth fault type for occurring, and it is convenient that operation is realized, can be common to DC600V trains
The various grounding fault diagnosis of electric power system, without the need for carrying out special algorithm design for different faults type, greatly reduce institute
The workload for needing, and the failure modes model obtained based on a large amount of fault waveform samples, can quickly, accurately when failure occurs
Diagnosis identify earth fault type, while can easily carry out upgrade maintenance by disaggregated model.
As shown in Fig. 2 main 8 kinds of earth fault types that the present embodiment specifically takes DC600V power supply system of train are examined
Disconnected identification, bears including rectification side anode (1), rectification side negative terminal (2), reactance prime (3), DC600V positive output ends (4), DC600V
The earth fault of output end (5), inverter side A phase (6), inverter side B phase (7) and inverter side C phase (8).Beforehand through actual measurement or
Emulation obtains the fault waveform collection in such as above-mentioned 8 kinds of different earth fault types, and fault waveform collection includes correspondence earth fault
The typical fault waveform that type is obtained in the earth fault of different operating modes (such as different loads, different grounding resistance);Work as target
When DC600V power supply system of train occurs earth fault, the failure modes model obtained by training enters to different earth fault patterns
Row identification.In addition to above-mentioned 8 kinds of earth fault types, other more earth fault classes can also be set certainly according to the actual requirements
Type.
In the present embodiment, fault waveform is specially the ripple that DC600V power supply system of train occurs half voltage during earth fault
Shape.In as shown in Figure 2, voltage sensor SV2 as detects the half voltage of DC600V power supply system of train, DC600V train power supplies
System uses neutral earthing protection circuit, and when earth fault occurs, half voltage waveform can change, and connect from different
Place is closely related, then can characterize DC600V power supply system of train difference earth fault class by the feature of the waveform of half voltage
Type.
In the present embodiment, step 1) in it is concrete concentrate fault waveform to carry out spectrum analysis each fault waveform respectively, and carry
Take the characteristic vector including spectrum information to be trained, obtain failure modes model;Step 2) in specifically to current failure waveform
Frequency analysis is carried out, and extracts the characteristic vector for including spectrum information, the characteristic vector extracted is carried out by failure modes model
Identification, obtains corresponding earth fault type.FFT is carried out after collecting fault waveform to fault waveform, the window of FFT is adopted
Sample cycle T is fixed, directly can realize spectrum analysis using fixed sample rate and sampling number release.The present embodiment leads to
Acquisition spectrum information is crossed as characteristic vector, is capable of the global feature of Efficient Characterization fault waveform, it is ensured that failure modes model
Accuracy of identification.
The minutia of half voltage waveform is likely to unstable on the vehicles such as complex working condition, different batches, load change
Fixed, and the electromagnetic environment in electric locomotive is extremely complex, in the presence of electromagnetic interference, the minutia of half voltage also can not
Lean on.By taking 1 point of earth fault as an example, different loads actual measurement fault waveform as shown in figure 3, wherein (a) for load current 50A when half electricity
Corrugating, (b) for load current 150A when half voltage waveform, be (c) load current 300A, and (d) be load current 600A
When half voltage waveform.From the figure 3, it may be seen that half voltage waveform has the characteristics that during different loads:1. global feature is consistent, but details
Feature difference is larger;2. waveform can occur longitudinal extension change, and amplitude is not of uniform size;3. the cycle stability of waveform, no to occur
Significant changes.The present embodiment characteristic vector includes that DC component information, front nth harmonic and fundamental voltage amplitude, than Mag, can be good at
Characterize the global feature of waveform, filter minutia, while meeting the characteristics such as constant, translation invariant of stretching, wherein n is specified
Harmonic number, can consider according to amount of calculation, classification accuracy and be chosen.To take the low 8 subharmonic number of half voltage waveform
According to, as shown in figure 4, by low 8 subharmonic data can fitting obtain meeting the consistent waveform of original waveform, show DC component,
Low-order harmonic data can be good at characterizing wave character.
Alternating Current Power Supply of the input of DC600V power supply system of train for 50Hz, inversion output is also 50Hz, then it is corresponding its
Earth fault waveform be also the cycle of 50Hz and its multiple harmonic components, i.e. fault waveform be stable;In addition, by three-phase alternating current
When causing earth fault, the harmonic components of fault waveform are identical, phase place is different.The present embodiment characteristic vector further comprises phase
Position information, can further improve the nicety of grading of failure modes model.
