A kind of intelligent phase-change switch system and method for adjusting three-phase imbalance
Technical field
The invention belongs to low-voltage network technical field, especially a kind of intelligent phase-change switch system for adjusting three-phase imbalance
System and its method.
Background technology
With the development of economic level, electric load is growing, and the random of a large amount of single-phase loads connects in low-voltage network
Enter to result in serious three-phase imbalance problem.Low-voltage distribution network three-phase imbalance can increase transformer loss and line loss, drop
Low power quality causes heavily loaded phase end low-voltage, equipment is caused to can not work normally, single-phase overload may be caused when serious
Tripping causes taiwan area power outage.
At present, mainly using the method for reactive-load compensation, it is mainly thyristor control that equipment, which is formed, for the improvement of three-phase imbalance
Reactor TCR (Thyristor Controlled Reactor) processed or static reactive generator SVG (Static Var
Generator), TCR is shifted by alternate power realizes that distribution transforming exports three-phase equilibrium, and SVG is by injecting reversed imbalance compensation
Electric current come realize distribution transforming export three-phase equilibrium, do not solve the problems, such as that practical load evenly distributes fundamentally, it is therefore desirable to right
Load is redistributed to solve the problems, such as three-phase imbalance.Some improvement projects are to carry out three-phase imbalance benefit using phase-change switch
It repays, also controls phase-change switch commutation using distribution transformer terminals, realized so as to which the method for being mutually transferred to underloading phase from heavy duty will be loaded
The three-phase equilibrium of distribute-electricity transformer district, but distribute-electricity transformer district typically contains a plurality of branch line, three-phase is still uneven on each branch line in this way
Weighing apparatus, therefore firstly the need of the three-phase equilibrium realized on branch line, then entire taiwan area also achieves three-phase for all branch line three-phase equilibriums
Balance.In addition, load is mutually transferred to underloading from heavy duty mutually only can guarantee instant three-phase equilibrium.In one day, the electricity of load
Flow valuve is variation, and the changing rule of different load is different, this may cause phase-change switch in one day repeatedly to act to protect
Instant three-phase equilibrium is demonstrate,proved, not only increases the workload of processor, also leads to the repeatedly action of universe switch, increases switch
Loss.Therefore it is unstable to be calculated in a manner of adjusting three-phase equilibrium with instant data, and the variation to load is needed to advise
Rule is summarized, and carries out redistributing for three-phase load with this, to achieve the effect that three-phase equilibrium.
Invention content
It is an object of the invention to overcome the deficiencies in the prior art, propose that a kind of intelligent commutation for adjusting three-phase imbalance is opened
System and method are closed, to reduce the action frequency of switch, ensures that load reaches steady three-phase balance state, carries
The high reliability of safety and the power supply of equipment.
The present invention solves its technical problem and following technical scheme is taken to realize:
A kind of intelligent phase-change switch system for adjusting three-phase imbalance, including multiple main controllers and multiple phase-change switches, institute
Main controller is stated in every branch line outlet, the three-phase line of each main controller one end access branch line front end, other end access
The three-phase line of branch line back end;The phase-change switch is mounted on user front end, and branch line three-phase line is accessed in phase-change switch one end, separately
One terminating user;It is communicatively coupled between the main controller and phase-change switch by wired or wireless mode, each branch
All phase-change switches connected under main controller and the branch line on line form an independent work system.
Further, the main controller includes collecting unit, storage unit, communication unit and processor unit, the processing
Device unit is connected respectively with collecting unit, storage unit, communication unit, and the three of collecting unit one end access branch line front end
Phase line, the other end access the three-phase line of branch line back end by current transformer, are acquired on branch line by current transformer
Three-phase current, and be sent to processor unit and handled;The communication unit is responsible between main controller and phase-change switch
Communication and the communication of main controller and distribution transformer terminals, by wired or wirelessly lead between main controller and phase-change switch
Letter, for receiving the load current information of phase-change switch acquisition and issuing commutation adjustment order, main controller and distribution transformer terminals it
Between by wired or wirelessly communicate, for sending three-phase imbalance data and commutation handling result on distribution transformer terminals.
