CN105932690A - Distribution network operation optimization method integrating reactive power optimization and network reconstruction - Google Patents

Abstract

The invention discloses a distribution network operation optimization method integrating reactive power optimization and network reconstruction. According to the method, coordination and cooperation of a reactive power optimization technology and a network reconstruction technology in distribution network operation optimization are fully considered. An optimization model employing annual comprehensive cost as a target function is firstly defined; a new coding strategy and an identification method of a solution are provided; and finally the best comprehensive optimization scheme is determined by a shuffled frog leaping algorithm. By the distribution network operation optimization method, the potential economic value existing in cooperation of the reactive power optimization technology and the network reconstruction technology can be fully explored; and the distribution network operation optimization method has certain practical reference value.

Description

A kind of comprehensive idle work optimization and the distribution running optimizatin method of network reconfiguration

Technical field

The invention belongs to Operation of Electric Systems analysis and control technical field, relating to a kind of comprehensive idle work optimization and net The distribution running optimizatin method of network reconstruct.

Technical background

Two important technical that Distribution system and distribution idle work optimization run as distribution optimization are distribution from The key technology of dynamicization, it is ensureing the quality of power supply, is reducing the aspect important roles such as via net loss.Distribution Reconstruct and obtain the network topology structure under optimum optimization desired value by the Guan Bi changing network switching；Distribution without Merit optimization is typically minimized by fixed point switching reactive-load compensation equipment and ensures higher electricity realizing active loss Voltage levels.Substantially, Distribution system is nonlinear combinatorial optimization problem, and distribution idle work optimization is non-linear integer Planning problem, the complex optimum of the two makes solving of problem more complicated, for this problem, has scholar respectively Propose to use the method for first reconstruct post-compensation alternating iteration, be main being optimized with reconstruct and idle work optimization respectively, The latter have employed intelligent algorithm and solves, and compares former alternative iteration method, improves computational accuracy, and they are not It is truly to synchronize to be reconstructed and idle work optimization, additionally, existing research is all using network loss as optimization aim, Not considering the economy of reactive-load compensation, the potential value of complex optimum is not well studied and excavates. Thus the invention discloses a kind of distribution running optimizatin method of comprehensive idle work optimization and network reconfiguration.

Summary of the invention

Goal of the invention: present invention aims to the deficiencies in the prior art, it is proposed that a kind of comprehensive idle excellent Change the distribution running optimizatin method with network reconfiguration, set up the complex optimum mould with year comprehensive cost as object function Type, and in solution procedure, simplify network, improve the identification solved, obtain comprehensive finally by the algorithm that leapfrogs Prioritization scheme, provides decision support for operations staff.

Technical scheme: the present invention provides a kind of distribution running optimizatin method of comprehensive idle work optimization and network reconfiguration, Comprise the following steps:

Step 1: set up using year comprehensive cost as the Integrated Optimization Model of object function；

Step 2: give the coding strategy and the recognition methods of solution made new advances；

Step 3: use shuffled frog leaping algorithm to try to achieve optimal complex optimum scheme.

Further, in described step 1, Integrated Optimization Model is:

$\min CT={\mathrm{\λT}}_{max}{P}_{loss}+(k_1+k_2)(C_1\underset{i=1}{\overset{k_3}{\Σ}}{W}_i+k_3{C}_2)$

In formula: CT is year comprehensive cost, λ is electricity price；T_maxHourage is lost for annual peak load；k₁For mending Repay the year maintenance cost rate of equipment；k₂For coefficient of investment recovery；W_iFor the reactive-load compensation amount of i-th node, C₁ Price for reactive-load compensation；C₂For the mounting cost of single compensation point, k₃For reactive-load compensation point number, P_lossFor The active loss of network, its size is affected with reactive-load compensation amount by network topology structure, and its value is every circuit The summation of active loss, expression formula is as follows:

$P_{loss}=\underset{k=1}{\overset{N_b}{\mathrm{\Σ}}}{H}_k{R}_k\frac{{P_k}^2+{Q_k}^2}{{V_k}^2}$

