CN107465197A - A kind of var Optimization Method in Network Distribution based on dynamic particle cluster algorithm on multiple populations - Google Patents

Abstract

The present invention relates to a kind of var Optimization Method in Network Distribution based on dynamic particle cluster algorithm on multiple populations, belong to reactive power optimization of power system control technology field.This method performs following steps：1) population is initialized, generates one group of population at random;2) division is carried out to particle according to roulette algorithm and generates some subgroups；3) particle in 2) is updated using velocity location more new formula, and judges the quality of particle and particle before renewal after renewal;4) continue to operate to 3) middle, carry out R times；5) judge whether particle reaches maximum iterationc, if so, output optimal result；If not reaching, note current optimal particle is optimal result；6) judge whether the particle of optimal result changes after a suboptimization, if nothing, optimal particle is final result；2) continue to optimize if so, then returning.The method of the present invention solves the limitation of Reactive Power Optimazation Problem under the insurmountable multi-constraint condition of prior art, and the result that carries out voltage and reactive power optimization is more excellent, the speed of service faster.

Description

A Reactive Power Optimization Method for Distribution Network Based on Dynamic Multi-swarm Particle Swarm Algorithm

technical field

The invention relates to a reactive power optimization method of a distribution network based on a dynamic multi-swarm particle swarm algorithm, and belongs to the technical field of reactive power optimization control of electric power systems.

Background technique

The distribution network is an important link in the distribution of electric energy in the power system. With the rapid development of social economy, the system load is increasing day by day, the topological lines of the distribution network are becoming more and more complex, and the network loss of the power grid is also increasing. Reasonable allocation of reactive power in the distribution network plays a vital role in balancing the power flow distribution of the distribution network, reducing the network loss of the distribution network, and stabilizing the voltage at the central load.

Particle swarm optimization algorithm is an optimization algorithm for dealing with nonlinear optimization problems. It belongs to a kind of evolutionary algorithm. It has the advantages of fast convergence speed, simple calculation, high precision, and easy to obtain the global optimum. It has been widely used in signal processing, neural network training, and reactive power optimization. The basic particle swarm optimization algorithm is aimed at unconstrained single-objective or multi-objective optimization problems, but in practical applications, optimization goals often have various constraints. Because the basic particle swarm optimization algorithm is suitable for optimization problems without constraints, the search is blind, and it is difficult to obtain the optimal solution that satisfies the constraints. Therefore, the present invention proposes a dynamic multi-swarm particle swarm optimization algorithm for constrained condition optimization problems.

The invention patent document with the application number 201410222352.1 discloses a technical solution named "A Multi-objective Reactive Power Optimization Method Based on Adaptive Chaos Particle Swarm Algorithm". This patent solves the problem that the control variable may fall into a local optimal solution when dealing with multi-objective reactive power optimization problems, and reasonably solves the problem that the speed of solving the optimal value is too slow. However, the patented algorithm is relatively complex and has limitations in dealing with multi-constraint problems. Therefore, there is a need for a more comprehensive method that can obtain the optimal solution of the voltage and reactive power optimization problem in a short time and deal with multiple constraints, and can be widely used in engineering practice.

Contents of the invention

The technical problem to be solved by the present invention is to propose an optimal solution to the problem of voltage and reactive power optimization in a short period of time for the deficiencies of the existing technology, and to deal with multiple constraint conditions more comprehensively based on dynamic multi-swarm particle swarm optimization Algorithmic reactive power optimization method for distribution network.

The technical solution proposed by the present invention in order to solve the above-mentioned technical problems is: a reactive power optimization method of distribution network based on dynamic population particle swarm algorithm, which performs the following steps:

Step 1: Initialize the population and randomly generate a group of particle groups;

And calculate the fitness value f(x) of each particle and the degree value g _im (x) of the node i violating the constraint condition m, l is the total number of closed branches in the distribution network; is the network active loss of branch b;

g _i1 (x) = U _imax - U _i ; g _i2 (x) = U _i - U _imin ; g _i3 (x) = Q _{imax -} Q _i ; g _i4 (x) = Q _i -Q _imin ;

