CN106970300A - Small Electric Current Earthing And Routing Device and its control method based on neural network processor - Google Patents
Small Electric Current Earthing And Routing Device and its control method based on neural network processor Download PDFInfo
- Publication number
- CN106970300A CN106970300A CN201710138838.0A CN201710138838A CN106970300A CN 106970300 A CN106970300 A CN 106970300A CN 201710138838 A CN201710138838 A CN 201710138838A CN 106970300 A CN106970300 A CN 106970300A
- Authority
- CN
- China
- Prior art keywords
- processor
- stages
- neural network
- data
- network processor
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 20
- 230000007935 neutral effect Effects 0.000 claims description 18
- 238000005070 sampling Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 4
- 230000001360 synchronised effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000010187 selection method Methods 0.000 claims description 3
- 230000001052 transient effect Effects 0.000 claims description 3
- 230000004888 barrier function Effects 0.000 claims description 2
- 238000013461 design Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
- Locating Faults (AREA)
Abstract
The present invention is a kind of Small Electric Current Earthing And Routing Device based on neural network processor, it is characterized in, including neural network processor, fpga chip, at least one memory, at least one high-speed ADC, DSP Processor, arm processor, local man-machine interface and network interface, described fpga chip respectively with neural network processor, at least one memory, at least one high-speed ADC, DSP Processor is connected, described DSP Processor is connected with arm processor, described arm processor respectively with neural network processor, local man-machine interface and network interface connection.And its control method is provided.With rational in infrastructure, manipulate conveniently, failure line selection high accuracy for examination.
Description
Technical field
The invention belongs to field of relay protection in power, and in particular to a kind of low current based on neural network processor
Line-selected earthing device and its control method.
Background technology
6~66kV power distribution networks of China are mainly using isolated neutral and through grounding through arc mode, and minority is used
Through High Resistance Grounding Mode, small current neutral grounding system is belonged to.The probability broken down in power distribution network is of a relatively high, the overwhelming majority
It is singlephase earth fault.Short-circuit loop, three voltages between phases are not formed when occurring singlephase earth fault due to small current neutral grounding system
Still symmetrical, China's electric power code regulation, small current neutral grounding system can continue to run with 2 hours with singlephase earth fault, so
The continuation and reliability of power supply can be improved.But actually most power supply departments are required at 30 points to avoid fault spread
Excision failure in clock.Because the power distribution network majority of China is by the way of neutral by arc extinction coil grounding so that search fault wire
Road turns into a problem.For a long time, domestic and international relay protection worker to single-phase grounded malfunction in grounded system of low current route selection,
Positioning has carried out untiringly research and discovery.To route selection problem, it is proposed that a variety of location designs based on different faults feature, and
And corresponding line selection apparatus is have developed, but actual motion effect is unsatisfactory.According to live incomplete statistics, current route selection
Device route selection accuracy rate is universal below 60%, so that most field operators are still searched with traditional order broaching tool lock method
Faulty line, greatly reduces power supply reliability.Therefore, the single-phase earth fault detecting method of power distribution network, tool are further studied
There is very strong theoretical and practical significance.
The content of the invention
In order to solve the above-mentioned technical problem, it is an object of the present invention to provide a kind of rational in infrastructure, manipulate conveniently, failure
The high Small Electric Current Earthing And Routing Device based on neural network processor of route selection accuracy rate, and its control method is provided.
Realizing the technical scheme that foregoing invention purpose is used is, a kind of small current neutral grounding choosing based on neural network processor
Line apparatus, it is characterized in that, it include neural network processor, fpga chip, at least one memory, at least one high-speed ADC,
DSP Processor, arm processor, local man-machine interface and network interface, described fpga chip respectively with Processing with Neural Network
Device, at least one memory, at least one high-speed ADC, DSP Processor connection, described DSP Processor connect with arm processor
Connect, described arm processor is connected with neural network processor, local man-machine interface and network interface respectively.
