CN104167814A - Method for realizing distribution network reconfiguration based on multiple agents - Google Patents

Abstract

The invention relates to a method for realizing distribution network reconfiguration based on multiple agents. The method comprises the following steps: (1) network area division is carried out, and agent structure information is initialized; (2) speed and position information of all agents in each area is initialized; (3) power flow and network loss are calculated; (4) whether or not to succeed in a competition and cooperation mechanism of a multi-agent system grid structure is judged; (5) self-learning operation is carried out on the successful agents, and the unsuccessful agents are corrected by a binary particle swarm algorithm; (7) the fitness values of all the agents are calculated; (7) whether the number of iterations is met is judged, the method goes to the next step if the number of iterations is met, or the method returns to step (3); (8) reconfiguration results in the areas are output to a distribution sub-station agent; and (8) the optimal solution of whole-network reconfiguration is output. By adopting the method of the invention, rapid convergence can be realized, and an accurate global optimal solution can be obtained.

Description

A kind of power distribution network reconfiguration implementation method based on many agencies

Technical field

The invention belongs to Power System and its Automation and intelligent distribution network technical field, particularly a kind of power distribution network reconfiguration implementation method based on many agencies.

Background technology

Power distribution network has closed loop design, open loop operation characteristic, i.e. whole network radial operation.Power distribution network reconfiguration is exactly to guarantee the radial operation of network, and meet under the prerequisites such as feeder line thermal capacitance, branch current, node voltage, transformer capacity, by changing a large amount of block switches and the state of interconnection switch, so that a certain index of power distribution network reaches optimum state, index is generally the line loss of power distribution network, makes load balancing or supply power voltage quality etc.Power distribution network reconfiguration need to be processed a large amount of switching values, thereby it is a multiple target, non-linear hybrid optimization problem.Power distribution network reconfiguration is to optimize the important means of distribution system operation, is one of important content of power distribution automation research, for its research, has great significance.

Algorithm research for power distribution network reconfiguration has a lot, mainly can be divided into three major types.One is traditional optimization algorithm, as linear programming technique, dynamic programming etc., this class algorithm is mainly optimized distribution net work structure based on mathematical theory, can not rely on initial network configuration, can obtain globally optimal solution in theory, but along with the complexity day by day of network, algorithm exists the problem of " dimension calamity ".It two is heuritic approach, as branch exchange method, optimal flow pattern etc., this class algorithm speed is very fast, but be difficult to obtain globally optimal solution, it three has produced a large amount of novel intelligent algorithms along with artificial intelligence combines with Distributed Calculation, as particle cluster algorithm, genetic algorithm, ant group algorithm etc.In many intelligent algorithms, binary particle swarm algorithm application is comparatively extensive, its cardinal principle: each particle has speed and two of positions amount, and position is the potential solution of problem to be optimized, and every one dimension component of speed represents that the every one dimension component in position gets 1 probability.Therefore compare with particle cluster algorithm, binary particle swarm algorithm speed more new formula is constant, and position more new formula change to some extent, utilized sigmoid function, specific as follows shown in:

$\left\{\begin{array}{cc}{x}_{\mathrm{ij}}=1& \mathrm{if}(\mathrm{rand}\left(\right)\≤S\left(v_{\mathrm{ij}}\right))\\ x_{\mathrm{ij}}=0& \mathrm{if}(\mathrm{rand}\left(\right)>S\left(v_{\mathrm{ij}}\right))\\ S\left(v_{\mathrm{ij}}\right)=\frac{1}{[1+\exp (-v_{\mathrm{ij}})]}& \end{array}\right.$

In formula, bring every one dimension component of speed into sigmoid function, the random number (rand() on its value and [0,1]) compare, thus the value of every one dimension component of decision position.The realization of binary particle swarm algorithm in power distribution network reconfiguration problem is exactly the positional information of particle to be regarded as to a kind of combination of on off state of power distribution network, according to target function, determine individuality that particle is initial and the optimal value of population, by continuous iteration, the more speed of new particle and positional information, thereby more the individuality of new particle and the optimal value of population, finally obtain the solution that population optimal solution is power distribution network reconfiguration.This algorithm may produce infeasible solution, so according to the content of graph theory, obtain the radial criterion of power distribution network, judges that formula props up way=effective nodes-1 as closure.