In formula, N_bRepresent circuitry number；R_kRepresent the resistance of branch road k；H_kRepresenting branch road closure state, 1 represents Guan Bi, 0 expression is opened；P_kRepresent the active power of branch road k；Q_kRepresent the reactive power of branch road k；V_kTable Show the terminal voltage of branch road k；

In Load flow calculation, except keeping network power balance external, in addition it is also necessary to consider following constraints:

V^min≤V_j≤V^max

$\left|S_k\right|\≤S_k^{\max }$

0≤W_i≤W_i,max

In formula, V^min, V^maxRepresent the bound of node voltage when distribution is properly functioning respectively；S_kRepresent branch road k Current-carrying capacity, S^maxRepresent the maximum carrying capacity of circuit k；W_i,maxRepresent the i-th compensation point compensation capacity upper limit.

Further, described step 2 comprises the following steps:

Step 201: network reduction and coding:

Distribution system obtains the network topology under optimum optimization desired value by the Guan Bi changing network switching and ties Structure, in order to improve the search efficiency solved, as a example by 33 node distributions, does following simplification: root saves to network Point is incorporated to 1 node；Owing to only allowing to disconnect a switch between node 2 and node 20, therefore can be by this Two internodal all branch roads are considered as a branch road group, by that analogy, simplify network, the network after simplification by 8 nodes, 12 branch road group compositions.Based on this, coding strategy based on independent loop circuit is used to have herein Effect ground reduces dimension, it is possible to increase the efficiency solved；

In shuffled frog leaping algorithm, every frog is equivalent to a prioritization scheme, and i-th frog is encoded to Z_i=[{ x₁,x₂,……,x_n1}；{y₁,y₂,……,y_n2], wherein, { x₁,x₂,…..,x_n1Represent what reconstruct disconnected Branch road numbering in branch road group.{y₁,y₂,……,y_n2Represent the reactive-load compensation amount selecting node, n1, n2 Represent branch road group and the number of reactive-load compensation point respectively；

Step 202: the identification of solution:

Utilizing degree of communication theory in graph theory that infeasible solution carries out identification, improving and making network is isolated island or looped network Solve so that solution meets the radial constraints of network, and the distribution network after equivalence can represent with a figure, its Each internodal annexation can represent with adjacency matrix A:

$A=\left[\begin{array}{cccc}{a}_{1,1}& a_{1,2}& ...& a_{1,M}\\ a_{2,1}& a_{2,2}& ...& a_{2,M}\\ \·& \·& & \·\\ \·& \·& & \·\\ \·& \·& & \·\\ a_{M,1}& a_{M,2}& ...& a_{M,M}\end{array}\right]$

Wherein, M is the node number of equivalent network.If node i is connected with j, then a_i,jIt is 1, is otherwise 0, structure Make the laplacian matrix B of figure:

B=diag (sum (A))-A

When rank (B)=M-1 equation is set up, and when the branch road disconnected is equal to the independent loop circuit of network, network Topological structure then meets radial requirement.

Further, the Algorithm for Solving process that leapfrogs in described step 3 includes following two step:

Step 301: global search

Step a: initiation parameter, including: quantity F of frog group；Quantity m of group；Rana nigromaculata in group Quantity n；Maximum allowable step-length S of beating_max；Globally optimal solution P_z；Locally optimal solution P_b；The most worst solution P_w； Global iterative evolution times N_g, local iteration's evolution times N_l, each compensation point reactive-load compensation upper limit W_i,max；

Step b: stochastic generation initial frog group, is calculated the evaluation of estimate of each frog by object function；

Step c: carry out ascending sort according to evaluation of estimate size, record optimal solution P_z, and frog group is pressed In the following manner is divided into group: the 1st frog puts into the 1st group, and the 2nd frog puts into the 2nd group, m The frog puts into m-th group, and the m+1 frog puts into the 1st group, by that analogy, until all frogs are put Into appointment position；

Step d: according to the following formula each group is carried out evolutional operation

SL=ceil (Rand () × (P_w-P_b))