A kind of control method of the Small Electric Current Earthing And Routing Device based on neural network processor, it comprises the following steps：
1) the S1 stages of sampled data：Three-phase voltage, the three-phase current of each power transmission sequence of synchronized sampling, and by sampled data according to
Queue mode sends into memory, into the S2 stages；
2) whether there are the S2 stages of residual voltage：If DSP Processor calculates system residual voltage Sudden Changing Rate more than threshold value,
Then think that earth fault occurs for system, record fault moment, into the S3 stages, otherwise return to the S1 stages；
3) the S3 stages of 20 fundamental wave data of sampling are continued：High-speed ADC is made to continue the data of 20 primitive period durations of sampling, with
The data of 20 primitive period durations are together as the data of small current line selection algorithm before fault moment, into the S4 stages；
4) the S4 stages of small current earthing wire-selecting algorithm are run：DSP Processor enters to data before and after the failure in memory first
The necessary processing of row, then notifies neural network processor to start, and carries out parallel computation and obtains faulty line output valve, while DSP
Processor runs transient zero-sequence component amplitude method, zero sequence quintuple harmonics electric current route selection method, wavelet analysis method failure judgement line successively
Road, into the S5 stages；
5) the S5 stages of route selection result are exported：Arm processor reads this event from DSP Processor and neural network processor
The route selection result of barrier, and export on the touchscreen, while uploading to local side server.Three kinds that if DSP Processor is run
There are two kinds of obtained route selection results consistent in algorithm, then only show a kind of this result, three results are otherwise shown, for operations staff
With reference to early stage device puts into operation, the route selection result that neural network processor is provided runs people without what reference value
Member can ignore it, into the S6 stages；
6) the route selection result whether correct S6 stages：Operations staff is if it find that actual earth fault line is given with DSP Processor
The deduction gone out is not inconsistent, then to becoming in local man-machine interface or in local side remote manual input fault circuit number and trouble point
Line length at the line outlet of power station, subsequently into the S7 stages；
7) the S7 stages of neutral net are trained：Neural network processor using correct faulty line data as desired throughput,
Failure post-sampling data continue to train neutral net as input quantity, are then stored in the neural network parameter trained and store
In device, terminate whole flow process.
The Small Electric Current Earthing And Routing Device based on neural network processor and its control method that the present invention is provided are had
Advantage be embodied in：
1）Rational in infrastructure, manipulation is convenient；
2）Many algorithms judge ground path jointly, can improve small current earthing wire-selecting accuracy rate；
3）Pass through the continual training to neutral net, it is ensured that its small current earthing wire-selecting accuracy rate can be increasingly
It is high.
Brief description of the drawings
The signal of the Small Electric Current Earthing And Routing Device attachment structure based on neural network processor that Fig. 1 provides for the present invention
Figure；
The control method flow chart for the Small Electric Current Earthing And Routing Device based on neural network processor that Fig. 2 provides for the present invention.
Embodiment
The small current neutral grounding based on neural network processor provided below in conjunction with the accompanying drawings with specific embodiment the present invention
Line selection apparatus and its control method are described in detail.
As shown in figure 1, the Small Electric Current Earthing And Routing Device based on neural network processor that the present invention is provided, including nerve
Network processing unit 1, fpga chip 2, at least one memory 3, at least one high-speed ADC 4, DSP Processor 5, arm processor 6,
Local man-machine interface 7 and network interface 8.Described fpga chip 2 respectively with neural network processor 1, at least one memory
3rd, at least one high-speed ADC 4, DSP Processor 5 are connected, and described DSP Processor 5 is connected with arm processor 6, described ARM
Processor 6 is connected with neural network processor 1, local man-machine interface 7 and network interface 8 respectively.
Described neural network processor 1 uses the embedded neural network processor chip " star of Vimicro company
Light intelligence one ", described memory 3 includes multiple Large Copacity dual port RAMs and a high capability flash chip, and RAM is used to deposit
Sampled data is stored up, Flash is used to store neural network parameter, and high-speed ADC 4 is responsible for the three-phase electricity of each feeder line of synchronized sampling
Pressure, three-phase electricity flow valuve, its number are more than circuit number so as to following enlarging, fpga chip 2 be responsible for neural network processor 1,
High speed data transfer passage is built between memory 3, DSP Processor 5 and high-speed ADC 4, data transmission efficiency is improved.DSP processing
Device 5 is responsible for three kinds of conventional small current earthing wire-selecting methods, arm processor 6 using 2000 serial 32 bit processors of TI companies
Embedded Android operation system, the local man-machine interface 7 and 100 m ethernet network interface 8 for being responsible for 8 cun of touch-screens compositions of processing is carried
The data cube computation task gone out.
