CN106093638A - A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation - Google Patents
A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation Download PDFInfo
- Publication number
- CN106093638A CN106093638A CN201610398935.9A CN201610398935A CN106093638A CN 106093638 A CN106093638 A CN 106093638A CN 201610398935 A CN201610398935 A CN 201610398935A CN 106093638 A CN106093638 A CN 106093638A
- Authority
- CN
- China
- Prior art keywords
- voltage
- pattern
- voltage drop
- whole network
- mean
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R19/00—Arrangements for measuring currents or voltages or for indicating presence or sign thereof
- G01R19/02—Measuring effective values, i.e. root-mean-square values
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R19/00—Arrangements for measuring currents or voltages or for indicating presence or sign thereof
- G01R19/165—Indicating that current or voltage is either above or below a predetermined value or within or outside a predetermined range of values
- G01R19/17—Indicating that current or voltage is either above or below a predetermined value or within or outside a predetermined range of values giving an indication of the number of times this occurs, i.e. multi-channel analysers
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The present invention relates to a kind of Voltage Drop root-mean-square value and fall the mode identification method that the frequency is estimated, its technical characterstic is: comprise the following steps: step 1, the whole network node voltage setting up different faults type fall pattern database；Step 2, structure monitoring node Voltage Drop pattern；Step 3, voltage sag type it is identified and classifies, falling pattern database according to the whole network node voltage that short trouble type selecting is corresponding；Step 4, the Voltage Drop root-mean-square value of the whole network node is estimated；Step 5, statistics obtain the Voltage Drop frequency estimated value under each Voltage Drop threshold value.The present invention estimates principle and implementation process is simple, pattern recognition precision is high and the calculating time is short.
Description
Technical field
The invention belongs to the Voltage Drop estimation technique field of electric power quality, especially a kind of Voltage Drop is equal
Root value and the mode identification method falling frequency estimation.
Background technology
The extensive application in power supply network such as computer technology, automatic technology, Power Electronic Technique and microelectric technique,
The Short Duration Power Quality Disturbances such as the quality of power supply to power supply network, especially Voltage Drop (Voltage Dip) propose higher wanting
Ask, make Voltage Drop become current power user, society and academia most concerned about the power quality problem with concern.
Estimate to be short-circuited in electrical network fault time the Voltage Drop root-mean-square value of bus nodes and special time in electricity
Pressure falls the frequency, is the premise solving Voltage Drop problem, and it is short that Voltage Drop is defined as supply side power-frequency voltage root-mean-square value
Suddenly the electrical energy power quality disturbance event then automatically recovered is declined in time.IEEE(Institute of Electrical and
Electronic Engineers, Institute of Electrical and Electric Engineers) Voltage Drop is defined as: rms voltage is down to volume
Determining the 10%-90% of voltage, the persistent period is 0.5 cycle short time disturbance phenomenon to 1min, in engineer applied, for convenience of amount
Change statistics, be down to rated voltage in various degree according to rms voltage, generally take 9 discrete Voltage Drop threshold value typical cases
Value: 0.9pu, 0.8pu, 0.7pu, 0.6pu, 0.5pu, 0.4pu, 0.3pu, 0.2pu, 0.1pu (pu representation unit rated voltage).
Method of estimation conventional in existing Voltage Drop estimation technique includes: actual measurement statistic law, Stochastic prediction method, state
The estimation technique etc..Separately below method of estimation conventional in existing Voltage Drop estimation technique is specifically described:
(1) actual measurement statistic law
Actual measurement statistic law is by occurring the frequency of Voltage Drop to carry out Voltage Drop estimation on each bar bus of statistics, this
Plant method of estimation relatively direct reliably, but must be when being estimated in bus nodes installation electric energy quality monitor device and carrying out long
Between monitor, its result is only applicable to the situation of electric network model and parameter determination.Voltage Drop has the biggest randomness,
If the method using quality of power supply generaI investigation, at synchronization, many bus nodes are carried out Voltage Drop monitoring simultaneously, can only
The average voltage of acquisition system falls situation, is difficult to hold the most meticulously the Voltage Drop situation at each node of system.As
Fruit uses Voltage Drop long-term fixed point monitoring method, and the length in monitoring cycle directly affects again the accuracy of analysis result.According to
Foreign literature is investigated, and all can trip the sensitivity load bus of 1 time because of Voltage Drop for every day, as to obtain 50% accurate
Degree need to observe 2 weeks；The accuracy obtaining 10% need to observe 1 year；And for being often only the non-sensibility of 1 trip accident of experience
Node, the accuracy that will obtain 50% need to observe 16 years；The accuracy obtaining 10% need to observe 400 years.The knot of fixed point monitoring
Fruit has reference value for the load frequently that trips because of Voltage Drop, and for general load bus, if the monitoring time is not
Long enough then has little significance, even if simultaneously to sensitivity load bus, the monitoring mode of fixed point monitoring can only obtain the most specific
The Voltage Drop situation of monitoring point, then cannot know during for non-monitored point.
If bus nodes all to the whole network are monitored, then need to be respectively mounted on each node monitoring instrument, thus
Considerably increase estimated cost, and owing to the generation of Voltage Drop has randomness, the length of monitoring time directly affects knot
The accuracy of fruit, therefore the method does not have predictability and replicability.
(2) the random estimation technique
Stochastic prediction method, based on stochastic modeling, carries out stochastic simulation by setting up probabilistic model to Voltage Drop, tool
Proactive and replicability.Wherein, fault position method is considered as more practical method.Stochastic prediction method is typically to utilize system
Historical statistical data such as components and parts probability of malfunction, fault type probability etc., estimate not monitor the voltage occurred on bus and fall
The frequency fallen.But element failure rate is very big owing to the impact of different factors changes every year, and these factors include weather conditions, dimension
Repair situation etc..Therefore the probability essence of random method makes this method be appropriate only for for a long time to estimate, obtain in statistical significance
Voltage Drop distribution probability, and within the time period that some is concrete, the estimation frequency of Voltage Drop obtains with actual monitoring
To the frequency may have very big gap.If additionally system is keeped in repair or extended, historical data originally also will be the most applicable,
Random method is caused to become unavailable.
(3) state estimate
Due to the restriction of measurement apparatus, the estimation of Voltage Drop has significant limitation.Therefore researcher is had to propose
State Estimation is applied in Voltage Drop estimation.Traditional state estimation is to utilize at sequence of threads, processes telemechanical in real time
Remote measurement that device is sent here and the communication information, thus draw sign power system practical structures and the reliable value of operation, make various mistake
The impact of difference and interference minimizes.Voltage Drop root-mean-square value estimation technique is not that Legacy Status estimation technique is in electric energy matter
The simple extension in amount field, but all can have a lot of changes at the theory of estimation technique, algorithm and research that some are relevant.