Binary particle swarm algorithm itself exists the inherent shortcoming that is easily absorbed in locally optimal solution, and in the time of simultaneously in applying it to Complicated Distribution Network, it exists the long problem of coding.

Agency's (Agent) technology is a research branch of distributed artificial intelligence, its communication capacity has shown powerful advantage in research power distribution network reconfiguration problem, as shown in Figure 1, in certain distribution region, by Agent, send the control command of distribution terminal (FTU) in order to control the on off state in distribution network.Mainly comprise distribution system and distribution terminal FTU thereof, interface circuit mainly comprises distribution substation Agent module and isolation drive unit, is finally digital signal processor (DSP) control unit.Distribution substation Agent is connected with each FTU Agent by communication network, gather the information of each switching value in power distribution network, the output of distribution substation Agent module is connected with the input of DSP control unit, assurance is sent to the information of each switching value of power distribution network in DSP control unit, DSP control unit carries out C programming to algorithm for distribution network reconfiguration, and the information providing according to distribution substation Agent module produces one group of optimal solution.The output of DSP control unit is connected with the input of isolation drive unit, in order to the control signal output that optimal solution is changed into.The output of isolation drive unit is connected with the input of power distribution network substation Agent module, and control signal is exported to power distribution network substation Agent module.Power distribution network substation Agent module is assigned control command to each FTU Agent by communication network again.FTU Agent receives after control command by FTU self switch of controlling is operated.As shown in Figure 2, wherein the normal structure of FTU Agent, comprises data acquisition module, gathers the information of each switching value; Message processing module, the information of processing related switch amount is sent to output module and communication module; Output module, mainly returns to information to FTU; Communication module, mainly complete each intermodule and Agent between communication.Optimization in region is all undertaken by said process.Interregional structure optimization, mainly by each power distribution network substation Agent module, send the structure optimization information in regional to center control agents, center control agents cooperates with each power distribution network substation Agent, and whole distribution system is done to whole optimization.

In the research of this power distribution network reconfiguration, for the algorithm of reconstruct, be weaker, it has only utilized the communicativeness of Agent simultaneously, and other powerful character of Agent fail to bring into play.And for complicated power distribution network, the solution that Dan You center control agents and power distribution network substation Agent coordination produce is not accurate enough.

Summary of the invention

The object of the invention is to design a kind of power distribution network reconfiguration method and method thereof based on many agencies realizes, it combines the plurality of advantages of distributed artificial intelligence, also having overcome many deficiencies of existing intelligent algorithm, is the power distribution network reconfiguration implementation method based on many agencies that a kind of principle is simple and be easy to restrain.

The present invention solves its technical problem and takes following technical scheme to realize:

A power distribution network reconfiguration implementation method based on many agencies, method step is as follows:

(1) take different feeder lines is divided into several regions by network as unit, each region initialization several distribution terminals agency and a structural information that power distribution network substation is acted on behalf of, the structural information of whole electrical network initialization Yi Ge center control agent;

(2) whole proxy informations in each region of initialization, obtain speed and the positional information of each agency in method at random;

(3) calculate trend and network loss: for every group, act on behalf of corresponding network and do trend calculating, and calculate the loss of distribution network simultaneously;

(4) for acting on behalf of population, adopt the competition and cooperation of multi-agent system network machine-processed, all agencies are divided into the agency of triumph and failed agency;

(5) agency for triumph carries out self study operation, promotes agency itself and solves ability, makes algorithm convergence speedup speed, for failed agency, adopts binary particle swarm algorithm correction;

(6) calculate all agencies' appropriateness value, adopt the selection mechanism of the survival of the fittest to enter into iteration next time;

(7) judge whether iterations meets, if meet, carries out next step, turns back to step (3) if cannot meet;

(8) reconstruction result in output area is acted on behalf of to distribution substation;

(9) in the region that power distribution network substation agency obtains, optimal solution sends center control agent to as the initial value of the whole network reconstruct, and center control agent repeats the improvement algorithm steps of above (2)-(8), the optimal solution of output the whole network reconstruct.