The present embodiment specifically takes the front nth harmonic of frequency spectrum with fundamental voltage amplitude than Mag, phase place Phase and DC component information
It is as feature vector, X, more one-dimensional than what Mag, phase place Phase and DC component information were constituted with fundamental voltage amplitude by front nth harmonic
Characteristic vector, it is reasonable that characteristic vector is chosen, characterization failure wave character that can be complete, and there is no the rotation of two dimensional image etc.
Problem, while meeting the features such as stretching constant, translation invariant such that it is able to obtain high-precision failure modes model.Training event
During barrier disaggregated model, feature vector, X, composing training collection, to training are extracted respectively to each fault waveform in fault waveform sample set
Collection carries out repetition training and correction obtains stable failure modes model, it is ensured that the precision of failure modes model.
When carrying out spectrum analysis in the present embodiment, if periodic signal, then treating analysis waveform carries out periodic signal positioning,
So that phase place has stable phase reference, while being easy to simple realization, periodic signal positioning specifically can be using the ripple that will be input into
Shape Jing glide filter, search minimum of a value or additive method are realized;If multicycle and the signal of mechanical periodicity, then to ripple to be analyzed
Shape is classified, and classification specifically can be realized using fixed sample points, change sampling rate.
During the present embodiment training failure modes model, also including model optimization step, concretely comprise the following steps:By failure modes mould
Type provides prediction probability of the test sample in different earth fault types, to obtain indirectly the degree of correlation of different faults pattern,
Failure modes model is optimized according to the prediction probability of each earth fault type;And increase required connecing in training set newly
Earth fault type, fault waveform, or delete old sample and re-start training modeling, after checking correction is repeated, obtain final
Failure modes model.The test sample concretely simulated test of DC600V power supply system of train earth fault, historical data
Or emulation data etc..Failure modes model after above-mentioned checking repeatedly and optimization disclosure satisfy that the sensitive of different faults type
Identification, realizes the effective classification to various faults pattern, while new fault type can be supported or to same according to the change of sample
In one fault type, under new operating mode, the diagnosis of waveform is recognized, be with good expansibility energy and maintainability performance.
In the present embodiment, step 2) in when there is earth fault in target DC600V power supply system of train, concrete elder generation's basis
The half voltage of target DC600V power supply system of train isolates DC ground fault, if not DC ground fault, then in real time
Collection current failure waveform, and the earth fault type according to corresponding to failure modes Model Diagnosis current failure waveform.
The present embodiment can be realizing isolating DC ground fault using following two modes:
1. separated with fundamental voltage amplitude ratio based on DC component information DC.
The half voltage waveform of DC ground fault is direct current signal, and Jing after fft analysis, flip-flop DC is much larger than fundamental wave
Amplitude, can diagnose DC ground fault or AC earth failure with fundamental voltage amplitude ratio by DC component information DC, such as
Fruit is DC ground fault, further may recognize that DC600V positive output ends (4), DC600V according to the size of DC component DC
Negative output terminal (5) earth fault, i.e., offset up compared with half voltage under nominal situation (specifically taking 300V) for DC600V it is just defeated
Go out end (4) earth fault, DC600V negative output terminals (5) earth fault.
2. judge to be separated based on half voltage value.
Under DC600V electric power system normal operations, the half voltage that voltage sensor SV2 is measured in such as Fig. 2 is fixed
Value (specifically takes 300V), and when there is DC side earth fault, the decline of cable insulating grade, half voltage can be gradually deviated from
In the definite value, such as Fig. 2, DC600V positive output ends (4), DC600V negative output terminals (5) earth fault are DC ground fault, its
In when there is earth fault in DC600V positive output ends (4), half voltage Ujd and equivalent earth resistance Rjd are met and are closed with minor function
It is formula:
When DC600V negative output terminals (5) occur earth fault, half voltage Ujd and equivalent earth resistance Rjd meets following
Functional relation:
That is, when the half voltage Uj of DC600V electric power systems shows as DC voltage, DC ground fault is shown as, further
By judging with the departure degree of the size (such as 300V) of half voltage during nominal situation, half voltage Uj can recognize that DC600V is just defeated
Go out end (4), DC600V negative output terminals (5) earth fault.Concretely comprising the following steps for DC Line Fault is separated based on half voltage value:If current
Half voltage is DC voltage, then be judged to DC ground fault, and if the value of current half voltage half when being upwardly deviated from nominal situation
Voltage, is diagnosed as DC600V positive output ends and earth fault occurs, and if the value of current half voltage when being deflected downwardly nominal situation
Half voltage, is diagnosed as DC600V negative output terminals and earth fault occurs.Judge to separate the side of DC ground fault based on half voltage value
Formula, realizes simple and reliable, and can reduce compared to detached mode is carried out with fundamental voltage amplitude ratio based on DC component information DC
The extraction of characteristic quantity.