U _imin ≤ U _i ≤ U _imax ; Q _{imin ≤} Q _i ≤ Q _imax ;

U _i is the voltage of node i in the distribution network, U _imax is the upper limit of the node voltage amplitude allowed by the operation of the distribution network, U _imin is the lower limit of the node voltage amplitude allowed by the distribution network operation, and Q _i is the compensation of node i Q _imax sets the upper limit of reactive power compensation for node i, Q _imin sets the lower limit of reactive power compensation for node i, n is the number of load nodes, and Q _set is the compensation of the specified distribution network system The upper limit of the total reactive power capacity;

Step 2: Calculate the degree q _m of each particle violating the constraint conditions in the multi-dynamic population, and divide the particles according to the roulette algorithm to generate several subgroups;

Select optimized particles from the allocated subgroups through the strategy formula, the strategy formula is as follows,

x _subswarm (u,m)＝f _find [s _sort (G _mi (x),d)]

u=1,2,…,k, k is the total number of subswarms; x _subswarm( u,m) represents the particles in the uth subswarm that optimize the mth constraint condition; s _sort (G _mi (x),d ) means to sort G _mi (x) from small to large, G _mi (x) is the degree to which the particles in the population violate the mth constraint condition, and select the first d elements in G _mi (x), d is the set value, its size is less than the total number of particles in the subgroup (in the embodiment, half is selected as the value of d); the f _find function is used to find the position of the corresponding particle, so as to select the first d particles that violate the constraint degree less;

Step 3: Utilize the speed and position update formula to update the speed and position of the optimized particle in step 2, and at the same time calculate the fitness value f(x) of the updated particle and the value g _im (x) of the degree of violating the constraint condition m,

And compare the advantages and disadvantages of the particles after the update and the particles before the update according to the quality judgment rules,

If the updated particles are better, then otherwise

After comparing the pros and cons of the particles, in the subgroup, according to the pros and cons judgment rules, the regional optimal and with Compare the pros and cons, if better, newer for otherwise still for

Velocity position update formula,

with are the velocity and position of particle j in the e-dimensional space at the t+1 iteration; c ₁ and c ₂ are acceleration factors, which are non-negative constants; r ₁ and r ₂ are between [0,1]. A random positive real number between ; Find the position of the optimal value of the individual in the e-th dimension for particle j to the t-th iteration; is the optimal position of the subgroup;

Step 4: Continue to update the updated particles in step 3, update the R generation, if the R generation is not updated, return to step 2;

Step 5: Determine whether the particle reaches the maximum number of iterations c, if so, (output the optimal result; if it does not reach the maximum number of iterations c, record the current optimal particle as the optimal result.

Step 6: Determine whether the particle of the optimal result has not changed after a times of optimization, if there is no change, the optimal particle is the final result; if there is a change, skip to step 2 to continue optimizing the particle.

The improvement of the above-mentioned technical scheme is: the grouping of the roulette wheel algorithm in the step 2,

G _mi (x)={max{g _mi (x),0}, m=1,2,...,5};

Among them, G _mi (x) is the degree to which the particles in the population violate the mth constraint condition, x _j is the jth particle in the population x, and N is the total number of particles in the particle population.

The improvement of the above-mentioned technical scheme is: the good and bad judging rules in the step 3,

1.f _obj (a)=f _obj (b)=0, if f(x _a )<f(x _b ); particle a is better;

2.f _obj (a)＝f _obj (b)＝m, if g _im (x _a )＜g _im (x _b )or g _im (x _a )＝g _im (x _b )&&f(x _a )＜ f(x _b ); Particle a is better;

Define f _obj (x)=m, which means that the xth subgroup optimizes the m constraint condition; f _obj (x)=0, means that the xth subgroup optimizes the objective function, namely

The beneficial effect of adopting the above technical solution in the present invention is that in order to ensure that the degree of voltage offset meets the requirements for safe and stable operation of the system, it is necessary to set constraints on the voltage of each node. In order to improve the economy of the reactive power compensation system and avoid system reactive power surplus, there are certain constraints on the reactive power compensation capacity of each load node and the system as a whole.