As shown in Fig. 2 the manipulation side for the Small Electric Current Earthing And Routing Device based on neural network processor that the present invention is provided
Method, including the following steps performed in order：
1) the S1 stages of sampled data：High-speed ADC 4 is high by the sample frequencys of 6000 times per second under the control of fpga chip 2
Three-phase voltage, the three-phase current of fast each power transmission sequence of synchronized sampling, and sampled data is sent into memory 3 according to queue mode,
Into the S2 stages；
2) whether there are the S2 stages of residual voltage：The processor of DSP Processor 5 is appointed from memory 3 takes newest the three of three-line
Phase voltage data, calculate residual voltage, if the residual voltage Sudden Changing Rate of two of which circuit is more than threshold value, then it is assumed that system
Generation earth fault, records fault moment, into the S3 stages, otherwise returns to the S1 stages；
3) the S3 stages of 20 fundamental wave data of sampling are continued：DSP Processor 5 is waited for, and waits high-speed ADC 4 to continue to adopt
20 fundamental wave durations of sample（I.e. 0.4 second）Data, totally 2400 groups of data, then before trace back 2400 groups of numbers of 20 fundamental waves before fault moment
According to by the address locking in memory 3 where this 4800 groups of data, as the data of small current line selection algorithm, into the S4 stages；
4) the S4 stages of small current earthing wire-selecting algorithm are run：DSP Processor 5 is first to data before and after the failure in memory 3
Necessary processing is carried out, zero-sequence current, residual voltage, various phase angles etc. is calculated, the fixing address block of memory 3 is stored in.At DSP
Reason device 5 notifies neural network processor 1 to start by arm processor 6, builds distributed radial basis function neural network, from depositing
The data that needs are extracted in reservoir 3 are inputted as neutral net, are carried out parallel computation and are obtained faulty line output valve.DSP processing
Device 5 runs transient zero-sequence component amplitude method, zero sequence quintuple harmonics electric current route selection method, wavelet analysis method failure judgement circuit successively,
Into the S5 stages；
5) the S5 stages of route selection result are exported：Arm processor 6 reads this from DSP Processor 5 and neural network processor 1
The route selection result of failure, and exported on the touch-screen of local man-machine interface 7, taken while uploading to local side by network interface 8
Business device.If there are two kinds of obtained route selection results consistent in three kinds of algorithms that DSP Processor 5 is run, only show that this is a kind of
As a result, three results are otherwise shown, are referred to for operations staff.Early stage device puts into operation, neural network processor 1 is provided
Route selection result without what reference value, operations staff can ignore it, into the S6 stages；
6) the route selection result whether correct S6 stages：Operations staff is if it find that actual earth fault line is given with DSP Processor 5
The deduction gone out is not inconsistent, then is arrived in local man-machine interface 7 or in local side remote manual input fault circuit number and trouble point
The line length in substation line exit, subsequently into the S7 stages；
7) the S7 stages of neutral net are trained：Neural network processor 1 using correct faulty line data as desired throughput,
Failure post-sampling data continue to train neutral net as input quantity, are then stored in the neural network parameter trained and store
In the Flash of device 3, terminate whole flow process.
The principle of the present invention is to introduce neural network algorithm in Single-phase Earth-fault Selection in Distribution Systems flow, has device
Have one can constantly deep learning, the brain constantly upgraded, the accuracy rate of its failure line selection will more and more higher.But, god
Parallel computation amount through network algorithm is big, and data volume is big, very high to hardware requirement, so neural network algorithm is all difficult all the time
In being applied in embedded system.There is artificial intelligence depth present invention employs the recent design research and development of Vimicro company of China
First embedded neural network processor chip NPU " starlight intelligence one " of the China of learning functionality is selected as small current neutral grounding
One core processor of line apparatus, neural network algorithm is incorporated into the practical application of small current earthing wire-selecting.Because dress
Put site of deployment and do not allow for us and go the neutral net of prolonged, continual trainer, without any return,
So also needing to add some conventional algorithms for identifying ground wire in the auxiliary nervous network algorithm at initial stage of device work
Road, improves constantly accuracy rate.