In the present embodiment, step 1) afterwards, step 2) before also include failure detection steps, concretely comprise the following steps:Real-time detection mesh
The mark half voltage of DC600V power supply system of train, leakage current, comprehensive detection to half voltage, leakage current determine whether ground connection
Failure, if it is determined that to there is earth fault, proceeding to execution step 2).Leakage current is earth current, during normal condition, protecting field
In no electric current flow through, constitute loop when earth fault occurs, cause the generation of leakage current, and the waveform of half voltage occur
Change, the present embodiment considers load effect in actual environment, caused by electromagnetic interference situations such as half voltage waveform disturbances, comprehensive
Half voltage, leakage current judge whether system occurs earth fault, can effectively improve accuracy, the robustness of fault detect, from
And action can be quickly responded when failure occurs.
The present embodiment specifically obtains leakage current I by the detection of SC1 current sensorsLeak, SV2 voltage sensor senses are to half electricity
Pressure UhalfWhen, if while meeting formula (3), (4), judgement occurs earth fault.
Wherein, IthFor current threshold set in advance.
Wherein, UthFor voltage threshold set in advance.
In the present embodiment, step 1) in be specifically based on random forest grader training failure modes model, naturally it is also possible to
Using other disaggregated models.Random forest grader has that speed is fast, antinoise, and the advantages of highly versatile, the present embodiment is based on
Random forest grader trains failure modes model, the random forest disaggregated model obtained by training to have complex work environment
Very high stability, carries out constantly expanding modification by the training set to model, you can to realize required nicety of grading, while logical
Cross increase new samples or delete old sample re -training model, in that context it may be convenient to realize the capacity upgrade of random forest grader,
Therefore generalisation properties are good, robustness, identification capability and adaptable, it is easy to carry out upgrade maintenance.
The present embodiment further illustrates the present invention as a example by based on random forest grader training failure modes model.
Distribute corresponding class label number for different faults type first, as shown in table 1;Again by the event of all kinds of fault types
Barrier waveform collection training probabilistic classifier, obtains failure modes model.
Table 1:Fault category classification chart.
In inverter side C phase earth fault (8) ground connection 600A loads, 1 Ω earth resistance operating modes in the specific embodiment of the invention
When, characteristic vector specifically takes front 7 subharmonic, obtains frequency spectrum list as shown in table 2, DC component DC=309.0V.
Table 2:The phase earth fault frequency spectrum list of inverter side C.
The then feature vector, X of above-mentioned fault waveformiFor:
Xi=[309.0,40.28%, 100.00,239.8,41.87,251.3,12.98,266.0,8.89, -4.7,
9.31,16.5,4.88,58.4,5.60,118.5], corresponding class label number is 8.Features described above vector XiConstitute a sample
This, constantly carries out feature extraction to the fault waveform under different faults pattern, different loads grounding requirement using same method,
Constitute the training set comprising various faults waveform sample;Then Random Forest model is trained, increases sample in training set newly
This, Random Forest model can support new fault type or the diagnosis identification to waveform under new operating mode in same fault type;
Check test is carried out to Random Forest model finally, and it is different faults type current test sample to be provided by random forest grader
Probability, guidance further optimize random forest grader, obtain final probabilistic classifier.