It is that the Voltage Drop frequency utilizing limited monitoring node to obtain is as known quantity that Voltage Drop root-mean-square value is estimated
Estimate, obtain the Voltage Drop frequency of non-monitored node, accurately obtain the Voltage Drop level of the whole network, close for sensitive load
Reason selects access point to provide decision-making foundation.It is said that in general, for the mathematic(al) representation such as formula (1) of state estimation:
H=MX+E (1)
Formula (1) is highly to owe fixed system of linear equations, and wherein, measurement H represents the monitoring instrument prison being arranged on bus
The rms voltage measured is less than or equal to the Voltage Drop quantity of set voltage threshold.Therefore, by the prison of monitoring point
Survey data and can directly form measurement H.Quantity of state X is state variable to be estimated, and X represents search time in actual electric network
Interior generation is the number of faults distribution situation in all part of paths on system line.The boundary of part of path is bent by rms voltage
Line determines with the intersection point of Voltage Drop threshold value.Calculation matrix M is used for contacting measurement and state variable.Generally measure noise vector
E is left in the basket.After considering the Integer constrained characteristic condition of line fault, by methods such as linear programmings, this state equation is solved to obtain shape
State amount X.Owing to equation height owes fixed, with or without array feasible solution.Existing processing method arbitrarily takes one group of feasible solution often
The actual distribution situation of indicator electric network fault, it is impossible to the truth of faults distribution.
Voltage Drop root-mean-square value method of estimation solution procedure is summarized as follows:
The first step: application formula H=MX, M are the calculation matrix of monitoring bus, according to Voltage Drop threshold value and line sectionalizing
Result determines that in M, the value of element is 0 or 1；If the Voltage Drop of the monitoring bus that the fault in line sectionalizing causes is mean square
Root is less than or equal to given Voltage Drop threshold value, and the element value in corresponding M is 1, is otherwise 0.In X is each part of path
The number of faults occurred, essence is exactly that the Voltage Drop quantity of the monitoring bus actual measurement according to monitoring point record estimates each
The number of faults occurred in line sectionalizing.
Second step: structural regime estimates model.Voltage Drop root-mean-square value estimates that equation is the underdetermined equations solved more, but
Real physical characteristics in view of state variable, it is necessary to consider following constraints:
1) no matter how line sectionalizing divides, the fault total amount in all line sectionalizings is occurred to be consistently equal to occur in institute
There is the fault total amount on circuit.It is to say, the shape that the line sectionalizing corresponding to one circuit of a voltage threshold gained comprises
The shape that the total amount of state variable should be comprised with the line sectionalizing corresponding to another Voltage Drop threshold value gained same circuit
The total amount of state variable is equal.
2) number of faults occurred on system neutral road can only take the integer more than or equal to 0.
3) number of faults occurred in state variable represents line sectionalizing, it is contemplated that some line sectionalizing is completely contained in
In All other routes segmentation, therefore the value of these line sectionalizing state variables is necessarily less than or equal to comprising the change of this line sectionalizing state
The value of amount.
Can obtain the constraints as shown in formula (2) in sum:
H=MX
Wherein Z represents that positive integer, t represent different Voltage Drop threshold values (0.1pu-0.9pu).A, b represent that part of path begins
End position (a, b belong between [0,1]) on whole piece circuit.
Apply existing linear programming function that this integer programming model can solve to obtain state variable X, i.e. line sectionalizing fault
Distribution situation.
3rd step: application equation below (3), the non-non-monitored node voltage that h is to be estimated falls frequency vector, M1It is right
Answering the calculation matrix of non-monitored node, X is the interior number of faults occurred of each part of path obtained.All non-monitored can be tried to achieve
The Voltage Drop frequency of node.Essence is to utilize the interior number of faults occurred of each part of path to estimate the voltage of non-monitored node
Fall the frequency.
H=M1X (3)
The Voltage Drop frequency that all nodes occur can be estimated by above step.But below it is only for one
Individual given Voltage Drop threshold value, to carry out Voltage Drop estimation under other Voltage Drop threshold conditions, needs again to set
Determine Voltage Drop threshold value, form the M calculation matrix under different Voltage Drop threshold value, reconstruct Voltage Drop root-mean-square value estimation side
Journey, and repeat above step.
The method of estimation of above-mentioned three kinds of Voltage Drop, has the disadvantage that and problem: during the required monitoring of (1) actual measurement statistic law
Between long, monitoring point quantity is many, and time cost and Financial cost are huge；(2) Stochastic prediction method is limited to components and parts fault rate etc. no
Determine factor, and be only applicable to the estimation of long time scale, and the Voltage Drop frequency that can not accurately be given in the concrete period is estimated
Meter result；(3) the state estimation equation of state estimate is highly to owe fixed, has multiple feasible solution, not bright from physical significance
Really which kind of solution is the optimal solution of closing to reality situation.(4) standing state estimates that equation and constraints thereof constitute complexity, and needs
Carrying out equation reconstruct according to different Voltage Drop threshold values, application linear programming for solution time consumption is big.(5) standing state estimation side
Method is only capable of whether decision node occurs Voltage Drop can not provide its concrete Voltage Drop root-mean-square value, lost a lot of weight
The Voltage Drop information wanted.
In sum, actual measurement statistic law needs to install monitoring instrument record for a long time on a large scale, therefore economy and time
Cost is all difficult for Practical Project and is accepted.Stochastic prediction method is to emulate theoretically according to power grid topology model and parameter
Measuring and calculating, has preferable extensibility and economy.But it is limited to the uncertain factors such as components and parts fault rate, and cannot be provided
Voltage Drop frequency estimated result in the body time period.And though the standing state estimation technique can provide the node voltage in special time
Falling frequency estimated value, but mathematical principle and model are complex, the highly underdetermined system of equations solves difficulty, different Voltage Drop thresholds
Need to rebuild state equation under value to calculate, and be only capable of obtaining the Voltage Drop frequency, lost node voltage root-mean-square
The important Voltage Drop information such as value.Therefore, Voltage Drop root-mean-square value and the estimation principle of the frequency and technology need further
Improve and optimize, in order to being applicable to practical engineering application.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, it is provided that a kind of reasonable in design, estimation principle and implementing
The Voltage Drop root-mean-square value that journey is simple, pattern recognition precision is high and the calculating time is short and the pattern recognition side falling frequency estimation
Method.
The present invention solves it and technical problem is that and take techniques below scheme to realize:
A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation, comprise the following steps:
Step 1, employing fault position method and short circuit calculation method, the Voltage Drop pattern caused according to different faults type, point
The whole network node voltage not setting up different faults type falls pattern database；
Step 2, according to monitoring node Voltage Drop root-mean-square value record data, build monitoring node Voltage Drop pattern；
Step 3, employing determine the direct criterion pair of voltage sag type according to monitoring node Voltage Drop root-mean-square value
Voltage sag type is identified and classifies, and falls mode data according to the whole network node voltage that short trouble type selecting is corresponding
Storehouse, and carry out pattern recognition in this whole network Voltage Drop pattern database；
Step 4, with the monitoring node Voltage Drop pattern of described step 2 for identifying feature, with identical or closest
Emulation the whole network node voltage fall pattern for identify target, with the pattern recognition of the rms voltage of monitoring point corresponding node
Trueness error is minimised as identification condition and carries out pattern recognition, find out with its closest to the whole network node voltage fall pattern,
Reduce the rms voltage of all the whole network nodes, thus complete the Voltage Drop root-mean-square value to the whole network node and estimate；
Step 5, setting voltage fall threshold value, fall according to the whole network node voltage under short troubles all in specific time period
The simulation estimate value of the Voltage Drop root-mean-square value of pattern, the voltage carried out one by one under statistics each Voltage Drop threshold value of acquisition falls
The frequency that falls estimated value.