In the basic particle swarm optimization algorithm, the flight speed of particles is determined by the individual optimal position and the global optimal position, while in the dynamic multi-swarm particle swarm optimization algorithm, since the particles are divided into different subgroups, the particle flight speed It will be determined by the optimal location of the individual and the optimal location of the subgroup.

In this way, the limitations of the particle swarm optimization algorithm in solving the reactive power optimization problem with multiple constraints are solved, and the dynamic particle swarm optimization algorithm has better results and faster running speed for voltage and reactive power optimization.

Description of drawings

The present invention will be further described below in conjunction with accompanying drawing.

Fig. 1 is a schematic flowchart of a reactive power optimization method for a distribution network based on a dynamic multi-swarm particle swarm algorithm according to an embodiment of the present invention.

Fig. 2 is the topology of distribution lines in a station area in the embodiment of the present invention.

detailed description

Example

A reactive power optimization method for distribution network based on dynamic population particle swarm algorithm in this embodiment

method, as shown in Figure 1, perform the following steps:

Step 1: Initialize the population and randomly generate a group of particle groups;

And calculate the fitness value f(x) of each particle and the degree value g _im (x) of the node i violating the constraint condition m, l is the total number of closed branches in the distribution network; is the network active loss of branch b, and the network loss can be obtained directly by using the calculation method of forward push back instead of power flow;

When using the dynamic multi-swarm particle swarm algorithm to deal with the reactive power optimization method of the distribution network, set the constraint conditions,

g _i1 (x)=U _imax −U _i ; g _i2 (x)=U _i −U _imin ; g _i3 (x)=Q _imax −Q _i ; g _i4 (x)=Q _i −Q _imin ;

In order to ensure that the degree of voltage offset meets the requirements for safe and stable operation of the system, it is necessary to set constraints on the voltage of each node. In order to improve the economy of the reactive power compensation system and avoid system reactive power surplus, there are certain constraints on the reactive power compensation capacity of each load node and the system as a whole. The constraints that the system needs to satisfy are as follows:

U _imin ≤ U _i ≤ U _imax ; Q _{imin ≤} Q _i ≤ Q _imax ;

U _i is the voltage of node i in the distribution network, U _imax is the upper limit of the node voltage amplitude allowed by the operation of the distribution network, U _imin is the lower limit of the node voltage amplitude allowed by the distribution network operation, and Q _i is the compensation of node i Q _imax sets the upper limit of reactive power compensation for node i, Q _imin sets the lower limit of reactive power compensation for node i, n is the number of load nodes, and Q _set is the compensation of the specified distribution network system The upper limit of the total reactive power capacity;

The idea of dynamic multi-swarm particle swarm algorithm is to dynamically group particles according to the difficulty of optimizing each constraint condition, that is, each particle is arranged to optimize a certain constraint condition.

Particles with the same task are arranged in the same subgroup, that is, a subgroup corresponds to optimizing a constraint condition.

Step 2: In order to objectively reflect the degree of difficulty for particles to optimize different constraint conditions, formula (10) defines the degree rate q _m of particles violating each constraint condition. It is used to reflect the degree of difficulty of particle optimization for different constraints.

Calculate the degree q _m of each particle violating the constraint conditions in the multi-dynamic population, and divide the particles according to the roulette algorithm to generate several subgroups, so as to group the particles objectively and reasonably.

According to the roulette strategy, the constraints that are more difficult to optimize will have more subgroups to optimize them. Define f _obj (x)=m, which means that the xth subgroup optimizes the m constraint condition; f _obj (x)=0, means that the xth subgroup optimizes the objective function, namely Since the particles for optimizing each constraint condition are randomly allocated, the degree to which the particle violates the constraint condition is also different. In order to improve the optimization ability of the population, the particles that violate the constraint condition m to a lesser extent should be selected to optimize the constraint condition m.