As shown in figure 5, the ground connection of DC600V power supply system of train is diagnosed in the specific embodiment of the invention based on random forest
During failure, upon the occurrence of a ground fault, the fault waveform of Real-time Collection DC600V power supply system of train is used as input signal, by week
Phase signal framing device isolates DC Line Fault after carrying out periodic signal positioning to input signal, if AC fault, then carries out
FFT simultaneously extracts feature vector, X [front nth harmonic is with fundamental voltage amplitude than Mag, phase place Phase, DC component information], obtains
Preliminary classification feature vector, XM, by preliminary classification feature vector, XMOutput diagnosis is input into the random forest grader for training
Recognition result.
Above-mentioned simply presently preferred embodiments of the present invention, not makees any pro forma restriction to the present invention.Although of the invention
It is disclosed above with preferred embodiment, but it is not limited to the present invention.Therefore, it is every without departing from technical solution of the present invention
Content, according to the technology of the present invention essence to any simple modification made for any of the above embodiments, equivalent variations and modification, all should fall
In the range of technical solution of the present invention protection.
Claims (13)
1. a kind of earth fault diagnostic method of DC600V power supply system of train, it is characterised in that step includes:
1) fault waveform sample set of the DC600V power supply system of train in different earth fault types is obtained, each described event
A kind of earth fault type of barrier waveform sample collection correspondence;Each fault waveform sample set is trained, failure modes are obtained
Model;
2) when target DC600V power supply system of train occurs earth fault, Real-time Collection current failure waveform, and according to described
Earth fault type corresponding to failure modes Model Diagnosis current failure waveform.
2. the earth fault diagnostic method of DC600V power supply system of train according to claim 1, it is characterised in that:It is described
Fault waveform is the waveform that DC600V power supply system of train occurs half voltage during earth fault.
3. the earth fault diagnostic method of DC600V power supply system of train according to claim 2, it is characterised in that:It is described
Step 1) in it is concrete concentrate fault waveform to carry out spectrum analysis each fault waveform respectively, and extract and include spectrum information
Characteristic vector is trained, and obtains failure modes model;The step 2) in frequency analysis is carried out to current failure waveform specifically,
And the characteristic vector for including spectrum information is extracted, the characteristic vector extracted is known by the failure modes model
Not, obtain corresponding earth fault type.
4. the earth fault diagnostic method of DC600V power supply system of train according to claim 3, it is characterised in that:It is described
Characteristic vector includes the Amplitude Ration and DC component information of front nth harmonic and fundamental wave, and wherein n is the harmonic number specified;It is described
Characteristic vector also includes phase information.
5. the earth fault diagnostic method of DC600V power supply system of train according to claim 4, it is characterised in that:It is described
When carrying out spectrum analysis, if periodic signal, then treating analysis waveform carries out periodic signal positioning;If multicycle and cycle change
The signal of change, then treat analysis waveform and classified.
6. the earth fault diagnostic method of the DC600V power supply system of train according to any one in Claims 1 to 5, its
It is characterised by:The step 1) also include model optimization step, concretely comprise the following steps:Test specimens are provided by the failure modes model
This prediction probability in different earth fault types, according to the prediction probability of each earth fault type to the failure modes
Model is optimized;And increase required earth fault type, fault waveform in training set newly, checking correction is repeated
Afterwards, obtain final failure modes model.
7. the earth fault diagnostic method of the DC600V power supply system of train according to any one in Claims 1 to 5, its
It is characterised by:The step 1) in be specifically based on random forest grader training failure modes model.
8. the earth fault diagnostic method of the DC600V power supply system of train according to any one in Claims 1 to 5, its
It is characterised by:The earth fault type include rectification side anode, rectification side negative terminal, reactance prime, DC600V positive output ends,
The earth fault of DC600V negative output terminals, inverter side A phase, inverter side B phase and inverter side C phase.
9. the earth fault diagnostic method of the DC600V power supply system of train according to any one in Claims 1 to 5, its
Be characterised by, the step 2) in when there is earth fault in target DC600V power supply system of train, it is concrete first according to target
The half voltage of DC600V power supply system of train isolates DC ground fault, if not DC ground fault, then Real-time Collection
Current failure waveform, and the earth fault type according to corresponding to the failure modes Model Diagnosis current failure waveform.