And, the concrete steps of described step 1 include:
(1) according to topological structure and the model parameter of electrical network, application and trouble point method, arrange in whole power system network
There is singlephase earth fault, two-phase short-circuit fault, double earthfault and three phase short circuit fault in simulation optional position, trouble point
The type of four kinds of short troubles；
(2) according to Power System Shortcuts computational methods, in simulating grid, the anticipation short trouble of four kinds of fault types, passes through
Emulated computation method builds the whole network node voltage of four kinds of short trouble types respectively and falls the data base of pattern.
And, described step 3 determine directly sentencing of voltage sag type according to monitoring node Voltage Drop root-mean-square value
The concrete criterion of other method is:
(1) three phase short circuit fault: three-phase voltage root-mean-square value is below 0.9pu and every phase root-mean-square value is identical；
(2) singlephase earth fault: a certain phase voltage root-mean-square value is less than 0.9pu, and remaining is biphase higher than 1pu；
(3) two-phase short-circuit fault: certain phase voltage root-mean-square value is normal, fluctuates in the little scope of about 1pu, and remaining is biphase low
In 0.9pu；
(4) double earthfault: certain phase voltage root-mean-square value is apparently higher than 1pu, and remaining is biphase less than 0.9pu.
And, the concrete steps of the mode identification method of described step 4 include:
(1) monitoring node Voltage Drop root-mean-square value record data and the whole network node voltage are fallen corresponding node in pattern
Rms voltage ask for pattern recognition error precision value as the following formula and keep a record, preserve minimum pattern recognition error precision
Value and corresponding the whole network node voltage thereof fall the sequence number of pattern.
(2) the whole network node voltage is fallen all the whole network node voltages in pattern database and falls the above behaviour of pattern repetition
Make, finally show that all the whole network node voltages fall in pattern database, the whole network node electricity that pattern recognition trueness error is minimum
Pressure falls pattern, thus identifies and be best able to represent true the whole network node voltage and fall Voltage Drop root-mean-square value imitative of pattern
True estimated value；
MRIi=Σ (MSj-DSi,j)2/Mnumber (4)
In above formula, MRI represents that monitoring node Voltage Drop pattern MS falls the corresponding joint of pattern DS with the whole network node voltage
The pattern recognition trueness error of rms voltage of point, MnumberRepresent be triggered under certain fault monitoring instrument number,
I takes and falls all over the whole network node voltage that all of the whole network node voltage in pattern database falls pattern, j takes all over all prisons that are triggered
Survey the monitoring node numbering of instrument, MS represents that the rms voltage of monitoring node Voltage Drop pattern, DS represent at actual electricity
Actual the whole network node voltage that in net, monitoring node Voltage Drop pattern is corresponding falls the rms voltage of pattern.
Advantages of the present invention and good effect be:
1, actual measurement statistic law and Stochastic prediction method are effectively combined by the present invention, according to the most limited monitoring node
The actual measurement Voltage Drop data obtained, Voltage Drop state and the frequency thereof to non-monitored point are estimated effectively.Due to reality
Survey data and can react the concrete practical situation of Voltage Drop in this period, can be had by Stochastic prediction model and short circuit calculation method
Spread to effect the Voltage Drop state of non-monitored point.Utilize the mode identification method of the present invention, look for according to limited Monitoring Data
Go out the system fault condition of optimal coupling therewith, thus realize obtaining all node voltages of the whole network by the monitoring of minority node and fall
Fall state, i.e. completes the state estimation to the whole network node, and the Voltage Drop threshold value according to arbitrarily setting can add up to obtain each threshold
The Voltage Drop frequency under Zhi.The present invention's it is crucial that introduce monitoring node Voltage Drop pattern and fault mode Land use models
Recognition methods obtains the fault mode pressed close to most with physical fault situation thus obtains node voltage and fall root-mean-square value and the frequency.
The estimation principle of the present invention and process are simple, it is not necessary to complicated solves software, calculates the time short.
2, the present invention effectively combine measurement method and Stochastic prediction method grid voltage sags root-mean-square value and the frequency estimate
The advantage of meter mode identification method, combines actual measurement statistic law and can know Voltage Drop information and Stochastic prediction in specific time period
Method, has the advantage that replicability is strong, can effectively reduce the monitoring time and not limited by factors such as components and parts fault rates, is more conducive to
Practical engineering application.
3, the principle of the invention and implementation process are simple, it is to avoid standing state method of estimation numerous and diverse.The present invention proposes
Monitoring node Voltage Drop pattern and the whole network node voltage fall the related notions such as pattern, and the thought of pattern recognition is introduced voltage
Fall estimation and give specific implementation method.Relative to Legacy Status method of estimation, the present invention is without solving the higher-dimension of complexity
The underdetermined system of equations, it is not necessary to consider complicated physical constraint condition, be that condition can recognize that by minimal mode accuracy of identification error
The grid nodes Voltage Drop state optimization solution of closing to reality situation.Can greatly reduce monitoring and calculating time, raising is estimated
Meter efficiency.(traditional method application integer programming method solving state estimates that equation needs tens of hours, and the pattern recognition of the present invention
Method completes Voltage Drop root-mean-square value and the Voltage Drop frequency is estimated only to need the several seconds).
4, the present invention falls pattern database carry out mould by setting up the whole network node voltage respectively for different faults type
Formula identification, is effectively improved pattern recognition precision.The Voltage Drop that the present invention causes according to different faults type, establishes four respectively
The whole network node voltage planting different faults type falls pattern database.Use and determine that voltage falls according to Voltage Drop root-mean-square value
The direct criterion of the type that falls, carried out type judgement to Voltage Drop before pattern recognition, further according to the difference of fault type,
Fall at corresponding the whole network node voltage and pattern database carries out pattern recognition, comprise combining of all fault types compared to setting up
Close the whole network node voltage to fall pattern database and can efficiently reduce identification error, improve pattern recognition precision (precision 90% with
On).
5, the present invention only need to set according to 0.9pu Voltage Drop threshold value and configuration monitoring point, can estimate all nodes simultaneously
Rms voltage and the frequency, required monitoring point is few, economy is high.The present invention only needs to carry out by 0.9pu Voltage Drop threshold value
Monitoring, it is to avoid traditional method need to estimate the trouble of equations according to different threshold value reconstituted state.And all nodes tool can be given
The rms voltage estimated value of body, and it is not only the Voltage Drop frequency, Voltage Drop information can be retained as much as possible.Electricity
It is to install limited monitoring instrument to obtain effective monitoring information that pressure falls the premise that root-mean-square value is estimated and the frequency is estimated.Tradition shape
The Voltage Drop threshold value that the state estimation technique is estimated is different, and required monitoring point quantity variance is huge, particularly estimates Low Dropout threshold
During value, required monitoring point is many, and monitoring cost is huge.And the mode identification method of the present invention only needs by the configuration monitoring of 0.9pu threshold value
Point, required monitoring point quantity is few and allocation plan unique (required monitoring point quantity be traditional method 20% within), facilitate work
Cheng Shiyong and can greatly reduce Financial cost.
Accompanying drawing explanation
Fig. 1 is the grid voltage sags pattern recognition schematic flow sheet of the present invention；
Fig. 2 is standard test system (the IEEE-14 node power distribution network) schematic diagram of the specific embodiment of the invention.