Further, the phase-change switch include collecting unit, communication unit, computing unit, protection location, control unit and
Switch unit;The load current information of the collecting unit acquisition user, information of voltage, electric leakage stream information;The communication unit
Main controller will be sent in the load current information data of acquisition and receive the commutation order that main controller issues;The computing unit
It is responsible for whether calculated load electric current, voltage, leakage current reach preset protection act condition, is opened if protection act condition is reached
The corresponding protection act of dynamic protection location;Described control unit performs the commutation order of main controller, and control switch unit carried out
Zero commutation operates;The protection location is equipped with under-voltage/over-voltage protection, short-circuit protection, overload protection, earth leakage protective, for controlling
Switch carries out protection act.
A kind of implementation method for the intelligent phase-change switch system for adjusting three-phase imbalance, includes the following steps:
Step 1, initialization:Set three-phase current unbalance degree threshold value UNthWith the frequency threshold value N that transfinitesth, set branch line three
Phase current acquisition time interval t0, set three-phase current unbalance transfinite times N initial value as 0;
Three-phase current in step 2, main controller acquisition branch line, calculates three-phase current unbalance degree:
UNI=max | IA-Iave|,|IB-Iave|,|IC-Iave|}/Iave
Wherein, UNIRepresent three-phase current unbalance degree, IA、IB、ICA, B, C the three-phase electricity flow valuve respectively sampled,
IaveRepresent three-phase average current, calculation formula is;
Iave=(IA+IB+IC)/3
By three-phase current unbalance degree UNIWith the three-phase current unbalance degree threshold value UN of settingthIt is compared, if beyond threshold
Value, then the three-phase current unbalance times N that transfinites add 1 and perform step 3, if without departing from three-phase current unbalance degree threshold value UNth, then
Wait for regular time t0Continue to execute step 2;
Step 3 compares three-phase current unbalance and transfinites and times N and transfinites frequency threshold value NthSize, if three-phase current is not
The balance times N that transfinites transfinites frequency threshold value Nth, then step 4 is performed, if without departing from regular time t is waited for0Perform step
Rapid 2;
The current data of all loads a few days ago in step 4, the processing unit reading storage unit of main controller, to former
The data at it same time point take arithmetic average, the characteristic current curve as the recent load;
The characteristic current curve of all loads is normalized in step 5, maps the data into the range of [0,1],
Then classified using fuzzy C-means clustering method to load, by all loads according to the difference of its characteristic current curve
Changing rule is divided into different classes, and it is similar that same type load represents its current curve changing rule, and step is performed to each type load
Rapid 6;
Step 6, in same type load, 0-1 codings are carried out to all on off state, it is mutually that A phases are then that switch, which is closed,
[1,0,0]T, switch closure is mutually that B phases are then [0,1,0]T, switch closure is mutually that C phases are then [0,0,1]T, it is assumed that switch number
For n, then the matrix that the element value that the state of on-load switch can be expressed as a 3 × n is 0 or 1 randomly generates on off state
Initial matrix, sets that object function is minimum as branch current degree of unbalancedness and switch motion number is minimum, using differential evolution
Algorithm finds the optimum state of switch;
Step 7, according to the optimized switching state that calculates, itself and actual switch state are compared, determine to need into
The switch of row commutation operation, main controller are handed down to phase-change switch by commutation instruction and commutation are separate by communication;
Step 8, the commutation that the phase-change switch for carrying out commutation operation is needed to receive main controller instruct, and control unit control is opened
It closes unit and performs corresponding phase switching operation.