10. the earth fault diagnostic method of DC600V power supply system of train according to claim 9, it is characterised in that institute
State and isolate concretely comprising the following steps for DC ground fault:If current half voltage is DC voltage, it is judged to DC ground fault,
And if the value of current half voltage half voltage when being deflected downwardly nominal situation, be diagnosed as DC600V positive output ends and earth fault occur,
And if the value of current half voltage half voltage when being upwardly deviated from nominal situation, be diagnosed as DC600V negative output terminals there is ground connection therefore
Barrier.
The earth fault diagnostic method of the 11. DC600V power supply system of train according to any one in Claims 1 to 5,
It is characterized in that:The step 1) afterwards, step 2) before also include failure detection steps, concretely comprise the following steps:Real-time detection target
The half voltage and leakage current of DC600V power supply system of train, half voltage and leakage current that comprehensive detection is arrived determine whether to send out
Raw earth fault, if it is determined that to there is earth fault, proceeding to execution step 2).
12. a kind of earth fault diagnostic equipments of DC600V power supply system of train, it is characterised in that include:
Disaggregated model training module, for obtaining failure of the DC600V power supply system of train when there is different earth fault types
Waveform sample collection, a kind of corresponding earth fault type of each described fault waveform sample set;By each fault waveform sample
Collection is trained, and obtains failure modes model;
Fault diagnosis module, for when target DC600V power supply system of train occur earth fault when, Real-time Collection current failure
Waveform, and the earth fault type according to corresponding to the failure modes Model Diagnosis current failure waveform.
The earth fault diagnostic equipment of 13. DC600V power supply system of train according to claim 12, it is characterised in that institute
Stating disaggregated model training module includes:
Fisrt feature amount extraction unit, for concentrating fault waveform to carry out spectrum analysis each fault waveform respectively, and carries
Take the characteristic vector composition failure modes model training collection including spectrum information;
Training unit, for being trained to the failure modes model training collection, obtains failure modes model;
The fault diagnosis module includes:
Data acquisition unit, for when target DC600V power supply system of train occur earth fault when, Real-time Collection current failure
Waveform;
Second feature amount extraction unit, for frequency analysis is carried out to current failure waveform, and extracts described including spectrum information
Characteristic vector, obtain characteristic vector to be sorted;
Diagnosis recognition unit, for carrying out diagnosis identification to the characteristic vector to be sorted by the failure modes model, obtains
Corresponding earth fault type.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611064578.9A CN106526400B (en) | 2016-11-28 | 2016-11-28 | The earth fault diagnostic method and device of DC600V power supply system of train |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611064578.9A CN106526400B (en) | 2016-11-28 | 2016-11-28 | The earth fault diagnostic method and device of DC600V power supply system of train |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106526400A true CN106526400A (en) | 2017-03-22 |
CN106526400B CN106526400B (en) | 2019-11-22 |
Family
ID=58356633
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611064578.9A Active CN106526400B (en) | 2016-11-28 | 2016-11-28 | The earth fault diagnostic method and device of DC600V power supply system of train |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106526400B (en) |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108583629A (en) * | 2018-05-04 | 2018-09-28 | 兰州容大信息科技有限公司 | A kind of railcar business fault handling method |
CN109142946A (en) * | 2018-06-29 | 2019-01-04 | 东华大学 | Transformer fault detection method based on ant group algorithm optimization random forest |
CN109203995A (en) * | 2018-07-10 | 2019-01-15 | 中南大学 | A kind of trailer system major loop ground fault diagnosis method and system |
CN109212373A (en) * | 2017-06-29 | 2019-01-15 | 株洲中车时代电气股份有限公司 | The grounding fault diagnosis and insulation detecting method and device of power supply system of train |