Select optimized particles from the allocated subgroups through the strategy formula, the strategy formula is as follows,

x _subswarm (u,m)＝f _find [s _sort (G _mi (x),d)]

u=1,2,…,k, k is the total number of subswarms; x _subswarm( u,m) represents the particles in the uth subswarm that optimize the mth constraint condition; s _sort (G _mi (x),d ) means to sort G _mi (x) from small to large, G _mi (x) is the degree to which the particles in the population violate the mth constraint condition, and select the first d elements in G _mi (x), d is the set value, its size is less than the total number of particles in the subgroup (in the embodiment, half is selected as the value of d); the f _find function is used to find the position of the corresponding particle, so as to select the first d particles that violate the constraint degree less;

Step 3: In the basic particle swarm optimization algorithm, the flying speed of the particles is determined by the individual optimal position and the global optimal position, while in the dynamic multi-swarm particle swarm optimization algorithm, since the particles are divided into different subgroups, the particle The flight speed of will be determined by the optimal location of the individual and the optimal location of the subgroup.

Use the speed and position update formula to update the speed and position of the optimized particle in step 2, and calculate the fitness value f(x) of the updated particle and the value g _im (x) of the degree of violating the constraint condition m at the same time,

And compare the advantages and disadvantages of the particles after the update and the particles before the update according to the quality judgment rules,

If the updated particles are better, then otherwise

After comparing the pros and cons of the particles, in the subgroup, according to the pros and cons judgment rules, the regional optimal and with Compare the pros and cons, if better, newer for otherwise still for

Velocity position update formula,

In the voltage and reactive power optimization problem, assuming that the total number of load nodes in the low-voltage distribution network is N, _{Xi = (x i1 ,} x _i2 ,…, x _iN ) is the position information of the i-th particle, representing the low-voltage distribution network The reactive power compensation capacity of each load node in the power grid, V _i =(v _i1 ,v _i2 ,...,v _iN ) is the velocity information of the i-th particle, which represents the correction amount of the position information.

with are the velocity and position of particle j in the e-dimensional space at the t+1 iteration; c ₁ and c ₂ are acceleration factors, which are non-negative constants; r ₁ and r ₂ are between [0,1]. A random positive real number between ; Find the position of the optimal value of the individual in the e-th dimension for particle j to the t-th iteration; is the optimal position of the subgroup.

Since r ₁ and r ₂ are random positive real numbers, the random results can be rounded up to ensure that the particle flight velocity is an integer.

Step 4: Continue to update the updated particles in step 3, update the R generation, if the R generation is not updated, return to step 2;

Step 5: Determine whether the particle reaches the maximum number of iterations c, if so, (output the optimal result; if it does not reach the maximum number of iterations c, record the current optimal particle as the optimal result.

Step 6: Determine whether the particle of the optimal result has not changed after a times of optimization. If there is no change, the optimal particle is the final result; if there is a change, skip to step 2 to continue optimizing the particle.

The grouping of the roulette algorithm in the step 2 of the present embodiment,

G _mi (x)={max{g _mi (x),0}, m=1,2,...,5};

Among them, G _mi (x) is the degree to which the particles in the population violate the mth constraint condition, x _j is the jth particle in the population x, and N is the total number of particles in the particle population.

Since the particles need to complete the dual optimization of the objective function and the constraint conditions, the quality of the particles cannot be judged only by the size of the fitness value, so the following rules are set to judge the quality of the particles: the rules for judging the quality of particles in step 3,

1.f _obj (a)=f _obj (b)=0, if f(x _a )<f(x _b ); particle a is better;

2.f _obj (a)＝f _obj (b)＝m, if g _im (x _a )＜g _im (x _b )or g _im (x _a )＝g _im (x _b )&&f(x _a )＜ f(x _b ); Particle a is better;

Define f _obj (x)=m, which means that the xth subgroup optimizes the m constraint condition; f _obj (x)=0, means that the xth subgroup optimizes the objective function, namely

In order to demonstrate the superiority of a distribution network reactive power optimization method based on the dynamic population particle swarm optimization algorithm in this embodiment, a comparison will be made in combination with actual cases below. Comprehensive optimization analysis of voltage and reactive power for distribution lines in a station area. The line topology diagram is shown in Figure 2.