NewP_w=P_w+SL,-SL_min≤SL≤SL_max

In formula, ceil represents and rounds, and rand () represents the random number producing 0～1, and SL represents the step-length leapfroged, SL_max, L_minStep-length for leapfroging limits, NewP_wRepresent the P after updating_w；

Step e: after all groups update, calculates the evaluation of estimate of all frogs in frog group；

Step f: judge whether to meet stop condition.If meeting, stopping search, otherwise forwarding step c to；

Step 302: Local Search, refers to the specific descriptions launching above-mentioned steps d, and process is as follows:

Step d₁: set I_M=I_N=0, I_MRepresent the enumerator that group evolves, I_NRepresent Local Evolution enumerator；

Step d₂: select the P of current group_bAnd P_w, I_MAdd 1；

Step d₃: I_NAdd 1；

Step d₄: improve the worst frog in group according in step d two formulas；

Step d₅: if upper step improves the worst frog, then replace the worst frog with this new frog, otherwise use P_zSubstituted P in step d_b, again evolve；

Step d₆: if upper step still without improving the worst frog, then randomly generates a feasible solution to replace the worst frog；

Step d₇If: I_NLess than Local Evolution number of times L_N, then step d is proceeded to₃；

Step d₈If: I_MLess than group number m, then proceed to step d₂, otherwise enter step d of global search₅。

Operation principle: the present invention first defines using year comprehensive cost as the Optimized model of object function, then Give the coding strategy and the discrimination method of solution made new advances, finally, use shuffled frog leaping algorithm to try to achieve the most excellent Change scheme.The present invention can fully excavate idle work optimization and two technology of network reconfiguration and cooperate the potential warp existed Ji is worth, and has certain practical reference value.

Beneficial effect: compared with existing complex optimum technology, the invention have the advantages that and technique effect:

1) model that the method is announced can not only reduce network loss, lifting section point voltage further, the most fully examines Consider the economy of reactive-load compensation so that distribution complex optimum more conforms to economy principle, i.e. fully excavated The economic worth of complex optimum, also provides new evaluation index for synthesis optimizing and scheduling.

2) except proposing new optimization aim, present invention employs and should be readily appreciated that, be easily programmed realization and optimizing The shuffled frog leaping algorithm that ability is strong, and combine network reduction strategy based on separate branches method, it is possible to fast search To optimal solution, it is ensured that model solution efficiency.

Accompanying drawing explanation

Fig. 1 is distribution network simplification figure of the present invention；

Fig. 2 is general flow chart of the present invention；

Fig. 3 is the algorithm flow chart that leapfrogs in the present invention；

Fig. 4 is node voltage scattergram of the present invention；

Fig. 5 is idle work optimization enabling objective value convergence curve of the present invention；

Fig. 6 is restructuring procedure desired value convergence curve of the present invention；

Fig. 7 is complex optimum enabling objective value convergence curve of the present invention.

Detailed description of the invention:

Below in conjunction with accompanying drawing and example, the enforcement of the present invention is described further, but the enforcement of the present invention and comprising It is not limited to this.

A kind of comprehensive idle work optimization and the distribution running optimizatin method of network reconfiguration, comprise the following steps:

Step 1: set up using year comprehensive cost as the Integrated Optimization Model of object function；

Step 2: give the coding strategy and the recognition methods of solution made new advances；

Step 3: use shuffled frog leaping algorithm to try to achieve optimal complex optimum scheme.