CN109228871A (en) * | 2017-07-10 | 2019-01-18 | 比亚迪股份有限公司 | Power supply system of train and its detection of electrical leakage recovery device, method and train |
CN109239511A (en) * | 2017-07-10 | 2019-01-18 | 比亚迪股份有限公司 | Train, power supply system of train and its detection of electrical leakage positioning device, method |
CN109406923A (en) * | 2017-08-17 | 2019-03-01 | 株洲中车时代电气股份有限公司 | A kind of power supply system of train earth leakage failure prediction method and device |
CN109599827A (en) * | 2018-11-21 | 2019-04-09 | 宁波恒晨电力建设有限公司 | Power network neutral point dynamic earth fault method and system based on big data |
CN109765878A (en) * | 2018-12-24 | 2019-05-17 | 上海大郡动力控制技术有限公司 | The aided analysis method of new-energy automobile CAN bus network failure |
CN110794243A (en) * | 2019-11-14 | 2020-02-14 | 南方电网科学研究院有限责任公司 | Fault diagnosis method, system and equipment for direct current system |
CN111751643A (en) * | 2020-06-02 | 2020-10-09 | 武汉中元华电软件有限公司 | Train electrical diagnosis early warning recording system |
CN112101077A (en) * | 2019-06-18 | 2020-12-18 | 北京映翰通网络技术股份有限公司 | Method for identifying fault type of power distribution network |
CN112240964A (en) * | 2019-07-16 | 2021-01-19 | 北京映翰通网络技术股份有限公司 | Method for identifying fault type of power distribution network |
CN112305387A (en) * | 2020-10-31 | 2021-02-02 | 贵州电网有限责任公司 | Ground insulation detection and diagnosis system |
CN112816831A (en) * | 2021-03-18 | 2021-05-18 | 华北电力大学(保定) | Single-phase earth fault positioning method for collecting wire of wind power plant |
CN113848508A (en) * | 2021-09-10 | 2021-12-28 | 南方电网调峰调频发电有限公司 | Battery energy storage system alternating current side single-point grounding fault positioning method through leakage current |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140058698A1 (en) * | 2012-08-27 | 2014-02-27 | Futurewei Technologies, Inc. | System and Method for Hybrid Board-Level Diagnostics |
CN104090244A (en) * | 2014-07-16 | 2014-10-08 | 广西大学 | Train power supply device detection system |
CN105093066A (en) * | 2015-08-12 | 2015-11-25 | 华北电力大学 | Line fault judgment method based on wavelet analysis and support vector machine |
-
2016
- 2016-11-28 CN CN201611064578.9A patent/CN106526400B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140058698A1 (en) * | 2012-08-27 | 2014-02-27 | Futurewei Technologies, Inc. | System and Method for Hybrid Board-Level Diagnostics |
CN104090244A (en) * | 2014-07-16 | 2014-10-08 | 广西大学 | Train power supply device detection system |
CN105093066A (en) * | 2015-08-12 | 2015-11-25 | 华北电力大学 | Line fault judgment method based on wavelet analysis and support vector machine |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109212373A (en) * | 2017-06-29 | 2019-01-15 | 株洲中车时代电气股份有限公司 | The grounding fault diagnosis and insulation detecting method and device of power supply system of train |
CN109212373B (en) * | 2017-06-29 | 2021-02-12 | 株洲中车时代电气股份有限公司 | Ground fault diagnosis and insulation detection method and device for train power supply system |
CN109239511A (en) * | 2017-07-10 | 2019-01-18 | 比亚迪股份有限公司 | Train, power supply system of train and its detection of electrical leakage positioning device, method |
CN109228871B (en) * | 2017-07-10 | 2020-08-07 | 比亚迪股份有限公司 | Train power supply system, leakage detection and recovery device and method thereof, and train |
CN109228871A (en) * | 2017-07-10 | 2019-01-18 | 比亚迪股份有限公司 | Power supply system of train and its detection of electrical leakage recovery device, method and train |
CN109406923A (en) * | 2017-08-17 | 2019-03-01 | 株洲中车时代电气股份有限公司 | A kind of power supply system of train earth leakage failure prediction method and device |
CN109406923B (en) * | 2017-08-17 | 2021-06-25 | 株洲中车时代电气股份有限公司 | Method and device for predicting ground leakage fault of train power supply system |
CN108583629A (en) * | 2018-05-04 | 2018-09-28 | 兰州容大信息科技有限公司 | A kind of railcar business fault handling method |
CN109142946A (en) * | 2018-06-29 | 2019-01-04 | 东华大学 | Transformer fault detection method based on ant group algorithm optimization random forest |
CN109203995A (en) * | 2018-07-10 | 2019-01-15 | 中南大学 | A kind of trailer system major loop ground fault diagnosis method and system |
CN109599827A (en) * | 2018-11-21 | 2019-04-09 | 宁波恒晨电力建设有限公司 | Power network neutral point dynamic earth fault method and system based on big data |