And described step (1) is divided into several regions by network, each region initialization several distribution terminals agency and a power distribution network substation agency's structural information, the structural information of whole electrical network initialization Yi Ge center control agent; Specifically refer to, distribution terminal is acted on behalf of major control device action, accepting device information passes to the distribution terminal agency with layer, the transmission data of simultaneously also will communicating by letter with higher level, distribution substation agency can be looked at as an interface proxy, contacting between main responsible center control agent and bottom distribution terminal agency, simultaneously also as the control agent in a region, the reconstruct optimal solution of intra-zone is passed to center control agent, center control agent is mainly data and the optimal solution information of accepting each distribution substation agency, finally obtains the whole network reconstruction strategy.

And, in described step (4), adopt the competition and cooperation of multi-agent system network machine-processed, its particular content is: set up a trellis, each agency is at a grid the inside, L _size* L _sizethe sum that is grid is also to act on behalf of the number of acting on behalf of in population, and N is defined as and acts on behalf of a _ijneighborhood, act on behalf of a _ijneighbours form and to act on behalf of a _ijneighborhood, arbitrarily agency's competition and cooperation all launch in its neighborhood, the expression formula of N is:

$N=\{a_{i^1,j},a_{i,j^1},a_{i^2,j},a_{i,j^2}\}$

$i^1=\left\{\begin{array}{cc}i-1& i\≠1\\ L_{\mathrm{size}}& i=1\end{array}\right.;$ $i^2=\left\{\begin{array}{cc}i+1& i\≠L_{\mathrm{size}}\\ 1& i=L_{\mathrm{size}}\end{array}\right.$

$j^1=\left\{\begin{array}{cc}j-1& j\≠1\\ L_{\mathrm{size}}& j=1\end{array}\right.;$ $j^2=\left\{\begin{array}{cc}j+1& j\≠L_{\mathrm{size}}\\ 1& j=L_{\mathrm{size}}\end{array}\right.$

Wherein be respectively and act on behalf of a _ijfour neighbours agency's names, each agency has four neighbours agencies, and four neighbours agencies are at war with it, cooperation and knowledge sharing, now suppose to act on behalf of a _mto act on behalf of a _ijin four neighbours agencies, appropriateness value is minimum, if meet f (a _ij)≤f (a _m), act on behalf of a _ijin competition, be winner, its positional information remains unchanged, and carries out self study.If do not meet f (a _ij)≤f (m), acts on behalf of a _min competition, be loser, it need to adopt the speed of binary particle swarm algorithm and the more new formula of position to revise.

And, agency's the self study of winning in described step (5) operates concrete operation and is: the minimum search range of setting agency, with the agency of minimum appropriateness value in scope, replace current agency, each agency revises the appropriateness value of self, raising agency's quality itself in its own minimum search range like this.

And, the selection of survival of the fittest mechanism in described step (6), detailed process is: the fitness value that calculates each agency, agency's appropriateness value is higher, and it just more has more duplicator meeting, and contrary agency's duplicator can be fewer, even be faced with this agency and be eliminated, wherein adopt formula convert target function f to appropriate value function f' _i, then adopt formula p' _i=f _i'/∑ f _i' calculate each and act on behalf of selecteed probability.

And the iterations in described step (7) is that algorithm Program person sets.

And the reconstruction result in described step (8) output area, to distribution substation agency, specifically refers to, the result of output is the combination of one group of 0-1, the state of 0 representation switch is for opening, and the state of 1 representation switch is closed, then by distribution terminal, is acted on behalf of the change of the concrete state of control switch.