Further, in described step 1, Integrated Optimization Model is:

$\min CT={\mathrm{\λT}}_{max}{P}_{loss}+(k_1+k_2)(C_1\underset{i=1}{\overset{k_3}{\Σ}}{W}_i+k_3{C}_2)$

In formula: CT is year comprehensive cost, λ is electricity price；T_maxHourage is lost for annual peak load；k₁For mending Repay the year maintenance cost rate of equipment；k₂For coefficient of investment recovery；W_iFor the reactive-load compensation amount of i-th node, C₁ Price for reactive-load compensation；C₂For the mounting cost of single compensation point, k₃For reactive-load compensation point number, P_lossFor The active loss of network, its size is affected with reactive-load compensation amount by network topology structure, and its value is every circuit The summation of active loss, expression formula is as follows:

$P_{loss}=\underset{k=1}{\overset{N_b}{\mathrm{\Σ}}}{H}_k{R}_k\frac{{P_k}^2+{Q_k}^2}{{V_k}^2}$

In formula, N_bRepresent circuitry number；R_kRepresent the resistance of branch road k；H_kRepresenting branch road closure state, 1 represents Guan Bi, 0 expression is opened；P_kRepresent the active power of branch road k；Q_kRepresent the reactive power of branch road k；V_kTable Show the terminal voltage of branch road k；

In Load flow calculation, except keeping network power balance external, in addition it is also necessary to consider following constraints:

V^min≤V_j≤V^max

$\left|S_k\right|\≤S_k^{\max }$

0≤W_i≤W_i,max

In formula, V^min, V^maxRepresent the bound of node voltage when distribution is properly functioning respectively；S_kRepresent branch road k Current-carrying capacity, S^maxRepresent the maximum carrying capacity of circuit k；W_i,maxRepresent the i-th compensation point compensation capacity upper limit.

Further, described step 2 comprises the following steps:

Step 201: network reduction and coding:

Distribution system obtains the network topology under optimum optimization desired value by the Guan Bi changing network switching and ties Structure, in order to improve the search efficiency solved, as a example by 33 node distributions, does following simplification: root saves to network Point is incorporated to 1 node；Owing to only allowing to disconnect a switch between node 2 and node 20, therefore can be by this Two internodal all branch roads are considered as a branch road group, by that analogy, simplify network, the network after simplification by 8 nodes, 12 branch road group compositions.Based on this, coding strategy based on independent loop circuit is used to have herein Effect ground reduces dimension, it is possible to increase the efficiency solved；

In shuffled frog leaping algorithm, every frog is equivalent to a prioritization scheme, and i-th frog is encoded to Z_i=[{ x₁,x₂,……,x_n1}；{y₁,y₂,……,y_n2], wherein, { x₁,x₂,…..,x_n1Represent what reconstruct disconnected Branch road numbering in branch road group.{y₁,y₂,……,y_n2Represent the reactive-load compensation amount selecting node, n1, n2 Represent branch road group and the number of reactive-load compensation point respectively；

Step 202: the identification of solution:

Utilizing degree of communication theory in graph theory that infeasible solution carries out identification, improving and making network is isolated island or looped network Solve so that solution meets the radial constraints of network, and the distribution network after equivalence can represent with a figure, its Each internodal annexation can represent with adjacency matrix A:

$A=\left[\begin{array}{cccc}{a}_{1,1}& a_{1,2}& ...& a_{1,M}\\ a_{2,1}& a_{2,2}& ...& a_{2,M}\\ \·& \·& & \·\\ \·& \·& & \·\\ \·& \·& & \·\\ a_{M,1}& a_{M,2}& ...& a_{M,M}\end{array}\right]$

Wherein, M is the node number of equivalent network.If node i is connected with j, then a_i,jIt is 1, is otherwise 0, structure Make the laplacian matrix B of figure:

B=diag (sum (A))-A

When rank (B)=M-1 equation is set up, and when the branch road disconnected is equal to the independent loop circuit of network, network Topological structure then meets radial requirement.

Further, the Algorithm for Solving process that leapfrogs in described step 3 includes following two step:

Step 301: global search

Step a: initiation parameter, including: quantity F of frog group；Quantity m of group；Rana nigromaculata in group Quantity n；Maximum allowable step-length S of beating_max；Globally optimal solution P_z；Locally optimal solution P_b；The most worst solution P_w； Global iterative evolution times N_g, local iteration's evolution times N₁, each compensation point reactive-load compensation upper limit W_i,max；

Step b: stochastic generation initial frog group, is calculated the evaluation of estimate of each frog by object function；