The line has 11 nodes and 10 branches. Branch 1 is a S11-200 type distribution transformer with a voltage regulation range of ±5×2.5% U _N . Bit 100% U _N side. The grid reference capacity is selected as 200kVA, the voltage reference value of the high voltage side of the transformer is 10kV, and the voltage reference value of the low voltage side is 380V. Node 1 is selected as a balanced node with a voltage amplitude of 10.4kV and a phase angle of 0. The specific parameters of the system are shown in Table 1.

Table 1 System branch parameters

The power flow calculation of the line is carried out by using the forward push-back method, and the network loss of the system is obtained as 27.68kW. The voltage results of each node are shown in Table 2.

Table 2 Power flow calculation node voltage

It can be seen from Table 2 that the node voltage at the end of the line is already in the normal range of serious voltage. In order to ensure the safe, economical and stable operation of the power system, the constraints of the voltage and reactive power optimization problem are set as:

0.9≤U _i ≤1.05

0≤Q _i ≤60kvar

Comparing the results of the dynamic grouping particle swarm optimization algorithm with the results obtained by the basic particle swarm optimization algorithm, the results are as follows:

Table 3 Running speed and optimization results of different optimization algorithms

It can be seen from Table 3 that the network loss calculated by the dynamic multi-swarm particle swarm optimization algorithm is high, but the minimum voltage is also high, and there is a huge difference in the running time, that is, the redundancy of the method in this embodiment is far greater. Much smaller than the basic particle swarm algorithm.

It can be seen from the above table that the basic particle swarm optimization algorithm cannot solve the problem with constraints, and in order to obtain the result that meets the constraints, it will generate more than ten million redundant calculations, which greatly reduces the running speed and cannot meet the requirements of voltage and reactive power optimization. Timeliness requirements in questions.

The dynamic multi-swarm particle swarm algorithm, that is, the method in this embodiment, has a wide application prospect in engineering practice because of its simple algorithm, fast running speed, easy convergence, and the result of the voltage and reactive power optimization problem can be obtained in a short time.

The present invention is not limited to the above-described embodiments. All technical solutions formed by equivalent replacements fall within the scope of protection required by the present invention.

Claims (3)

1. a kind of var Optimization Method in Network Distribution based on dynamic demes particle cluster algorithm, it is characterised in that perform following steps：

Step 1：Population is initialized, generates one group of population at random；

And calculate the degree value g that the fitness value f (x) of each particle and node i in population violate constraints m_im(x),

<mrow> <mi>min</mi> <mi> </mi> <mi>f</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>b</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>l</mi> </munderover> <msubsup> <mi>P</mi> <mrow> <mi>l</mi> <mi>o</mi> <mi>s</mi> <mi>s</mi> </mrow> <mi>b</mi> </msubsup> </mrow>

L is the branch road sum closed in power distribution network；For branch road b network active loss；

g_i1(x)=U_imax-U_i；g_i2(x)=U_i-U_imin；g_i3(x)=Q_imax-Q_i；g_i4(x)=Q_i-Q_imin；

U_imin≤U_i≤U_imax；Q_imin≤Q_i≤Q_imax；

U_iFor the voltage of the node i of power distribution network, U_imaxTo run the node voltage amplitude upper limit of permission, U in power distribution network_iminFor with The node voltage amplitude lower limit that operation of power networks allows, Q_iFor the reactive power of the compensation of node i, Q_imaxFor node i setting compensation without The upper limit of work(, Q_iminFor the lower limit that node i setting compensation is idle, n is the number of load bus, Q_setFor defined power distribution network system The upper limit of the compensating reactive power total capacity of system；

Step 2：Calculate each particle in more dynamic demes and run counter to constraints degree q_m, and according to roulette algorithm to grain Son carries out division and generates some subgroups；

Optimization particle is chosen by tactful formula from the subgroup after distribution, the tactful formula is as follows,

x_subswarm(u, m)=f_find[s_sort(G_mi(x),d)]

U=1,2 ..., k, k are subgroup sum；x_subswarm(u, m) represents to optimize m-th of constraints in u-th of subgroup Particle；s_sort(G_mi(x), d) represent to G_mi(x) it is ranked up from small to large, G_mi(x) violated m-th for the particle in population The degree of constraints, and choose G_mi(x) preceding d element, d are setting value in, and its size is less than total number of particles in subgroup and (implemented Value of the half as d is chosen in example)；f_findFunction is used for the position for finding corresponding particle, so as to select run counter to degree of restraint compared with Small preceding d particle；

Step 3：The speed for optimizing particle in step 2 is updated with position using velocity location more new formula, calculated simultaneously Go out the fitness value f (x) of particle and violation constraints m degree value g after updating_im(x),

And compare the quality of particle and particle before renewal after renewal according to good and bad judgment rule,

If particle is more excellent after renewal,Otherwise

After the quality of completeer particle, it is optimal in subgroup according to good and bad judgment rule to seek regionAnd withIt is more excellent It is bad, ifIt is more excellent, renewalForOtherwiseRemain as

Velocity location more new formula,

WithSpeed and position of the respectively particle j in the t+1 times iteration in e dimension spaces；c₁、c₂For accelerated factor, Take nonnegative constant；r₁、r₂For the random arithmetic number between [0,1]；Looked for for particle j untill the t times iteration in e dimensions Position to where the optimal value of individual；For the position at the optimal place in subgroup；

Step 4：Continue to be updated the particle after updating in step 3, update R generations, if not updating R generations, return to step 2；

Step 5：Judge whether particle reaches maximum iteration c, if so, output optimal result；If not up to greatest iteration time Number c, note current optimal particle is optimal result；

Step 6：Judge whether the optimal particle does not change after by a suboptimization, it is described optimal if not changing Particle is final result；If changing, jump to step 2 and continue to optimize.

2. the var Optimization Method in Network Distribution according to claim 1 based on dynamic demes particle cluster algorithm, its feature exist In：The packet of roulette algorithm in step 2,

<mrow> <mi>f</mi> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mover> <mi>q</mi> <mo>&amp;OverBar;</mo> </mover> <mo>,</mo> <msub> <mi>q</mi> <mi>m</mi> </msub> <mo>=</mo> <mo>&amp;lsqb;</mo> <msub> <mi>q</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>q</mi> <mn>2</mn> </msub> <mo>,</mo> <mo>...</mo> <mo>,</mo> <msub> <mi>q</mi> <mn>5</mn> </msub> <mo>&amp;rsqb;</mo> <mo>,</mo> <msub> <mi>p</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <mrow> <mover> <mi>q</mi> <mo>&amp;OverBar;</mo> </mover> <mo>&amp;CenterDot;</mo> <msub> <mi>q</mi> <mi>m</mi> </msub> </mrow> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>m</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>5</mn> </munderover> <msub> <mi>q</mi> <mi>m</mi> </msub> </mrow> </mfrac> <mo>,</mo> <mi>m</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mn>5</mn> <mo>;</mo> </mrow>

<mrow> <msub> <mi>q</mi> <mi>m</mi> </msub> <mo>=</mo> <mfrac> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>G</mi> <mrow> <mi>m</mi> <mi>i</mi> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>m</mi> <mo>=</mo> <mn>1</mn> </mrow> <mn>5</mn> </munderover> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>G</mi> <mrow> <mi>m</mi> <mi>i</mi> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>,</mo> <mi>m</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mn>5</mn> </mrow>

G_mi(x)={ max { g_mi, 0 }, (x) m=1,2 ..., 5 }；

Wherein, G_mi(x) degree of m-th of constraints, x are violated for the particle in population_jFor j-th of particle in population x, N For total number of particles in population.

3. the var Optimization Method in Network Distribution according to claim 1 based on dynamic demes particle cluster algorithm, its feature exist In：Good and bad judgment rule in step 3,

1.f_obj(a)=f_obj(b)=0, if f (x_a) ＜ f (x_b)；Particle a is more excellent；

2.f_obj(a)=f_obj(b)=m, if g_im(x_a) ＜ g_im(x_b)or g_im(x_a)=g_im(x_b)&&f(x_a) ＜ f (x_b)；Particle A is more excellent；

Define f_obj(x)=m, represent that x-th of subgroup optimizes m-th of constraints；f_obj(x) x-th of subgroup optimization=0, is represented Object function, i.e.,