Detailed description of the invention
Below in conjunction with accompanying drawing, the embodiment of the present invention is described in further detail:
When the factors such as power system operating mode, fault type, abort situation all determine, breaking down in electrical network, it is intrinsic
Fault mode.Result from the region of the Voltage Drop in electrical network and the degree of depth the most all determines that, referred to as electrical network inherent voltage
Fall pattern.Obviously there is inevitable one-to-one relationship between electrical network is intrinsic fault mode and Voltage Drop pattern.According to
Under a certain fault mode, the Voltage Drop root-mean-square value record data of monitoring node, application model recognition methods, find out
Fall pattern for close the whole network node voltage, just can reduce the rms voltage of all nodes, thus complete all non-
The Voltage Drop root-mean-square value of monitoring node is estimated.Here it is the Voltage Drop root-mean-square value of present invention proposition and the frequency estimate mould
The foundation of formula method of identification.
Obviously, the premise of the method is that the whole network part of nodes is optimized monitoring point configuration, and providing for pattern recognition must
The Monitoring Data wanted is as feature foundation.Owing to the optimal allocation of monitoring point is only used as background and the premise of the inventive method, and
The claimed technology contents of non-invention, the most only introduces the core requirement that monitoring point is configured by the present invention.Existing voltage
Fall the optimal allocation technology of monitoring point typically to meet Voltage Drop the whole network ornamental as core requirement.Voltage Drop the whole network can
Sight refers to: in electrical network, optional position occurs any type of short trouble to cause Voltage Drop at least can be supervised by 1 monitoring instrument
Measuring, when thus guarantee is short-circuited, fault causes Voltage Drop, at least 1 monitoring instrument has record data, it is provided that
To mode identification procedure as feature foundation.When putting into practice the mode identification method of the present invention, as long as utilizing existing monitoring point
Optimal allocation technology ensures the whole network ornamental of Voltage Drop, and only configures monitoring point according to 0.9pu Voltage Drop threshold value, i.e.
The Voltage Drop root-mean-square value of the present invention and the mode identification method falling frequency estimation can be carried out.
The processing method of distributed power source (distributed generation, DG) in active distribution network: actively distribution
One big feature of net is i.e. the access having many DG.The DG type of main flow in present situation, including double-fed induction wind driven generator, combustion gas
Turbine, photovoltaic generation, fuel cell, and direct drive type permanent magnetism synchronous wind power generator, the DG in addition to double-fed induction wind driven generator
Electricity generation system structurally has certain similarity all to be needed by inverter grid-connected, collectively referred to here in as inversion grid connection type
DG.So DG can simplify double-fed induction wind driven generator and the big class of inversion grid connection type DG two being divided into direct grid-connected.Inverse type DG
External characteristics is equivalent to the model that controlled voltage source is connected with equivalent impedance, and wherein controlled voltage source is equivalent to inverter switching device function
Being multiplied with DC voltage-stabilizing capacitance voltage, impedance is inverter outlet to the equivalent impedance of public access point, including wave filter, isolation
Transformator, line impedance etc..The double-fed induction wind driven generator external characteristics entering direct grid-connected in terms of stator side is equivalent to time transient state
Electromotive force and subtranient impedance series connection.
In traditional concept, pattern is typically to be made up of with certain structure a series of element.And the voltage of the present invention falls
The mode identification method that falling root-mean-square value and falls frequency estimation is used to find element and structure is all most like with target pattern
Individuality.The present invention is from the physical characteristic of short circuit malfunction, according to grid failure state and Voltage Drop root-mean-square value
Between mapping relations, in conjunction with power system node voltage fallen and estimates and the engineering of Voltage Drop source module location is actual
Demand, proposes Voltage Drop and the mode identification method of frequency estimation.Wherein, Voltage Drop pattern, the optimizing of fault mode are known
It is crucial that other method and the whole network node voltage fall the foundation of pattern database.In order to be adapted to reality engineer applied, the present invention
First the related notion related to mode identification method in Voltage Drop estimates application is defined.
1, monitoring node Voltage Drop pattern
At a time, a certain position of the power system fault that is short-circuited causes some region in electrical network to occur voltage to fall
Falling, now in electrical network, the monitoring instrument of electric energy quality monitoring node may be triggered, these in monitoring instrument being triggered
Record data are element, are structural order according to node serial number size, may make up the monitoring data sequent of Voltage Drop root-mean-square value, claim
For monitoring node Voltage Drop pattern.
2, the whole network node voltage falls pattern
At a time, a certain position of the power system fault that is short-circuited causes some region in electrical network to occur voltage to fall
Fall, now in electrical network part of nodes generation Voltage Drop and there is not Voltage Drop in remaining node, the most all of node is the most right
Answer the rms voltage of reality, although owing to monitoring point these rms voltages limited are not all of actual measuring,
But it is the most in esse, with these in esse all node voltage root-mean-square values as element, according to node serial number size
For structural order, constitute the whole network node voltage and fall mode sequences.The whole network node voltage falls the target that pattern is pattern recognition
Pattern.
3, the whole network node voltage falls pattern database
According to Power System Shortcuts computational methods, in simulating grid, all possible anticipation short trouble, is counted by emulation
Calculation method obtains all possible the whole network node voltage and falls pattern database.Any possible fault due to analogue simulation,
Including fault type and abort situation etc., this data base contains all possible the whole network node voltage and falls pattern.From theory
On say, discrete trouble point is chosen the most more intensive, and the whole network node voltage of obtaining of emulation falls pattern and can preferably simulate entirely
The situation of net optional position fault.But incorporation engineering is actual, EQUILIBRIUM CALCULATION FOR PROCESS amount and computational accuracy, every circuit takes about 10
The error precision of discrete trouble point can meet requirement, can the most accurately simulate the whole network node electricity under the fault of optional position
Pressure falls pattern (can be according to required precision and network topology localization of faults number in concrete practice).
4, Voltage Drop pattern recognition
Monitoring node Voltage Drop pattern is constituted, as feature foundation, at the whole network node according to monitoring point record data
Voltage Drop pattern database carries out pattern recognition, searches out the whole network node voltage closest to actual truth and fall
The process of pattern.
A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation, as it is shown in figure 1, include following step
Rapid:
Step 1, employing fault position method and short circuit calculation method, the Voltage Drop pattern caused according to different faults type, point
The whole network node voltage not setting up different faults type falls pattern database.
The concrete steps of described step 1 include:
(1) according to topological structure and the model parameter of electrical network, application and trouble point method, arrange in whole power system network
There is singlephase earth fault, two-phase short-circuit fault, double earthfault and three phase short circuit fault in simulation optional position, trouble point
The type of four kinds of short troubles；
(2) according to Power System Shortcuts computational methods, in simulating grid, the anticipation short trouble of four kinds of fault types, passes through
Emulated computation method builds the whole network node voltage of four kinds of short trouble types respectively and falls the data base of pattern.
The concrete steps of (2nd) step of described step 1 include:
1. to a certain fault type, every circuit is uniformly arranged K trouble point (the concrete value foundation of K and precision
Require and system scale), the rms voltage of all N number of nodes is calculated by power system existing short circuit calculation method, with
All node voltage root-mean-square values are element, are structural order according to node serial number size, the equal structure in trouble point of each simulation
The whole network node voltage is become to fall mode sequences.Therefore it is 1*N dimension that each the whole network node voltage falls pattern, a circuit K
Individual trouble point can obtain the whole network node voltage of K 1*N dimension and fall mode sequences.