Step c: carry out ascending sort according to evaluation of estimate size, record optimal solution P_z, and frog group is pressed In the following manner is divided into group: the 1st frog puts into the 1st group, and the 2nd frog puts into the 2nd group, m The frog puts into m-th group, and the m+1 frog puts into the 1st group, by that analogy, until all frogs are put Into appointment position；

Step d: according to the following formula each group is carried out evolutional operation

SL=ceil (Rand () × (P_w-P_b))

NewP_w=P_w+SL,-SL_min≤SL≤SL_max

In formula, ceil represents and rounds, and rand () represents the random number producing 0～1, and SL represents the step-length leapfroged, SL_max, L_minStep-length for leapfroging limits, NewP_wRepresent the P after updating_w；

Step e: after all groups update, calculates the evaluation of estimate of all frogs in frog group；

Step f: judge whether to meet stop condition.If meeting, stopping search, otherwise forwarding step c to；

Step 302: Local Search, refers to the specific descriptions launching above-mentioned steps d, and process is as follows:

Step d₁: set I_M=I_N=0, I_MRepresent the enumerator that group evolves, I_NRepresent Local Evolution enumerator；

Step d₂: select the P of current group_bAnd P_w, I_MAdd 1；

Step d₃: I_NAdd 1；

Step d₄: improve the worst frog in group according in step d two formulas；

Step d₅: if upper step improves the worst frog, then replace the worst frog with this new frog, otherwise use P_zSubstituted P in step d_b, again evolve；

Step d₆: if upper step still without improving the worst frog, then randomly generates a feasible solution to replace the worst frog；

Step d₇If: I_NLess than Local Evolution number of times L_N, then step d is proceeded to₃；

Step d₈If: I_MLess than group number m, then proceed to step d₂, otherwise enter step d of global search₅。

Operation principle: the present invention first defines using year comprehensive cost as the Optimized model of object function, then Give the coding strategy and the discrimination method of solution made new advances, finally, use shuffled frog leaping algorithm to try to achieve the most excellent Change scheme.The present invention can fully excavate idle work optimization and two technology of network reconfiguration and cooperate the potential warp existed Ji is worth, and has certain practical reference value.

Embodiment

The present invention is shown in accompanying drawing 2 using IEEE33 node system as explanation case, general flow chart, this system voltage etc. Level is 12.66kV, meritorious total load 3715kW, idle total load 2300kvar.For network reconfiguration, idle Optimize, three kinds of prioritization schemes of complex optimum are analyzed.

The algorithm parameter that leapfrogs is provided that frog group's size is 80, and group's number is 20, and overall situation evolution number of times is 50, Local Evolution number of times is 3, and flow chart is shown in accompanying drawing 3.Reactive-load compensation point selection load or burden without work the heaviest 23,24,29 Node, on the one hand meets reactive power compensation on the spot principle.Reactive-load compensation amount is with 1.2 times of works of the total load or burden without work of system For the upper limit of reactive-load compensation, using 10kvar as minimum step-size in search, if y₁∈ [0,60], y₂∈ [0,60], y₃∈[0,120].Other parameter: λ=0.5 yuan/kW h, T_max=5000h, k₁=0.13, k₂=0.1, C₁=60 Unit/kvar, C₂=5000 yuan/node.

Three kinds of scheme optimization Comparative result such as table 1 below, under each prioritization scheme, each node voltage of system is distributed such as accompanying drawing 4, solve iterative process target function value convergence situation such as accompanying drawing 5 to accompanying drawing 7.

Three kinds of scheme optimization Comparative result tables of table 1 present invention

In table 1, the contrast of each scheme optimum results shows, complex optimum scheme compares simple reconstruct and idle work optimization, Quality of voltage can be improved further: lowest section point voltage 0.9575p.u, compare individually reconstruct and idle work optimization Result be respectively increased 2.1%, 3.5%；Network active loss can be reduced further: active loss 102.09kW, the result comparing individually reconstruct and idle work optimization reduces 26.7%；Reduce year Integrated Cost further With: year comprehensive cost be 27.68 ten thousand, compare the result of individually reconstruct and idle work optimization reduce 20.5% respectively, 25.8%.Additionally, single reactive-load compensation is compared in the reactive-load compensation in complex optimum scheme, compensation dosage reduces 15.0%.Demonstrate reasonability and the effectiveness of the carried Integrated Optimization Model of the present invention.

Understanding with Fig. 4 in conjunction with table 1, independent fixed point idle work optimization and reconstruct can effectively reduce the meritorious damage of network Consumption, but promoting the system horizontal aspect of each node voltage, quality reconstruction is more preferable compared with idle work optimization.Complex optimum side Case compares reconstruct, and each node voltage of system is integrated with promoting further, thus demonstrates the conclusion of table 1 further.

From Fig. 5 to Fig. 7, SFLA during solving three Optimized models, there is good convergence effect, Solve the time all within 25 seconds, it is possible to Efficient Solution carried model herein.

To sum up, case illustrates correctness and the practicality of the present invention.

Claims (4)

1. a comprehensive idle work optimization and the distribution running optimizatin method of network reconfiguration, it is characterised in that: comprise the following steps:

Step 1: set up using year comprehensive cost as the Integrated Optimization Model of object function；

Step 2: give the coding strategy and the discrimination method of solution made new advances；

Step 3: use shuffled frog leaping algorithm to try to achieve optimal complex optimum scheme.

Comprehensive idle work optimization the most according to claim 1 and the distribution running optimizatin method of network reconfiguration, its feature It is: in described step 1, Integrated Optimization Model is:

$\min CT={\mathrm{\λT}}_{max}{P}_{loss}+(k_1+k_2)(C_1\underset{i=1}{\overset{k_3}{\Σ}}{W}_i+k_3{C}_2)$

In formula: CT is year comprehensive cost, λ is electricity price；T_maxHourage is lost for annual peak load；k₁For compensating equipment Year maintenance cost rate；k₂For coefficient of investment recovery；W_iFor the reactive-load compensation amount of i-th node, C₁For reactive-load compensation Price；C₂For the mounting cost of single compensation point, k₃For reactive-load compensation point number；P_lossFor the active loss of network, its Size is affected with reactive-load compensation amount by network topology structure, and its value is the summation of every circuit active loss, and expression formula is such as Under:

$P_{loss}=\underset{k=1}{\overset{N_b}{\mathrm{\Σ}}}{H}_k{R}_k\frac{{P_k}^2+{Q_k}^2}{{V_k}^2}$

In formula, N_bRepresent circuitry number；R_kRepresent the resistance of branch road k；H_kRepresent that branch road closure state, 1 expression close, 0 Expression is opened；P_kRepresent the active power of branch road k；Q_kRepresent the reactive power of branch road k；V_kRepresent the end of branch road k Voltage；

In Load flow calculation, except keeping network power balance external, in addition it is also necessary to consider following constraints:

V^min≤V_j≤V^max

$\left|S_k\right|\≤S_k^{\max }$

0≤W_i≤W_i,max

In formula, V^min, V^maxRepresent the bound of node voltage when distribution is properly functioning respectively；S_kRepresent the current-carrying capacity of branch road k, S^maxRepresent the maximum carrying capacity of circuit k；W_i,maxRepresent the i-th compensation point compensation capacity upper limit.

The comprehensive idle work optimization of kind the most according to claim 1 and the distribution running optimizatin method of network reconfiguration, it is special Levy and be: described step 2 comprises the following steps:

Step 201: network reduction and coding:

Distribution system obtains the network topology structure under optimum optimization desired value by the Guan Bi changing network switching, in order to Improve the search efficiency solved, as a example by 33 node distributions, network is done following simplification: root node is incorporated to 1 node； Owing to only allowing to disconnect a switch between node 2 and node 20, therefore can be by internodal for the two all branch roads Being considered as a branch road group, by that analogy, simplify network, the network after simplification is by 8 nodes, 12 branch road group compositions； Based on this, the present invention uses coding strategy based on independent loop circuit can efficiently reduce dimension, it is possible to increase solve Efficiency；

In shuffled frog leaping algorithm, every frog is equivalent to a prioritization scheme, and i-th frog is encoded to Z_i=[{ x₁,x₂,……,x_n1}；{y₁,y₂,……,y_n2], wherein, { x₁,x₂,…..,x_n1Represent that the branch road that reconstruct disconnects exists Numbering in branch road group.{y₁,y₂,……,y_n2Representing the reactive-load compensation amount selecting node, n1, n2 represent branch road group respectively Number with reactive-load compensation point；

Step 202: the identification of solution:

Utilize degree of communication theory in graph theory that infeasible solution carries out identification, improve the solution making network be isolated island or looped network so that Solution meets the radial constraints of network, and the distribution network after equivalence can represent with a figure, its each internodal connection Relation can represent with adjacency matrix A:

$A=\left[\begin{array}{cccc}{a}_{1,1}& a_{1,2}& ...& a_{1,M}\\ a_{2,1}& a_{2,2}& ...& a_{2,M}\\ \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}\\ \\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ a_{M,1}& a_{M,2}& ...& a_{M,M}\end{array}\right]$

Wherein, M is the node number of equivalent network；If node i is connected with j, then a_i,jIt is 1, is otherwise 0；Structural map Laplacian matrix B:

B=diag (sum (A))-A

When rank (B)=M-1 equation is set up, and when the branch road disconnected is equal to the independent loop circuit of network, the topology knot of network Structure then meets radial requirement.

A kind of comprehensive idle work optimization the most according to claim 1 and the distribution running optimizatin method of network reconfiguration, its It is characterised by: the described step 3 Algorithm for Solving process that leapfrogs includes following two step:

Step 301: global search, including following step by step:

Step a: initiation parameter, including quantity F of frog group；Quantity m of group；Quantity n of Rana nigromaculata in group；? Allow greatly step-length SL of beating_max；Globally optimal solution P_z；Locally optimal solution P_b；The most worst solution P_w；Global iterative is evolved secondary Number N_g, local iteration's evolution times N_l, each compensation point reactive-load compensation upper limit W_i,max；

Step b: stochastic generation initial frog group, is calculated the evaluation of estimate of each frog by object function；

Step c: carry out ascending sort according to evaluation of estimate size, record optimal solution P_z, and by frog group in the following manner Being divided into group: the 1st frog puts into the 1st group, the 2nd frog puts into the 2nd group, and the m frog puts into m-th Group, the m+1 frog puts into the 1st group, by that analogy, until all frogs are placed into appointment position；

Step d: according to the following formula each group is carried out evolutional operation

SL=ceil (Rand () × (P_w-P_b))

NewP_w=P_w+SL,-SL_min≤SL≤SL_max

Wherein, ceil represents and rounds, and rand () represents the random number producing 0～1, and SL represents the step-length leapfroged, SL_max, SL_minStep-length for leapfroging limits, NewP_wRepresent the P after updating_w；

Step e: after all groups update, calculates the evaluation of estimate of all frogs in frog group；

Step f: judge whether to meet stop condition, if meeting, stopping search, otherwise forwarding step c to；

Step 302: Local Search, refers to the specific descriptions launching above-mentioned steps d, as follows:

Step d₁: set I_M=I_N=0, I_MRepresent the enumerator that group evolves, I_NRepresent Local Evolution enumerator；

Step d₂: select the P of current group_bAnd P_w, I_MAdd 1；

Step d₃: I_NAdd 1；

Step d₄: improve the worst frog in group according in step d two formulas；

Step d₅: if upper step improves the worst frog, then replace the worst frog with this new frog, otherwise use P_zIn substituted step d P_b, again evolve；

Step d₆: if upper step still without improving the worst frog, then randomly generates a feasible solution to replace the worst frog；

Step d₇If: I_NLess than Local Evolution number of times L_N, then step d is proceeded to₃；

Step d₈If: I_MLess than group number m, then proceed to step d₂, otherwise enter step d of global search₅。