Claims (2)
1. a kind of Small Electric Current Earthing And Routing Device based on neural network processor, it is characterized in that, it includes Processing with Neural Network
Device, fpga chip, at least one memory, at least one high-speed ADC, DSP Processor, arm processor, local man-machine interface and
Network interface, described fpga chip respectively with neural network processor, at least one memory, at least one high-speed ADC,
DSP Processor is connected, and described DSP Processor is connected with arm processor, described arm processor respectively with neutral net
Manage device, local man-machine interface and network interface connection.
2. the Small Electric Current Earthing And Routing Device according to claim 1 based on neural network processor, it is characterized in that, it
Control method comprises the following steps：
1) the S1 stages of sampled data：Three-phase voltage, the three-phase current of each power transmission sequence of synchronized sampling, and by sampled data according to
Queue mode sends into memory, into the S2 stages；
2) whether there are the S2 stages of residual voltage：If DSP Processor calculates system residual voltage Sudden Changing Rate more than threshold value,
Then think that earth fault occurs for system, record fault moment, into the S3 stages, otherwise return to the S1 stages；
3) the S3 stages of 20 fundamental wave data of sampling are continued：High-speed ADC is made to continue the data of 20 primitive period durations of sampling,
Data with 20 primitive period durations before fault moment are together as the data of small current line selection algorithm, into the S4 stages；
4) the S4 stages of small current earthing wire-selecting algorithm are run：DSP Processor enters to data before and after the failure in memory first
The necessary processing of row, then notifies neural network processor to start, and carries out parallel computation and obtains faulty line output valve, while DSP
Processor runs transient zero-sequence component amplitude method, zero sequence quintuple harmonics electric current route selection method, wavelet analysis method failure judgement line successively
Road, into the S5 stages；
5) the S5 stages of route selection result are exported：Arm processor reads this event from DSP Processor and neural network processor
The route selection result of barrier, and export on the touchscreen, while uploading to local side server；
If there are two kinds of obtained route selection results consistent in three kinds of algorithms that DSP Processor is run, this one kind knot is only shown
Really, three results are otherwise shown, are referred to for operations staff, early stage device puts into operation, what neural network processor was provided
Route selection result is without what reference value, and operations staff can ignore it, into the S6 stages；
6) the route selection result whether correct S6 stages：Operations staff is if it find that actual earth fault line is given with DSP Processor
The deduction gone out is not inconsistent, then to becoming in local man-machine interface or in local side remote manual input fault circuit number and trouble point
Line length at the line outlet of power station, subsequently into the S7 stages；
7) the S7 stages of neutral net are trained：Neural network processor using correct faulty line data as desired throughput,
Failure post-sampling data continue to train neutral net as input quantity, are then stored in the neural network parameter trained and store
In device, terminate whole flow process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710138838.0A CN106970300B (en) | 2017-03-09 | 2017-03-09 | Small Electric Current Earthing And Routing Device and its control method based on neural network processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710138838.0A CN106970300B (en) | 2017-03-09 | 2017-03-09 | Small Electric Current Earthing And Routing Device and its control method based on neural network processor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106970300A true CN106970300A (en) | 2017-07-21 |
| CN106970300B CN106970300B (en) | 2019-06-28 |
Family
ID=59329419
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710138838.0A Expired - Fee Related CN106970300B (en) | 2017-03-09 | 2017-03-09 | Small Electric Current Earthing And Routing Device and its control method based on neural network processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106970300B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113542043A (en) * | 2020-04-14 | 2021-10-22 | 中兴通讯股份有限公司 | Data sampling method, device, equipment and medium of network equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011223801A (en) * | 2010-04-13 | 2011-11-04 | Chugoku Electric Power Co Inc:The | Dc grounding position searching method, grounding current supply device and dc ground monitoring system |
| CN102608499A (en) * | 2012-04-03 | 2012-07-25 | 云南电网公司红河供电局 | Low-current line selection device and control method for inhibiting unbalanced current by way of differential filtration |