Advantage of the present invention and good effect are:

1. power distribution network reconfiguration method of the present invention, adopts the competition and cooperation mechanism of advanced multi-Agent well to overcome particle cluster algorithm itself and is easily absorbed in the shortcoming of locally optimal solution, and the system of implementing can Fast Convergent and obtained globally optimal solution comparatively accurately;

2. the present invention utilizes many agency's Distributed Calculation of (Multi-Agent) and the clear superiority of communicativeness, power distribution network reconfiguration problem is decomposed, the reconstruct of done subsystem respectively, the cooperated reconstruction task of whole complicated distribution of multi-Agent Cooperative, effectively raises rapidity and the correctness of reconstruct more

Accompanying drawing explanation

Fig. 1 is Agent system structural representation in current distribution region;

Fig. 2 is the normal structure schematic diagram of FTU Agent in prior art;

Fig. 3 is steps flow chart schematic diagram of the present invention;

Fig. 4 is multi-Agent layering schematic diagram in the present invention;

Fig. 5 is the competition and cooperation schematic diagram of mechanism that the present invention adopts Agent system network.

Embodiment

Below in conjunction with accompanying drawing, the embodiment of the present invention is further described, following examples are descriptive, are not determinate, can not limit protection scope of the present invention with this.

Basic principle

Whole distribution network is divided to some regions, cooperation and competition mechanism based on Agent in each region is improved binary particle swarm algorithm, reconstruct optimal solution in domain, these optimal solutions pass to center control agents as the initial value of interregional restructing algorithm by distribution substation Agent, interregionally select equally improved binary particle swarm algorithm.

A power distribution network reconfiguration implementation method based on many agencies, as shown in Figure 3, method step is as follows:

(1) take different feeder lines is divided into several regions by network as unit, the structural information of several distribution terminal agencies (FTU Agent) of each region initialization and a power distribution network substation Agent, the structural information of whole electrical network initialization Yi Ge center control agents.

(2) whole Agent information in each region of initialization, obtains at random each Agent(in method and regards particle as) speed and positional information;

(3) calculating trend and network loss: for every group of Agent(, regard particle as) corresponding network makes trend and calculates, and calculate the loss of distribution network simultaneously;

(4) for Agent population, adopt the competition and cooperation mechanism of Agent system network, all Agent are divided into the Agent of triumph and failed Agent;

(5) Agent for triumph carries out self study operation, promotes Agent itself and solves ability, makes algorithm convergence speedup speed, for failed Agent, adopts binary particle swarm algorithm to revise;

(6) calculate the appropriateness value of all Agent, adopt the selection mechanism of the survival of the fittest to enter into iteration next time;

(7) judge whether iterations meets, if meet, carries out next step, turns back to step (3) if cannot meet;

(8) reconstruction result in output area is to distribution substation Agent;

(9) in the region that power distribution network substation Agent obtains, optimal solution sends center control agents to as the initial value of the whole network reconstruct, and center control agents repeats the improvement algorithm steps of above (2)-(8), the optimal solution of output the whole network reconstruct.

In the concrete implementation step of the inventive method, in described step (1), divide network, the whole network configuration information of initialization, as shown in Figure 4, FTU Agent major control device action, accepting device information passes to the FTU Agent with layer, the transmission data of simultaneously also will communicating by letter with higher level, distribution substation Agent can be looked at as an interface Agent, contacting between main responsible center control agents and bottom FTU Agent, simultaneously also as the control agents in a region, the reconstruct optimal solution of intra-zone is passed to center control agents, center control agents is mainly data and the optimal solution information of accepting each distribution substation Agent, finally obtain the whole network reconstruction strategy,

In the concrete implementation step of the inventive method, in described step (4), adopt the competition and cooperation mechanism of Agent system network, its particular content is: as shown in Figure 5, set up a trellis, each Agent is at a grid the inside, L _size* L _sizethe sum that is grid is also the number of Agent in Agent population, and N is defined as Agent a _ijneighborhood, Agent a _ijneighbours form Agent a _ijneighborhood, arbitrarily the competition and cooperation of Agent all launch in its neighborhood, the expression formula of N is:

$N=\{a_{i^1,j},a_{i,j^1},a_{i^2,j},a_{i,j^2}\}$

$i^1=\left\{\begin{array}{cc}i-1& i\≠1\\ L_{\mathrm{size}}& i=1\end{array}\right.;$ $i^2=\left\{\begin{array}{cc}i+1& i\≠L_{\mathrm{size}}\\ 1& i=L_{\mathrm{size}}\end{array}\right.$

$j^1=\left\{\begin{array}{cc}j-1& j\≠1\\ L_{\mathrm{size}}& j=1\end{array}\right.;$ $j^2=\left\{\begin{array}{cc}j+1& j\≠L_{\mathrm{size}}\\ 1& j=L_{\mathrm{size}}\end{array}\right.$

Wherein be respectively Agent a _ijthe name of four neighbours Agent, each Agent has four neighbours Agent, and four neighbours Agent are at war with it, cooperation and knowledge sharing, now suppose Agent a _magent a _ijin four neighbours Agent, appropriateness value is minimum, if meet f (a _ij)≤f (a _m), Agent a _ijin competition, be winner, its positional information remains unchanged, and carries out self study.If do not meet f (a _ij)≤f (m), Agent a _min competition, be loser, it need to adopt the speed of binary particle swarm algorithm and the more new formula of position to revise.

In the concrete implementation step of the inventive method, the self study of Agent of winning in described step (5) operates concrete operation and is: as shown in Figure 5, set the minimum search range of Agent, with the Agent of minimum appropriateness value in scope, replace current Agent, each Agent revises the appropriateness value of self in its oneself minimum search range like this, improves the quality of Agent itself.

In the concrete implementation step of the inventive method, the selection of survival of the fittest mechanism in described step (6), detailed process is: as shown in Figure 5, calculate the fitness value of each Agent, the appropriateness value of Agent is higher, and it just more has more duplicator meeting, and the duplicator of contrary Agent can be fewer, even be faced with this Agent and be eliminated, wherein adopt formula convert target function f to appropriate value function f' _i, then adopt formula p' _i=f _i'/∑ f _i' calculate each and act on behalf of selecteed probability.

In the concrete implementation step of the inventive method, the iterations in described step (7) is that algorithm Program person sets, as long as rationally.

In the concrete implementation step of the inventive method, described step (8) Output rusults, the result of output is the combination of one group of 0-1, and the state of 0 representation switch is for opening, the state of 1 representation switch is closed, then by the change of the concrete state of FTU Agent control switch.

Claims (7)

1. the power distribution network reconfiguration implementation method based on many agencies, is characterized in that: method step is as follows:

(1) take different feeder lines is divided into several regions by network as unit, each region initialization several distribution terminals agency and a structural information that power distribution network substation is acted on behalf of, the structural information of whole electrical network initialization Yi Ge center control agent;

(2) whole proxy informations in each region of initialization, obtain speed and the positional information of each agency in method at random;

(3) calculate trend and network loss: for every group, act on behalf of corresponding network and do trend calculating, and calculate the loss of distribution network simultaneously;

(4) for acting on behalf of population, adopt the competition and cooperation of multi-agent system network machine-processed, all agencies are divided into the agency of triumph and failed agency;

(5) agency for triumph carries out self study operation, promotes agency itself and solves ability, makes algorithm convergence speedup speed, for failed agency, adopts binary particle swarm algorithm correction;

(6) calculate all agencies' appropriateness value, adopt the selection mechanism of the survival of the fittest to enter into iteration next time;

(7) judge whether iterations meets, if meet, carries out next step, turns back to step (3) if cannot meet;

(8) reconstruction result in output area is acted on behalf of to distribution substation;

(9) in the region that power distribution network substation agency obtains, optimal solution sends center control agent to as the initial value of the whole network reconstruct, and center control agent repeats the improvement algorithm steps of above (2)-(8), the optimal solution of output the whole network reconstruct.

2. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, it is characterized in that: described step (1) is divided into several regions by network, each region initialization several distribution terminals agency and a power distribution network substation agency's structural information, the structural information of whole electrical network initialization Yi Ge center control agent; Specifically refer to, distribution terminal is acted on behalf of major control device action, accepting device information passes to the distribution terminal agency with layer, the transmission data of simultaneously also will communicating by letter with higher level, distribution substation agency can be looked at as an interface proxy, contacting between main responsible center control agent and bottom distribution terminal agency, simultaneously also as the control agent in a region, the reconstruct optimal solution of intra-zone is passed to center control agent, center control agent is mainly data and the optimal solution information of accepting each distribution substation agency, finally obtains the whole network reconstruction strategy.

3. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, it is characterized in that: the competition and cooperation mechanism that adopts multi-agent system network in described step (4), its particular content is: set up a trellis, each agency is at a grid the inside, L _size* L _sizethe sum that is grid is also to act on behalf of the number of acting on behalf of in population, and N is defined as and acts on behalf of a _ijneighborhood, act on behalf of a _ijneighbours form and to act on behalf of a _ijneighborhood, arbitrarily agency's competition and cooperation all launch in its neighborhood, the expression formula of N is:

$N=\{a_{i^1,j},a_{i,j^1},a_{i^2,j},a_{i,j^2}\}$

$i^1=\left\{\begin{array}{cc}i-1& i\≠1\\ L_{\mathrm{size}}& i=1\end{array}\right.;$ $i^2=\left\{\begin{array}{cc}i+1& i\≠L_{\mathrm{size}}\\ 1& i=L_{\mathrm{size}}\end{array}\right.$

$j^1=\left\{\begin{array}{cc}j-1& j\≠1\\ L_{\mathrm{size}}& j=1\end{array}\right.;$ $j^2=\left\{\begin{array}{cc}j+1& j\≠L_{\mathrm{size}}\\ 1& j=L_{\mathrm{size}}\end{array}\right.$

Wherein be respectively and act on behalf of a _ijfour neighbours agency's names, each agency has four neighbours agencies, and four neighbours agencies are at war with it, cooperation and knowledge sharing, now suppose to act on behalf of a _mto act on behalf of a _ijin four neighbours agencies, appropriateness value is minimum, if meet f (a _ij)≤f (a _m), act on behalf of a _ijin competition, be winner, its positional information remains unchanged, and carries out self study.If do not meet f (a _ij)≤f (m), acts on behalf of a _min competition, be loser, it need to adopt the speed of binary particle swarm algorithm and the more new formula of position to revise.

4. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, it is characterized in that: the agency's that wins in described step (5) self study operates concrete operation and is: the minimum search range of setting agency, with the agency of minimum appropriateness value in scope, replace current agency, each agency revises the appropriateness value of self, raising agency's quality itself in its own minimum search range like this.

5. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, it is characterized in that: the selection of survival of the fittest mechanism in described step (6), detailed process is: the fitness value that calculates each agency, agency's appropriateness value is higher, it just more has more duplicator meeting, contrary agency's duplicator can be fewer, is even faced with this agency and is eliminated, and wherein adopts formula convert target function f to appropriate value function f' _i, then adopt formula p' _i=f _i'/∑ f _i' calculate each and act on behalf of selecteed probability.

6. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, is characterized in that: the iterations in described step (7) is that algorithm Program person sets.

7. the power distribution network reconfiguration implementation method based on many agencies according to claim 1, it is characterized in that: the reconstruction result in described step (8) output area is acted on behalf of to distribution substation, specifically refer to, the result of output is the combination of one group of 0-1, the state of 0 representation switch is for opening, the state of 1 representation switch is closed, then by distribution terminal, is acted on behalf of the change of the concrete state of control switch.