2. 1. the whole network L bar circuit being repeated step, the whole network node voltage that can obtain L*K 1*N dimension falls mode sequences.Should
L*K the whole network node voltage falls the whole network node voltage that mode sequences i.e. may make up under a certain fault type and falls pattern count
According to storehouse.
3. for four kinds of different fault types (singlephase earth fault, two-phase short-circuit fault, double earthfault, three-phases
Short trouble), respectively repeat steps 1.-step 2., the forecast failure of four kinds of fault types of simulation, respectively constitute four the whole networks joint
Point voltage falls pattern database.Four the whole network node voltages fall pattern database and all comprise the whole network node of L*K 1*N dimension
Voltage Drop mode sequences, therefore comprise four kinds of the whole network node voltages falling the integrated database size altogether of pattern is 4*L*K
The whole network node voltage of 1*N dimension falls mode sequences.
Step 2, according to monitoring node Voltage Drop root-mean-square value record data, build monitoring node Voltage Drop pattern.
The fault that is short-circuited in a certain moment power system result in Voltage Drop event, meets at monitoring point allocation plan
Under conditions of the faulty all standing monitoring of the whole network institute, in the whole network, all monitoring points at least 1 monitoring instrument is triggered, or many
Platform monitoring instrument all has monitoring record.Record case to all monitoring points processes, for less than 0.9pu Voltage Drop threshold
The rms voltage of value can trigger monitoring instrument and produce record data, and some monitoring node is higher than due to rms voltage
0.9pu thus do not have trigger voltage to fall monitoring instrument, no record data in device, then represent monitoring instrument by " # " non-registered
Situation, constitutes monitoring node voltage according to the rms voltage of monitoring node number order in electrical network and each monitoring point
Fall pattern.Obvious, in monitoring node Voltage Drop pattern under this failure condition of " # " positional representation, corresponding monitoring node without
Record, its implicit information is that this node voltage root-mean-square value is higher than Voltage Drop threshold value 0.9pu set.
Step 3, employing determine the direct criterion pair of voltage sag type according to monitoring node Voltage Drop root-mean-square value
Voltage sag type is identified and classifies, and falls mode data according to the whole network node voltage that short trouble type selecting is corresponding
Storehouse, and carry out pattern recognition in this whole network Voltage Drop pattern database.
Owing to power system internal short-circuit fault includes various faults type, unified comprise all fault types if setting up
The whole network node voltage falls pattern database and carries out pattern recognition, necessarily occurs identifying that increasing of deviation causes error precision to drop
Low.Therefore the present invention is directed to various fault type set up the whole network node voltage respectively and fall pattern database, for concrete list
Secondary short trouble, is first identified voltage sag type with voltage sag type direct criterion and classifies, further according to fault
Type is fallen at corresponding the whole network node voltage and is carried out pattern recognition in pattern database.
The direct criterion determining voltage sag type according to monitoring node Voltage Drop root-mean-square value of described step 3
Concrete criterion is:
(1) three phase short circuit fault: three-phase voltage root-mean-square value is below 0.9pu and every phase root-mean-square value is identical；
(2) singlephase earth fault: a certain phase voltage root-mean-square value is less than 0.9pu, and remaining is biphase higher than 1pu；
(3) two-phase short-circuit fault: certain phase voltage root-mean-square value is normal, fluctuates in the little scope of about 1pu, and remaining is biphase low
In 0.9pu；
(4) double earthfault: certain phase voltage root-mean-square value is apparently higher than 1pu, and remaining is biphase less than 0.9pu.
In following step 4 and step 5, refer both to the whole network node of monitoring node Voltage Drop pattern and corresponding fault type
Voltage Drop pattern database carries out pattern recognition, and a point fault type carries out pattern recognition, can be effectively improved pattern recognition precision,
It it is one of the BROAD SUMMARY of the present invention.
Step 4, with the monitoring node Voltage Drop pattern of described step 2 for identifying feature, with identical or closest
Emulation the whole network node voltage fall pattern DS (best) for identifying target, with the rms voltage of monitoring point corresponding node
Pattern recognition trueness error is minimised as identification condition and carries out pattern recognition, find out therewith closest to the whole network node voltage fall
Stamping die formula, reduces the rms voltage of all the whole network nodes, thus completes to estimate the Voltage Drop root-mean-square value of the whole network node
Meter.
Owing to breaking down in electrical network, when the factors such as power system operating mode, fault type, abort situation all determine then have
Its intrinsic fault mode.Result from the region of the Voltage Drop in electrical network and the degree of depth the most all determines that, referred to as electrical network
Inherent voltage falls pattern.Obviously inevitable one_to_one corresponding is had to close between electrical network is intrinsic fault mode and Voltage Drop pattern
System.Can find out according to the Voltage Drop root-mean-square value of monitoring node, application model recognition methods under a certain fault mode
Fall pattern for close the whole network node voltage, just can reduce the rms voltage of all nodes, thus complete all non-
The Voltage Drop root-mean-square value of monitoring node is estimated.
The whole network node voltage owing to setting up falls pattern database and contains all of fault type and simulate institute
Possible forecast failure.The whole network node voltage that Arbitrary Fault causes falls pattern and falls mode data at the whole network node voltage
Storehouse certainly exists one the most identical or closest to emulation the whole network node voltage fall pattern.Assume certain event
Under barrier, monitoring node Voltage Drop pattern is MS (t), and in actual electric network, actual the whole network node voltage of its correspondence falls pattern
For DS (t), the whole network node voltage obtained in emulation falls in pattern database, certainly exists emulation the whole network node voltage
Fall pattern DS (best) fall with actual the whole network node voltage pattern DS (t) identical or closest to.DS (best) is
In this step, mode identification method wants the target the whole network node voltage found to fall pattern, and this simulation value available replaces real
The whole network node voltage falls pattern.
The concrete steps of the pattern recognition of described step 4 include:
(1) monitoring node Voltage Drop root-mean-square value record data and the whole network node voltage are fallen corresponding node in pattern
Rms voltage ask for pattern recognition error precision value as the following formula and keep a record, preserve minimum pattern recognition error precision
Value and corresponding the whole network node voltage thereof fall the sequence number of pattern.
(2) the whole network node voltage is fallen all the whole network node voltages in pattern database and falls the above behaviour of pattern repetition
Make, finally show that all the whole network node voltages fall in pattern database, the whole network node electricity that pattern recognition trueness error is minimum
Pressure falls pattern, thus identifies and be best able to represent true the whole network node voltage and fall Voltage Drop root-mean-square value imitative of pattern
True estimated value；
MRIi=Σ (MSj-DSi,j)2/Mnumber (4)
In above formula, MRI represents that monitoring node Voltage Drop pattern MS falls the corresponding joint of pattern DS with the whole network node voltage
The pattern recognition trueness error of rms voltage of point, MnumberRepresent be triggered under certain fault monitoring instrument number,
I takes and falls all over the whole network node voltage that all of the whole network node voltage in pattern database falls pattern, j takes all over all prisons that are triggered
Survey the monitoring node numbering of instrument, MS represents that the rms voltage of monitoring node Voltage Drop pattern, DS represent at actual electricity
Actual the whole network node voltage that in net, monitoring node Voltage Drop pattern is corresponding falls the rms voltage of pattern.