| CN105759167A (en) * | 2016-01-28 | 2016-07-13 | 江苏省电力公司南京供电公司 | Wavelet neural network-based distribution network single-phase short circuit line selection method |
| CN105785149A (en) * | 2014-12-16 | 2016-07-20 | 东北电力大学 | Network tester for undercurrent grounding line selection apparatus of smart substation and network testing method therefor |
-
2017
- 2017-03-09 CN CN201710138838.0A patent/CN106970300B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011223801A (en) * | 2010-04-13 | 2011-11-04 | Chugoku Electric Power Co Inc:The | Dc grounding position searching method, grounding current supply device and dc ground monitoring system |
| CN102608499A (en) * | 2012-04-03 | 2012-07-25 | 云南电网公司红河供电局 | Low-current line selection device and control method for inhibiting unbalanced current by way of differential filtration |
| CN105785149A (en) * | 2014-12-16 | 2016-07-20 | 东北电力大学 | Network tester for undercurrent grounding line selection apparatus of smart substation and network testing method therefor |
| CN105759167A (en) * | 2016-01-28 | 2016-07-13 | 江苏省电力公司南京供电公司 | Wavelet neural network-based distribution network single-phase short circuit line selection method |
Non-Patent Citations (1)
| Title |
|---|
| 李荣明等: "基于神经网络的小电流接地系统综合智能选线", 《重庆大学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113542043A (en) * | 2020-04-14 | 2021-10-22 | 中兴通讯股份有限公司 | Data sampling method, device, equipment and medium of network equipment |
| CN113542043B (en) * | 2020-04-14 | 2024-06-07 | 中兴通讯股份有限公司 | Data sampling method, device, equipment and medium of network equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106970300B (en) | 2019-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kezunovic | Smart fault location for smart grids | |
| CN102508076B (en) | Fault diagnosis device and method based on multi-agent system and wavelet analysis | |
| CN207895024U (en) | A kind of medium voltage distribution network fault fast positioning device | |
| CN103904779B (en) | A kind of full station intelligent terminal analogue means of intelligent substation and using method thereof | |
| CN103424669B (en) | A kind of selection method utilizing fault feeder zero-sequence current matrix principal component analysis (PCA) first principal component | |
| CN107037322A (en) | Power distribution network low current grounding localization method based on steady state characteristic | |
| CN102944813B (en) | Resonant grounding system fault line selection method based on reactive power flow directions | |
| CN106199342A (en) | A kind of wire selection method for power distribution network single phase earthing failure | |
| CN104655987B (en) | A kind of transmission line lightning stroke point location new method based on OPGW polarization states | |
| CN102135591A (en) | Resonant grounding power grid single-phase ground fault db wavelet transient component line selection method | |
| CN104360227A (en) | Substation cable outlet fault monitoring method based on traveling wave method and transient basic frequency method | |
| CN106940411A (en) | A kind of single-phase ground fault line selecting method of small-electric current grounding system | |
| CN106646130B (en) | A kind of active distribution network Fault Locating Method compared based on current polarity and system | |
| CN101871988A (en) | System and method for positioning medium voltage distribution network using power line carrier communication | |
| CN102628914A (en) | Method for positioning grounding point of intelligent distribution line | |
| CN105759179A (en) | Zero-sequence-transient-analysis-based fault positioning system and method for low current grounding system | |
| CN104360226A (en) | Method for monitoring fault of cable outgoing lines of transformer substation on basis of current initial traveling wave polarity | |
| CN104375056A (en) | Substation cable outgoing line fault monitoring method based on voltage and current initial row waves | |
| CN111413643A (en) | Power distribution network single-phase earth fault identification method | |
| CN205670180U (en) | Based on zero sequence transient analysis low current neutral grounding system fault alignment system | |
| CN106569096B (en) | A kind of tuning on-line method of power distribution network single-phase fault | |
| Shiddieqy et al. | Power line transmission fault modeling and dataset generation for ai based automatic detection | |
| CN113156267B (en) | Power distribution network ground fault section selection method and system | |
| CN101943738A (en) | Single-phase grounding distance measuring method and distance measuring device | |
| CN106970300A (en) | Small Electric Current Earthing And Routing Device and its control method based on neural network processor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 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 | ||
| 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: 20190628 |