Further, the fuzzy C-means clustering method in the step 5 includes the following steps:
Step (1), setting cluster numbers c, iteration threshold Δ and maximum iteration;
(2), with the random number between 0~1 step initializes a subordinated-degree matrix U, make all degrees of membership and be 1:
In formula, uijFor i-th group of j-th of element in subordinated-degree matrix U, n is every group of element number;
(3), according to U step calculates cluster centre Matrix C, computational methods are:
In formula, ciFor i-th group in cluster centre Matrix C of cluster centre, xjFor j-th of sample value;
Step (4), calculating target function J, computational methods are:
In formula, dij=| | ci-xj| | the Euclidean distance for j-th of data to ith cluster center;
Step (5), compare the target function value that the value of object function is obtained with last iterative calculation, if knots modification is less than
The threshold value of setting reaches maximum iteration, then stops iteration, exports the subordinated-degree matrix being finally calculated, goes to step
(7), otherwise go to step (6);
Step (6), calculate new subordinated-degree matrix U ', computational methods are:
In formula, u 'ijFor the element of new subordinated-degree matrix U ', (3) return to step carries out new iteration;
Step (7), according to obtained subordinated-degree matrix, select maximum membership degree value as generic.
Further, the differential evolution algorithm in the step 6 includes the following steps:
Step (1), initialization:Definitive variation operator F0, crossover operator CR and maximum evolutionary generation Gm, determine fitness function
W;
Step (2), randomly generate initial population X0, evolutionary generation G=1;
The fitness value W of step (3), each individual in calculating initial population0;
Step (4), judge whether evolutionary generation G reaches maximum evolutionary generation Gm;It is terminated if so, evolving, it will at this time most
Good individual is as solution output;If it is not, continue step (5);
Step (5), choose three random parents into row variation and crossover operation, boundary condition is handled, is obtained temporarily
Population XG1;
Step (6), to interim population XG1It is evaluated, calculates the fitness value W in interim populationG1;
(7), to interim population and contemporary population step carries out selection operation, obtain new population XG;
Step (8), evolutionary generation G add 1, go to step (4).
The advantages and positive effects of the present invention are:
The phase-change switch that the present invention installs main controller control load front end on branch line carries out commutation operation, in master controller
It is middle to be classified using fuzzy C-means clustering to the characteristic current curve of load and each using differential evolution algorithm searching
The optimal commutation strategy of class reaches branch and the steady three-phrase burden balance of taiwan area to the full extent, ensures tri-phase unbalance factor
Minimum and switch motion number is minimum, so as to improve the reliability of the safety of equipment and power supply.
Description of the drawings
Fig. 1 is the intelligent phase-change switch system connection diagram of the present invention;
Fig. 2 is the master circuit block diagram of the present invention;
Fig. 3 is the phase-change switch circuit block diagram of the present invention;
Fig. 4 is the work flow diagram of the present invention;
Fig. 5 is the calculation flow chart of the fuzzy C-means clustering method of the present invention;
Fig. 6 is the calculation flow chart of differential evolution algorithm.
Specific embodiment
The embodiment of the present invention is further described below in conjunction with attached drawing.
A kind of intelligent phase-change switch system for adjusting three-phase imbalance, as shown in Figure 1, including main controller and phase-change switch,
Main controller is mounted in every branch line outlet, the three-phase line of main controller one end access branch line front end, after other end access branch line
The three-phase line at end, phase-change switch are mounted on user front end, one end access branch line three-phase line, another terminating user.Main controller
It is communicatively coupled between phase-change switch by wired or wireless mode, under the main controller on each branch line and the branch line
All phase-change switches of connection form an independent work system.Main controller is using fuzzy C-means clustering to the spy of load
Sign current curve is classified, and is found using differential evolution algorithm per a kind of optimal commutation strategy, is realized three on the branch line
It balances each other;All branch line three-phase equilibriums, then entire distribute-electricity transformer district also three-phase equilibrium.
Main controller generates for the commutation strategy of user, and clustering algorithm and differential evolution algorithm meter are used according to historical data
Optimal commutation strategy is calculated, and commutation order is issued to phase-change switch and reversing switch is controlled to carry out commutation action.Such as Fig. 2 institutes
Show, the main controller is mainly comprising collecting unit, storage unit, communication unit and processor unit, the access of collecting unit one end
The three-phase line of branch line front end, the other end are accessed the three-phase line of branch line back end by current transformer, pass through current transformer
The three-phase current on branch line is acquired, and is sent to processor unit and is handled.Communication unit is mainly responsible for main controller with changing
The communication of communication and main controller and distribution transformer terminals between mutually switching, passes through wired or nothing between main controller and phase-change switch
The mode of line communicates, for receive phase-change switch acquisition load current information and issue commutation adjustment order, main controller with
By wired or wirelessly communicate between distribution transformer terminals, for sending three-phase imbalance data and commutation on distribution transformer terminals
Handling result.Processor unit mainly calculates tri-phase unbalance factor according to three-phase current data, and judges whether it transfinites, if super
Limit then starts commutation strategy calculating process, continues acquisition if not transfiniting and calculates tri-phase unbalance factor.
As shown in figure 3, the phase-change switch is mainly comprising collecting unit, communication unit, computing unit, protection location, control
Unit and switch unit processed, collecting unit mainly acquire the load current information of user, information of voltage, electric leakage stream information etc..It is logical
Letter unit mainly will be sent to main controller in the load current information data of acquisition and receive the commutation order that main controller issues.Meter
It calculates unit and is mainly responsible for whether calculated load electric current, voltage, leakage current reach preset protection act condition, if it is dynamic to reach protection
Make the condition then corresponding protection act of starting protection unit.Control unit mainly performs the commutation order of main controller, control switch
Unit carries out zero crossing commutation operation.Protection location is used equipped with under-voltage/over-voltage protection, short-circuit protection, overload protection, earth leakage protective
It is switched in control and carries out protection act.
The course of work of intelligent phase-change switch system is:Phase-change switch had the load current acquired in real time every 1 hour
It is sent in valid value active to main controller, current value is stored in storage unit by main controller, and main controller is adopted every fixed duration t0
Collect the three-phase electricity flow valuve on branch line, and calculate three-phase current unbalance degree, and set the threshold value UNth of tri-phase unbalance factor and get over
Limit frequency threshold value Nth, if the tri-phase unbalance factor calculated exceed threshold value, while out-of-limit cumulative number also exceed threshold value when, then
Start commutation adjusting and optimizing policy calculation.
A kind of intelligent phase-change switch system for adjusting three-phase imbalance, as shown in figure 4, including the following steps:
Step 1, initialization, setting three-phase current unbalance degree threshold value UNthWith the frequency threshold value N that transfinitesth, set branch line three
Phase current acquisition time interval t0, and set three-phase current unbalance transfinite times N initial value as 0.
Three-phase current in step 2, main controller acquisition branch line, is calculated as follows three-phase current unbalance degree:
UNI=max | IA-Iave|,|IB-Iave|,|IC-Iave|}/Iave
Wherein, UNIRepresent three-phase current unbalance degree, IA、IB、ICA, B, C the three-phase electricity flow valuve respectively sampled,
IaveRepresent three-phase average current:
Iave=(IA+IB+IC)/3
By three-phase current unbalance degree UNIIt is compared with the three-phase current unbalance degree threshold value of setting, if beyond threshold value,
Then N adds 1 and performs step 3, if without departing from threshold value, waits for regular time t0Continue to execute step 2.
Step 3 compares three-phase current unbalance and transfinites times N and threshold value NthSize, if N exceed threshold value Nth, then perform
Step 4, if without departing from regular time t is waited for0Perform step 2.
Step 4, main controller processing unit are read before all loads in storage unit 5 days 0:00~24:00 electric current number
According to Ihis=(Id1,Id2,Id3,Id4,Id5), wherein Idm=[im1,im2,…,im24]T(m=1,2 ..., 5) represents the m days 24 small
When current sampling data, arithmetic average is taken to the data at the same time point of 5 days as the following formula, the spy as the recent load
Levy current curve:
The characteristic current curve of all loads is normalized in step 5, maps the data into the range of [0,1],
Then classified using fuzzy C-means clustering method to load, by all loads according to the difference of its characteristic current curve
Changing rule is divided into different classes, and it is similar that same type load represents its current curve changing rule, and step is performed to each type load
Rapid 6.
In this step, the characteristic current curve of all loads is normalized, maps the data into [0,1] model
Method in enclosing is as follows:
Iafter=(Ibefore-min{[Ibefore]})/(max{[Ibefore]}-min{[Ibefore]})
In formula, Ibefore、IafterThe preceding data with after normalization of normalization, [I are represented respectivelybefore] represent IbeforePlace
Set of data samples, max { [Ibefore] and min { [Ibefore] respectively represent set of data samples maxima and minima.
As shown in figure 5, the fuzzy C-means clustering method that this step uses includes the following steps:
Step (1), setting cluster numbers c, iteration threshold Δ and maximum iteration.
(2), with the random number between (0,1) step initializes a subordinated-degree matrix U, make all degrees of membership and be 1:
In formula, uijFor i-th group of j-th of element in subordinated-degree matrix U, n is every group of element number.
(3), according to U step calculates cluster centre Matrix C, computational methods are:
In formula, ciFor i-th group in cluster centre Matrix C of cluster centre, xjFor j-th of sample value.
Step (4), calculating target function J, computational methods are:
In formula, dij=| | ci-xj| | the Euclidean distance for j-th of data to ith cluster center.
Step (5), compare the target function value that the value of object function is obtained with last iterative calculation, if knots modification is less than
The threshold value of setting reaches maximum iteration, then stops iteration, exports the subordinated-degree matrix being finally calculated, goes to step
(7), otherwise go to step (6).
Step (6), calculate new subordinated-degree matrix U ', computational methods are:
In formula, u 'ijElement for new subordinated-degree matrix U '.(3) return to step carries out new iteration.
Step (7), according to obtained subordinated-degree matrix, select maximum membership degree value as generic.
Step 6, in same type load, 0-1 codings are carried out to all on off state, it is mutually that A phases are then that switch, which is closed,
[1,0,0]T, switch closure is mutually that B phases are then [0,1,0]T, switch closure is mutually that C phases are then [0,0,1]T, it is assumed that switch number
For n, then the matrix X=(x that the element value that the state of on-load switch can be expressed as a 3 × n is 0 or 1ij) (i=1,2,3;j
=1,2 ..., n;xij=0 or 1), randomly generate the initial matrix X of on off state0, object function is set as branch current injustice
Weighing apparatus degree is minimum and switch motion number is minimum, and switch is found using the variation of differential evolution algorithm, intersection and selection operation
Optimum state Xbest。
In this step, the qualifications of matrix generation are:
Known load current data IaWith branch actual switch state Xpre.Object function is set as branch current imbalance
Degree is minimum and switch motion number is minimum, i.e. min W=min (UNI+ Σ k), wherein UNIFor the three-phase current that is calculated not
The degree of balance, Σ k are the action frequency of all switches on branch line, and computational methods are with actually opening by the on off state X after transformation
Off status XpreBe compared by row, if certain row switch it is separate converted if the action frequency that switchs add 1.
As shown in fig. 6, the specific implementation of differential evolution algorithm includes the following steps:
Step (1), initialization, definitive variation operator F0, crossover operator CR and maximum evolutionary generation Gm, determine fitness function
W。
Step (2), randomly generate initial population X0, evolutionary generation G=1.
The fitness value W of step (3), each individual in calculating initial population0。
Step (4), judge whether evolutionary generation G reaches maximum evolutionary generation Gm.It is terminated if so, evolving, it will at this time most
Good individual is as solution output;It is (5) walked if it is not, continuing the.
Step (5), choose three random parents into row variation and crossover operation, boundary condition is handled, is obtained temporarily
Population XG1。
Step (6), to interim population XG1It is evaluated, calculates the fitness value W in interim populationG1。
(7), to interim population and contemporary population step carries out selection operation, obtain new population XG。
Step (8), evolutionary generation G add 1, turn the and (4) walk.
Step 7, according to the optimized switching state that calculates, itself and actual switch state are compared, determine to need into
The switch of row commutation operation, main controller are handed down to phase-change switch by commutation instruction and commutation are separate by communication.
Step 8, the commutation that the phase-change switch for carrying out commutation operation is needed to receive main controller instruct, and control unit control is opened
It closes unit and performs corresponding phase switching operation.
It is emphasized that embodiment of the present invention is illustrative rather than limited, therefore present invention packet
Include the embodiment being not limited to described in specific embodiment, it is every by those skilled in the art according to the technique and scheme of the present invention
The other embodiment obtained, also belongs to the scope of protection of the invention.