Step 5, setting voltage fall threshold value, fall according to the whole network node voltage under short troubles all in specific time period
The simulation estimate value of the Voltage Drop root-mean-square value of pattern, the voltage carried out one by one under statistics each Voltage Drop threshold value of acquisition falls
The frequency that falls estimated value.
In step 4, it may be determined that the whole network node voltage under certain fault condition falls the estimated value of pattern.To all of
Single failure repeat the mode identification procedure of step 1 to 4 can obtain in special time period the whole network node of faulty correspondence
The estimated value of Voltage Drop pattern, i.e. every time the rms voltage state of all nodes under fault.Determining according to Voltage Drop
Justice, generally takes 9 discrete Voltage Drop threshold values and carries out the Voltage Drop degree of depth and frequency statistics thereof, in concrete practice, and can basis
The number of Voltage Drop threshold value increase and decrease statistics point.
In the present embodiment, the mode identification method of the application present invention complete Voltage Drop root-mean-square value state and
The estimation of the Voltage Drop frequency, when there is any short trouble in electrical network, first according to the record data in monitoring instrument, builds
Monitoring node Voltage Drop pattern.Then monitoring node Voltage Drop root-mean-square value record data are fallen with the whole network node voltage
In pattern, the rms voltage of corresponding node asks for pattern recognition trueness error by formula (4), and keeps a record, and preserves minimum mould
Formula accuracy of identification error amount and corresponding the whole network node voltage thereof fall the sequence number of pattern, and the whole network node voltage is fallen mode data
Pattern of falling all the whole network node voltages in storehouse repeats above operation.Finally can fall pattern count by all the whole network node voltages
According in storehouse, the minimum the whole network node voltage of pattern recognition trueness error falls pattern, be i.e. have identified can represent true complete
Net node voltage falls the simulation value of pattern.
Below as a example by standard test system as shown in Figure 2, to a kind of Voltage Drop root-mean-square value of the present invention with fall
The mode identification method that the frequency that falls is estimated carries out checking and illustrates:
As in figure 2 it is shown, this test system is IEEE-14 node power distribution network, containing 14 bus nodes, 13 circuits；If
This grid reference capacity fixed is 100MVA, and reference voltage is 23kV, power distribution network open loop operation.
In order to consider the impact of distributed power source in active distribution network in the present embodiment, it is assumed that access DG at node 7 and 12.
Then grid voltage sags root-mean-square value and frequency estimation model recognition methods the step of the present invention are as follows:
Step 1, employing fault position method and short circuit calculation method, the Voltage Drop pattern caused according to different faults type, point
The whole network node voltage not setting up different faults type falls pattern database.
The process of distributed power source DG in active distribution network: account for exhausted vast scale in view of wind-powered electricity generation in China's distributed power generation
(more than 70%), and wherein apply most double-fed induction wind driven generators for direct grid-connected, in the present embodiment, with 1.5MW
Access the endpoint node 7 and 12 of distribution network as DG as a example by double-fed induction wind driven generator.
According to the present invention definition to the related notion that mode identification method relates in Voltage Drop estimates application, build
Desired data.
Topological structure according to electrical network as shown in Figure 2 and model parameter, application and trouble point method, at whole power system net
Network arranges simulation optional position, trouble point be short-circuited the situation of fault.
1. to a certain fault type, every circuit is uniformly arranged 11 trouble points, with the existing short circuit of power system
Computational methods calculate the rms voltage of all 14 nodes, with all node voltage root-mean-square values as element, according to node
Numbering size is structural order, and the trouble point of each simulation all constitutes the whole network node voltage and falls mode sequences.Therefore it is every
It is 1*14 dimension that one the whole network node voltage falls pattern, and 11 trouble points of a circuit can obtain the whole network node electricity of 11 1*14 dimensions
Pressure falls mode sequences.
2. 13 circuits of the whole network being repeated step A, the whole network node voltage that can obtain 143 1*14 dimensions falls mode sequences.
These 143 the whole network node voltages fall the whole network node voltage that mode sequences i.e. may make up under certain fault type and fall pattern count
According to storehouse.
3. for four kinds of different fault types (singlephase earth fault, two-phase short-circuit fault, double earthfault, three-phases
Short trouble), respectively repeat steps A-step B, the forecast failure of four kinds of fault types of simulation, respectively constitute four the whole network nodes
Voltage Drop pattern database.Four the whole network node voltages fall pattern database and all comprise the whole network node of 143 1*14 dimensions
Voltage Drop mode sequences, therefore comprise four kinds of the whole network node voltages falling the integrated database size altogether of pattern is 572 1*
The whole network node voltage of 14 dimensions falls mode sequences.
The present embodiment is only given as a example by the whole network node voltage of singlephase earth fault falls pattern database, such as table 1 institute
Showing, wherein, row represents that totally 143 the whole network node voltages fall pattern, and node 1 to node 14 is shown in list.
(table 1): singlephase earth fault the whole network node voltage falls pattern database (unit/pu)
Step 2, according to monitoring node Voltage Drop root-mean-square value record data, build monitoring node Voltage Drop pattern.
The fault that is short-circuited in a certain moment power system result in Voltage Drop event, meets at monitoring point allocation plan
Under conditions of the faulty all standing monitoring of the whole network institute, in the whole network, all monitoring points at least 1 monitoring instrument is triggered, or many
Platform monitoring instrument all has monitoring record.Record case to all monitoring points processes, for less than 0.9pu Voltage Drop threshold
The rms voltage of value can trigger monitoring instrument and produce record data, and some monitoring node is higher than due to rms voltage
0.9pu thus do not have trigger voltage to fall monitoring instrument, no record data in device, in the present embodiment with " # " represent monitoring point without
The situation of record, and fill in a form according to monitoring node number order in electrical network, constitute monitoring node Voltage Drop pattern,
As shown in table 2.Obvious, in monitoring node Voltage Drop pattern under this failure condition of " # " positional representation, corresponding monitoring node without
Record, its implicit information is that this node voltage root-mean-square value is higher than Voltage Drop threshold value 0.9pu set.
It is Voltage Drop root-mean-square value and the premise of frequency estimation owing to configuring Voltage Drop monitoring point in bus nodes, by
In the present invention without the whole network node is monitored entirely, it is only necessary to selected minority node is surveyed, and can effectively reduce monitoring point
Demand thus greatly reduce monitoring cost.The present embodiment is assumed in network unique according to 0.9pu Voltage Drop threshold value
Configuration monitoring point, monitoring node is: node 2, node 10, node 12.In table 2, altogether 3 row represent 3 monitoring nodes respectively, 40
Row represents that investigating period interior having altogether occurs 40 short troubles.
(table 2): monitoring node Voltage Drop pattern composition table:
Step 3, employing determine the direct criterion pair of voltage sag type according to monitoring node Voltage Drop root-mean-square value
Voltage sag type is identified and classifies, and falls mode data according to the whole network node voltage that short trouble type selecting is corresponding
Storehouse, and carry out pattern recognition in this whole network Voltage Drop pattern database.
For short trouble each time, with voltage sag type direct criterion voltage sag type is identified and
Classification, determines the direct criterion of voltage sag type according to Voltage Drop root-mean-square value:
(1) three phase short circuit fault: three-phase voltage root-mean-square value is below 0.9pu and every phase root-mean-square value is identical.
(2) singlephase earth fault: a certain phase voltage root-mean-square value is less than 0.9pu, and remaining is biphase higher than 1pu.
(3) two-phase short-circuit fault: certain phase voltage root-mean-square value is normal (fluctuation of about 1pu little scope), and remaining is biphase low
In 0.9pu.
(4) double earthfault: certain phase voltage root-mean-square value is apparently higher than 1pu, and remaining is biphase less than 0.9pu.
Can obtain through type identification for 40 faults altogether, singlephase earth fault 29 times, two-phase short-circuit fault 5 times,
Double earthfault 4 times, three phase short circuit fault 2 times.
Step 4, with the monitoring node Voltage Drop pattern of described step 2 for identifying feature, with identical or closest
Emulation the whole network node voltage fall pattern for identify target, with the pattern recognition of the rms voltage of monitoring point corresponding node
Trueness error is minimised as identification condition and carries out pattern recognition, find out with its closest to the whole network node voltage fall pattern,
Reduce the rms voltage of all the whole network nodes, thus complete the Voltage Drop root-mean-square value to the whole network node and estimate.
In step 4, according to the classification results of this step 3 monitoring node Voltage Drop pattern and corresponding fault type
The whole network node voltage falls pattern database and carries out pattern recognition.Divide fault type to carry out pattern recognition, identification can be effectively improved
Precision.
Monitoring node Voltage Drop root-mean-square value record data (table 2) is fallen with the whole network node voltage corresponding joint in pattern
The rms voltage of point asks for pattern recognition trueness error by formula (4), and keeps a record, and preserves minimum pattern recognition precision by mistake
Difference and corresponding the whole network node voltage thereof fall the sequence number of pattern, and all the whole network node voltages in data base are fallen pattern weight
Multiple above operation.The whole network node voltage that finally can obtain pattern recognition trueness error minimum falls pattern, is i.e. to have identified
True the whole network node voltage can be represented and fall the estimated value of pattern.All of single failure is repeated above mode identification procedure i.e.
Can obtain in special time period the whole network node voltage of faulty correspondence fall the estimated value of pattern, i.e. obtain under each fault
The rms voltage state of all nodes.
MRIi=Σ (MSj-DSi,j)2/Mnumber (4)
In formula (4), Mnumber represents the actual monitoring instrument quantity being triggered, it is clear that in the present embodiment, Mnumber takes
Value is between 1 to 3, and concrete value is according to how many decisions of " # " in table 2.Pattern recognition result is shown in Table 5.Wherein the 1st it is classified as pattern
Identification error, 2-the 15th is classified as the rms voltage estimated value of node 1-node 14, and 40 row are all in representing the investigation period
40 short troubles.
(table 3): pattern recognition error and rms voltage estimated result
Step 5, setting voltage fall threshold value, fall according to the whole network node voltage under short troubles all in specific time period
The simulation estimate value of the Voltage Drop root-mean-square value of pattern, the voltage carried out one by one under statistics each Voltage Drop threshold value of acquisition falls
The frequency that falls estimated value.
In step 4, it may be determined that the whole network node voltage under fault condition falls the estimated value of pattern.According to Voltage Drop
Definition add up, such as table 4 (the present embodiment take the most commonly used 9 discrete voltage fall threshold value add up).This enforcement
14 nodes altogether of distribution network used by example, therefore 2*14 row altogether.
(table 4): different depth Voltage Drop frequency statistics table
Being contrasted with the inventive method gained estimated value from Voltage Drop frequency actual value in table 4, estimated value is with actual
Closely, maximum absolute error number of times, within 2 times, can estimate that virtual voltage falls the frequency to value well.And this
Bright it is not only restricted to the uncertain factors such as components and parts fault rate, it is adaptable to the estimation of random time yardstick, when can accurately provide concrete
Voltage Drop frequency estimated result in Duan.Compare traditional method principle the simplest, put into practice (tradition side more convenient, quick
Method application integer programming method solving state estimate equation need tens of hours, and the present invention complete Voltage Drop root-mean-square value and
Its Voltage Drop frequency is estimated only to need the several seconds).
In conjunction with table 3 pattern recognition error and rms voltage estimated result, gained pattern recognition error of the present invention is little, energy
Effectively estimate that node voltage falls root-mean-square value, compare existing method for estimating state and can provide more useful Voltage Drop letter
Breath.Therefore explanation the inventive method can correctly and efficiently complete Voltage Drop root-mean-square value and the Voltage Drop frequency is estimated.
It is emphasized that embodiment of the present invention is illustrative rather than determinate, bag the most of the present invention
Include the embodiment being not limited to described in detailed description of the invention, every by those skilled in the art according to technical scheme
Other embodiments drawn, also belong to the scope of protection of the invention.
Claims (4)
1. a Voltage Drop root-mean-square value and the mode identification method falling frequency estimation, it is characterised in that: include following step
Rapid:
Step 1, employing fault position method and short circuit calculation method, the Voltage Drop pattern caused according to different faults type, build respectively
The whole network node voltage of vertical different faults type falls pattern database；
Step 2, according to monitoring node Voltage Drop root-mean-square value record data, build monitoring node Voltage Drop pattern；
According to monitoring node Voltage Drop root-mean-square value, step 3, employing determine that the direct criterion of voltage sag type is to voltage
Dip type is identified and classifies, and falls pattern database according to the whole network node voltage that short trouble type selecting is corresponding,
And carry out pattern recognition in this whole network Voltage Drop pattern database；
Step 4, with the monitoring node Voltage Drop pattern of described step 2 for identifying feature, with identical or immediate imitative
True the whole network node voltage falls pattern for identifying target, with the pattern recognition precision of the rms voltage of monitoring point corresponding node
Error minimize is that identification condition carries out pattern recognition, find out with its closest to the whole network node voltage fall pattern, reduction
The rms voltage of all the whole network nodes, thus complete the Voltage Drop root-mean-square value to the whole network node and estimate；
Step 5, setting voltage fall threshold value, fall pattern according to the whole network node voltage under short troubles all in specific time period
The simulation estimate value of Voltage Drop root-mean-square value, carry out the Voltage Drop frequency that statistics obtains under each Voltage Drop threshold value one by one
Secondary estimated value.
A kind of Voltage Drop root-mean-square value the most according to claim 1 and the mode identification method falling frequency estimation, its
It is characterised by: the concrete steps of described step 1 include:
(1) according to topological structure and the model parameter of electrical network, application and trouble point method, in whole power system network, fault is set
There are singlephase earth fault, two-phase short-circuit fault, double earthfault and four kinds of three phase short circuit fault in some simulation optional position
The type of short trouble；
(2) according to Power System Shortcuts computational methods, the anticipation short trouble of four kinds of fault types in simulating grid, by emulation
Computational methods build the whole network node voltage of four kinds of short trouble types respectively and fall the data base of pattern.
A kind of Voltage Drop root-mean-square value the most according to claim 1 and 2 and the mode identification method falling frequency estimation,
It is characterized in that: the direct differentiation determining voltage sag type according to monitoring node Voltage Drop root-mean-square value of described step 3
The concrete criterion of method is:
(1) three phase short circuit fault: three-phase voltage root-mean-square value is below 0.9pu and every phase root-mean-square value is identical；
(2) singlephase earth fault: a certain phase voltage root-mean-square value is less than 0.9pu, and remaining is biphase higher than 1pu；
(3) two-phase short-circuit fault: certain phase voltage root-mean-square value is normal, fluctuates in the little scope of about 1pu, and remaining biphase is less than
0.9pu；
(4) double earthfault: certain phase voltage root-mean-square value is apparently higher than 1pu, and remaining is biphase less than 0.9pu.
A kind of Voltage Drop root-mean-square value the most according to claim 3 and the mode identification method falling frequency estimation, its
It is characterised by: the concrete steps of the mode identification method of described step 4 include:
(1) monitoring node Voltage Drop root-mean-square value record data and the whole network node voltage are fallen the electricity of corresponding node in pattern
Pressure root-mean-square value is asked for pattern recognition error precision value as the following formula and keeps a record, preserve minimum pattern recognition error precision value and
Its corresponding the whole network node voltage falls the sequence number of pattern.
(2) the whole network node voltage is fallen all the whole network node voltages in pattern database and falls the above operation of pattern repetition,
Finally showing that all the whole network node voltages fall in pattern database, the whole network node voltage of pattern recognition trueness error minimum falls
Stamping die formula, thus identify the emulation being best able to represent the Voltage Drop root-mean-square value that true the whole network node voltage falls pattern and estimate
Evaluation；
MRIi=∑ (MSj-DSi,j)2/Mnumber
In above formula, MRI represents that monitoring node Voltage Drop pattern MS and the whole network node voltage fall the corresponding node of pattern DS
The pattern recognition trueness error of rms voltage, MnumberRepresent that the monitoring instrument number being triggered under certain fault, i take
Fall all over the whole network node voltage that all of the whole network node voltage in pattern database falls pattern, j takes all over all monitorings that are triggered
The monitoring node numbering of instrument, MS represent that the rms voltage of monitoring node Voltage Drop pattern, DS represent in actual electric network
Actual the whole network node voltage corresponding to middle monitoring node Voltage Drop pattern falls the rms voltage of pattern.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610398935.9A CN106093638A (en) | 2016-06-06 | 2016-06-06 | A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610398935.9A CN106093638A (en) | 2016-06-06 | 2016-06-06 | A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106093638A true CN106093638A (en) | 2016-11-09 |
Family
ID=57227432
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610398935.9A Pending CN106093638A (en) | 2016-06-06 | 2016-06-06 | A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106093638A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034461A (en) * | 2018-07-06 | 2018-12-18 | 国网山西省电力公司电力科学研究院 | A kind of voltage dip Stochastic prediction method based on actual electric network monitoring information |
| CN110646655A (en) * | 2019-11-22 | 2020-01-03 | 深圳市鼎阳科技股份有限公司 | RMS value-based triggering method and digital oscilloscope |
| CN111682551A (en) * | 2020-07-11 | 2020-09-18 | 李瑶 | Load policy management system of safety device |
| US20220206553A1 (en) * | 2020-12-30 | 2022-06-30 | Innogrit Technologies Co., Ltd. | Voltage drop management for vlsi and soc |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104730420A (en) * | 2015-03-26 | 2015-06-24 | 曾江 | Node voltage estimation and voltage sag rapid detection method |
-
2016
- 2016-06-06 CN CN201610398935.9A patent/CN106093638A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104730420A (en) * | 2015-03-26 | 2015-06-24 | 曾江 | Node voltage estimation and voltage sag rapid detection method |
Non-Patent Citations (4)
| Title |
|---|
| 唐琳等: "电压暂降状态和水平评估模式匹配法与监测装置多目标优化配置", 《中国电机工程学报》 * |
| 徐永海等: "《电网中电压暂降和短时间中断》", 31 December 2015 * |
| 王玲等: "基于暂降类型判断的短路故障类型识别研究", 《电测与仪表》 * |
| 陈丽等: "基于模式识别的配电网故障区段定位", 《电工电能新技术》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034461A (en) * | 2018-07-06 | 2018-12-18 | 国网山西省电力公司电力科学研究院 | A kind of voltage dip Stochastic prediction method based on actual electric network monitoring information |
| CN109034461B (en) * | 2018-07-06 | 2021-04-20 | 国网山西省电力公司电力科学研究院 | Voltage sag random estimation method based on actual power grid monitoring information |
| CN110646655A (en) * | 2019-11-22 | 2020-01-03 | 深圳市鼎阳科技股份有限公司 | RMS value-based triggering method and digital oscilloscope |
| CN110646655B (en) * | 2019-11-22 | 2020-03-31 | 深圳市鼎阳科技股份有限公司 | RMS value-based triggering method and digital oscilloscope |
| CN111682551A (en) * | 2020-07-11 | 2020-09-18 | 李瑶 | Load policy management system of safety device |
| US20220206553A1 (en) * | 2020-12-30 | 2022-06-30 | Innogrit Technologies Co., Ltd. | Voltage drop management for vlsi and soc |
| US11385698B1 (en) * | 2020-12-30 | 2022-07-12 | Innogrit Technologies Co., Ltd. | Voltage drop management for VLSI and SoC |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112636341B (en) | Power system inertia spatial distribution estimation method and device based on multiple innovation identification | |
| CN108168682A (en) | A kind of GIL On-line Faults monitoring system based on vibration signal support vector machines | |
| CN108832619A (en) | Transient stability evaluation in power system method based on convolutional neural networks | |
| CN110570122B (en) | Offshore wind power plant reliability assessment method considering wind speed seasonal characteristics and current collection system element faults | |
| CN106093638A (en) | A kind of Voltage Drop root-mean-square value and the mode identification method falling frequency estimation | |
| Ma et al. | The characteristic ellipsoid methodology and its application in power systems | |
| CN104156892A (en) | Active distribution network voltage drop simulation and evaluation method | |
| CN104407273B (en) | Electric energy quality disturbance source positioning method considering monitoring reliability | |
| CN110348114B (en) | Non-precise fault identification method for power grid completeness state information reconstruction | |
| CN104362622A (en) | Load model of regional power grid with small hydroelectric generating set and modeling method | |
| CN102508055A (en) | Device and method for detecting wind power generation grid-connected system | |
| Werho et al. | A potential island formation identification scheme supported by PMU measurements | |
| Hu et al. | Hierarchical fault diagnosis for power systems based on equivalent-input-disturbance approach | |
| CN104616061A (en) | Island detection method based on wavelet packet logarithmic energy entropy and genetic algorithm optimization | |
| CN106570790A (en) | Wind farm output power data restoration method considering segmental characteristics of wind speed data | |
| Shiddieqy et al. | Power line transmission fault modeling and dataset generation for ai based automatic detection | |
| CN110363334A (en) | Grid-connected grid line loss prediction technique based on Grey Neural Network Model | |
| CN105071380A (en) | Power system wide-area coherency identification method and device | |
| Wang et al. | A review of power system transient stability analysis and assessment | |
| CN105046011A (en) | Rapid analysis method used for state of electric device and based on topological computation of distributed power grid | |
| Wei et al. | Online distribution system topology monitoring with limited smart meter communication | |
| CN105956760A (en) | Intelligent power distribution network situation perception method Based on multivariate spatio-temporal information modeling | |
| Trevizan et al. | Distribution system state estimation sensitivity to errors in phase connections | |
| Jaramillo et al. | Identification of distributed energy resources in low voltage distribution networks | |
| CN107069710B (en) | Power system state estimation method considering new energy space-time correlation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161109 |