CN109765878A (en) * | 2018-12-24 | 2019-05-17 | 上海大郡动力控制技术有限公司 | The aided analysis method of new-energy automobile CAN bus network failure |
CN112101077A (en) * | 2019-06-18 | 2020-12-18 | 北京映翰通网络技术股份有限公司 | Method for identifying fault type of power distribution network |
CN112240964A (en) * | 2019-07-16 | 2021-01-19 | 北京映翰通网络技术股份有限公司 | Method for identifying fault type of power distribution network |
CN112240964B (en) * | 2019-07-16 | 2023-06-20 | 北京映翰通网络技术股份有限公司 | Method for identifying fault type of power distribution network |
CN110794243A (en) * | 2019-11-14 | 2020-02-14 | 南方电网科学研究院有限责任公司 | Fault diagnosis method, system and equipment for direct current system |
CN111751643A (en) * | 2020-06-02 | 2020-10-09 | 武汉中元华电软件有限公司 | Train electrical diagnosis early warning recording system |
CN111751643B (en) * | 2020-06-02 | 2023-04-07 | 武汉中元华电软件有限公司 | Train electrical diagnosis early warning recording system |
CN112305387A (en) * | 2020-10-31 | 2021-02-02 | 贵州电网有限责任公司 | Ground insulation detection and diagnosis system |
CN112816831A (en) * | 2021-03-18 | 2021-05-18 | 华北电力大学(保定) | Single-phase earth fault positioning method for collecting wire of wind power plant |
CN113848508A (en) * | 2021-09-10 | 2021-12-28 | 南方电网调峰调频发电有限公司 | Battery energy storage system alternating current side single-point grounding fault positioning method through leakage current |
Also Published As
Publication number | Publication date |
---|---|
CN106526400B (en) | 2019-11-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106526400B (en) | The earth fault diagnostic method and device of DC600V power supply system of train | |
CN107037313B (en) | The method for establishing deformation of transformer winding failure Yu frequency sweep impedance characteristic corresponding relationship | |
CN102680825B (en) | Interference source identification method in system-grade electromagnetic compatibility fault diagnosis | |
CN108062571A (en) | Diagnosing failure of photovoltaic array method based on differential evolution random forest grader | |
CN102944416B (en) | Multi-sensor signal fusion technology-based fault diagnosis method for wind turbine blades | |
CN103116090A (en) | Three-phrase pulse-width modulation (PWM) rectifier fault diagnosis method based on wavelet packet analysis and support vector machine | |
CN106841949B (en) | Method and device for monitoring stator insulation of three-phase asynchronous alternating current motor on line | |
CN103438797B (en) | Deformation of transformer winding online test method and system | |
CN103969579A (en) | Asymmetric fault diagnosis method and system for stator of permanent magnet synchronous motor | |
CN103064008A (en) | Nolinear analog circuit soft fault diagnostic method based on Hilbert-huang transform | |
CN109375010A (en) | Winding in Power Transformer method for diagnosing faults based on vibration distribution characteristic model | |
CN112098889B (en) | Single-phase earth fault positioning method based on neural network and feature matrix | |
CN110174585B (en) | Method for identifying open circuit fault of high-voltage capacitor of double-tuned alternating current filter | |
CN112200694A (en) | Dominant instability mode identification model construction and application method based on graph neural network | |
CN105738772A (en) | Compulsory disturbance source positioning method based on power and frequency fluctuation phase | |
CN104236933B (en) | A kind of potential faults method for early warning for train traction system | |
Li et al. | A fault pattern and convolutional neural network based single-phase earth fault identification method for distribution network | |
CN114460439A (en) | Digital integrated circuit test system | |
CN114021671B (en) | Active power distribution network online fault processing method and system | |
CN113702767B (en) | Island direct-current microgrid fault diagnosis method based on wavelet sliding window energy | |
CN104090244A (en) | Train power supply device detection system | |
CN103823158A (en) | Resonant earthed system fault line selection method with invariant moments | |
CN105279315A (en) | Related analysis and Mahalanobis distance based transformer online monitoring information aggregation analysis method | |
CN108875796A (en) | Diagnosing failure of photovoltaic array method based on linear discriminant analysis and support vector machines | |
CN109781442B (en) | Detection method for crack fault of bogie of maglev train |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |