CN103545827B - A kind of three-phase imbalance load distribution method being applicable to low-voltage network - Google Patents

Abstract

A kind of three-phase imbalance load distribution method being applicable to low-voltage network, the method is for the three-phase load unbalance situation often occurred in low-voltage network, by extracting user power utilization load data, fuzzy C-means clustering (FCM) algorithm is adopted to carry out cluster analysis to customer charge feature, according to categorization results, all types of user is allocated in low-voltage network by optimal algorithm, thus suppress the three-phase imbalance of low-voltage network, finally reach the target reducing low-voltage network loss and improve low-voltage network reliability.The present invention carries out pattern classification by user's day or monthly load curve to customer charge, the customer charge optimized algorithm being belonged to same load pattern searches out the phase sequential mode of best access electrical network, thus realize the time division balance of classification on the spot of load side, thus on source, at utmost reduce the degree of unbalance of distribution.The three-phase imbalance load that the present invention is applicable to low-voltage network distributes.

Description

A kind of three-phase imbalance load distribution method being applicable to low-voltage network

Technical field

The present invention relates to a kind of three-phase imbalance load distribution method being applicable to low-voltage network, belong to low-voltage distribution technical field.

Background technology

Low-voltage network has the advantages that quantity is many and scope is wide, low-voltage customer load mostly is single-phase, and electricity consumption behavior has randomness by seasonal effect, therefore low-voltage network load three-phase imbalance situation is very outstanding, the harm that low-voltage network load three-phase imbalance brings is mainly reflected in the loss adding distribution transformer and circuit, affect power consumption equipment normally to work, reduce distribution transformer and exert oneself and reduce low-voltage network power supply reliability etc.

For low-voltage network load three-phase imbalance problem, by extracting user power utilization load data, fuzzy C-means clustering (FCM) algorithm is adopted to carry out cluster analysis to customer charge feature, according to categorization results, all types of user is allocated in low-voltage network by optimal algorithm, thus suppress the three-phase imbalance of low-voltage network, finally reach the target reducing low-voltage network loss and improve low-voltage network reliability.

Summary of the invention

The object of the invention is, in order to solve the three-phase imbalance problem of low-voltage network, the present invention discloses a kind of three-phase imbalance load distribution method being applicable to low-voltage network.

Realizing technical scheme of the present invention is, by extracting user power utilization load data, fuzzy C-means clustering (FCM) algorithm is adopted to carry out cluster analysis to customer charge feature, according to categorization results, all types of user is allocated in low-voltage network by optimal algorithm, thus suppress the three-phase imbalance of low-voltage network, finally reach the target reducing low-voltage network loss and improve low-voltage network reliability.

The present invention sets up three-phase imbalance commutation control strategy, paste C mean cluster (FCM) algorithm is adopted to carry out cluster analysis to customer charge in low-voltage network, the daily load curve of user is divided in units of 1h, be divided into 24 deciles, for the data that customer charge feature clustering is analyzed, utilize differential evolution optimal algorithm to carry out dynamically optimized scheduling to low-voltage network unbalance condition, switch the separate of user by actuator, thus suppress the three-phase imbalance of low-voltage network.

Three-phase imbalance commutation control strategy of the present invention is based on customer charge feature, function is set up at three-phase current unbalance degree and user load connection type, the best phase sequence connection mode of load is searched by optimization method, thus make the object that the degree of unbalance of load current reaches minimum, that is:

J＝min{(max(|Dev _a|,|Dev _b|,|Dev _c|)),(X)}

Here | Dev _a|, | Dev _b|, | Dev _c| be respectively a, the difference of b, c phase current and three-phase average current, J is three-phase load unbalance degree.

${\mathrm{Dev}}_i=\frac{I_{\mathrm{ph},i}}{I_{\mathrm{ave}}}-1(i=a,b,c)$

$I_{\mathrm{ave}}=\frac{I_{\mathrm{ph},a}+I_{\mathrm{ph},b}+I_{\mathrm{ph},c}}{3}$

I _{ph, a}, I _{ph, b}, I _{ph, c}be respectively a by Three-phase Power Flow program computation after user load commutation, b, c phase current.

N is the number of users switched, and X is the number of times of commutation.

$X=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{x}_j$

X _jfor jth user's commutation mark, if commutation, be 1, not commutation is then 0.Under the distribution transforming tri-phase unbalance factor condition of satisfied setting, guarantee that commutation number of times X is minimum.

Three-phase load phase sequence balance is based on the research to load pattern, in order to obtain the information in order to improve operation of power networks decision-making.Customer charge pattern according to extracting carries out Cluster Classification to power customer, takes equilibrium assignment method of attachment to reach the balance of three-phase load according to the load pattern curve after sorting out in the specific time period to belonging to same load pattern class user.

The present invention is based on the phase sequence balance following several principles of load pattern identification:

(1) under same load pattern, phase sequence balances as far as possible;

(2) each user is required that within continuous time commutation number of times and total degree are few as much as possible;

(3) for accomplishing to balance under same load pattern, and user's overall balance under other load patterns.

The present invention adopts fuzzy C-means clustering (FCM) algorithm to carry out cluster analysis to customer charge in low-voltage network, and the power samples value that 48 characteristic values of use are daily load 24h, carries out cluster to the load curve of different load.

Control strategy adopts the load pattern identification of fuzzy C-means clustering (FCM) algorithm, and before employing fuzzy C-means clustering, the first elimination capacity order of magnitude is on the impact of pattern analysis, usually adopts extreme value by load sequence normalization.

$x_{\mathrm{ij}}^{\′}=\frac{x_{\mathrm{ij}}-x_{\min ,j}}{x_{\max ,j}-x_{\min ,j}}(i=\mathrm{1,2},...,m,j=\mathrm{1,2},...,n)$

After normalization, data between [0,1], in order to extract load form; Symbol in formula, x _ijfor daily load value; x _{min, j}for customer charge minimum value; x _{max, j}for customer charge maximum.

FCM algorithm is by asking planning matrix U, and cluster centre V, makes cluster the minimization of object function,

$\min J_m(U,V)=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{c}{\mathrm{\Σ}}}{u}_{\mathrm{ij}}^m{d}_{\mathrm{ij}}^2(x_j,v_i)$

d _ij(x _j,v _i)＝||v _i-x _j||

In formula, n is the number of sample data, and m is that to be generally 2, c be cluster centre number to Weighted Index, d _ijsample point and cluster centre distance, fuzzy matrix the element u of matrix U _ijrepresent jth (j=1,2 ..., n) individual data sample point belong to i-th (i=1,2 ..., the c) degree of membership of class.

The invention has the beneficial effects as follows, the present invention carries out pattern classification by user's day or monthly load curve to customer charge, the customer charge optimized algorithm being belonged to same load pattern searches out the phase sequential mode of best access electrical network, thus realize the time division balance of classification on the spot of load side, thus on source, at utmost reduce the degree of unbalance of distribution.

The three-phase imbalance load that the present invention is applicable to low-voltage network distributes.

Accompanying drawing explanation

Fig. 1 is the flow chart that low-voltage network three-phase imbalance load distributes;

Fig. 2 is that (ordinate is load to load pattern identification cluster curve; Abscissa is hour);

In order to distribute load curve in front low-voltage network, (ordinate is load to Fig. 3; Abscissa is hour);

In order to distribute load curve in rear low-voltage network, (ordinate is load to Fig. 4; Abscissa is hour);

Symbol in figure: C1 is C1 class user; C2 is C2 class user; C3 is C3 class user; Ia is A phase current; Ib is B phase current; Ic is C phase current.

Embodiment

The specific embodiment of the invention presses the flow chart that Fig. 1 low-voltage network three-phase imbalance load distributes.

The present embodiment low and medium voltage distribution network has 20 users, load curve as table 1, user and load characteristics clustering corresponding relation as table 2, by forming 3 class cluster curves after fuzzy C-means clustering (FCM) algorithm as Fig. 2.According to cluster result, determine the goal constraint that in user continuous time, commutation number of times and total degree are minimum, obtain user by optimal algorithm and distribute commutation operation table as table 3, before and after sharing of load commutation in low-voltage network load curve as Fig. 3 and Fig. 4.

Table 1 load curve

Table 2 load class and user's table of comparisons

In table, 1 represents that user belongs to this load class, and 0 expression does not belong to this load class.

Table 3 optimization distributes commutation operation table

User

U1

U2

U3

U4

U5

U6

U7

U8

U9

U10

Phase sequence

A

A

A

A

A

B

B

B

B

B

User

U11

U12

U13

U14

U15

U16

U17

U18

U19

U20

Phase sequence

C

C

C

C

A

B

C

A

B

C

Claims (2)

1. be applicable to a three-phase imbalance load distribution method for low-voltage network, it is characterized in that, described method establishment three-phase imbalance commutation control strategy, carries out cluster analysis to customer charge in low-voltage network; Described control strategy is based on customer charge feature, function is set up at three-phase current unbalance degree and user load connection type, searched the best phase sequence connection mode of load by optimization method, thus make the object that the degree of unbalance of load current reaches minimum, that is:

J＝min{(max(|Dev _a|,|Dev _b|,|Dev _c|)),(X)}

Here | Dev _a|, | Dev _b|, | Dev _c| be respectively a, the difference of b, c phase current and three-phase average current, J is three-phase load unbalance degree;

$\begin{array}{cc}{\mathrm{Dev}}_{ph}=\frac{I_{ph}}{I_{ave}}-1& (ph=a,b,c);\end{array}$

$I_{ave}=\frac{I_{ph,a}+I_{ph,b}+I_{ph,c}}{3};$

I _{ph, a}, I _{ph, b}, I _{ph, c}be respectively a by Three-phase Power Flow program computation after user load commutation, b, c phase current; I _avefor average current;

N is the number of users switched, and X is the number of times of commutation;

$X=\underset{j=1}{\overset{n}{\Σ}}{x}_j$

X _jfor jth user's commutation mark, if commutation, be 1, not commutation is then 0; Under the distribution transforming tri-phase unbalance factor condition of satisfied setting, guarantee that commutation number of times X is minimum.

2. a kind of three-phase imbalance load distribution method being applicable to low-voltage network according to claim 1, it is characterized in that, described to customer charge in low-voltage network carry out cluster analysis adopt Fuzzy C-Means Cluster Algorithm, the power samples value that 24 characteristic values used are daily load 24h, carries out cluster to the load curve of different load;

Control strategy adopts the load pattern identification of Fuzzy C-Means Cluster Algorithm, and before employing fuzzy C-means clustering, the first elimination capacity order of magnitude is on the impact of pattern analysis, adopts extreme value by load sequence normalization:

$x_{ij}^{\′}=\frac{x_{ij}-x_{\min ,j}}{x_{\max ,j}-x_{\min ,j}},(i=1,2,\mathrm{...},m,j=1,2,\mathrm{...},n)$

After normalization, data between [0,1], in order to extract load form; Symbol in formula, x _ijfor daily load value; x _{min, j}for customer charge minimum value; x _{max, j}for customer charge maximum:

Fuzzy C-Means Cluster Algorithm is by asking planning matrix U, and cluster centre V, makes cluster the minimization of object function,

${\mathrm{minJ}}_m(U,V)=\underset{j=1}{\overset{n}{\Σ}}\underset{i=1}{\overset{c}{\Σ}}{u}_{ij}^m{d}_{ij}^2(x_j,v_i)$

d _ij(x _j,v _i)＝||v _i-x _j||

In formula, n is the number of sample data, and m is that to be generally 2, c be cluster centre number to Weighted Index, d _ijsample point and cluster centre distance, fuzzy matrix the element u of matrix U _ijrepresent a jth data sample point, j=1,2 ... n, belongs to the degree of membership of the i-th class, i=1,2 ... c.