CN104102956B - Distribution network expansion planning method based on strategy adaption differential evolution - Google Patents

Abstract

The invention discloses a distribution network expansion planning method based on strategy adaption differential evolution. A capital time value is considered, calculation is carried out according to equal amount installment payment capital recovery, a model is established by taking least annual investment and operation maintenance cost as a target, a differential evolution algorithm based on strategy adaption on the basis of the model is proposed, an optimal solution obtained by the algorithm is decoded, a planning line is drawn on an electric power GIS (geographic information system) platform according to a line number obtained by decoding to carry out structure repair to an individual, and the individual is enabled to satisfy a radiative network structure. An adaption evolutionary strategy is continuously updated in an iteration process to prevent the algorithm from being trapped in local optimum, and a planning process of a distribution network line is enabled to exhibit better interactivity, more intuitive planning result and flexibility in planning scheme regulation by a GIS.

Description

A kind of distribution network planning method evolved based on tactful adaptive differential

Technical field

The present invention relates in a kind of power system distribution network Expansion Planning, more particularly to, it is a kind of based on strategy The distribution network planning method that adaptive differential is evolved.

Background technology

Distribution network planning referred on the basis of present situation electrical network analysis and future load forecast of distribution, from possible In power transformation station location and capacity, Connection Mode, feeder line number, path and model, an optimum or suboptimal design conduct are found out Extension modification scheme, makes investment, operation, maintenance, network loss and reliability failure costs sum minimum.

According to the different disposal method to economy and reliability index, the mathematical model of distribution network planning can divide For economical, reliable and comprehensive 3 class.

The object function of economical model only considers economic index, and fail-safe analysis is verified generally only as posteriority n-1. According to the difference of economic index, shipping model and least cost model can be further divided into.The former thinks all load moments When comprehensive minimum, the mode of connection is most short, with line power as control variable；The latter with invest recovery cost, equipment depreciation expense and Electric energy loss expense sum is object function, and the model more meets the requirement of economy in engineering compared with the former.

Reliable model is based on certain economic level, the plan model with reliability as target.

Comprehensive model is then, with certain conversion regime, reliability index to be converted to reliability benefit integrated economics Property model reliability cost composition object function.

In three of the above model, economy model has certain economic value, but reliability is typically poor.Reliability Model can embody the improvement of reliability index and the relation of fund input, but practicality is poor, is normally only used for local expansion Planning.Comprehensive model seeks balance is obtained between reliability cost and reliability benefit, so that distribution network extension rule Draw and reach global optimum, with higher comprehensive social benefit.

The content of the invention

Local optimum, calculating process are easily trapped into for distribution network planning easily to produce when solving with evolution algorithm The problems such as raw a large amount of infeasible solutions, the present invention propose it is a kind of be prevented effectively from be absorbed in local optimum, reliability it is good based on strategy The distribution network planning method that adaptive differential is evolved.

The technical solution adopted for the present invention to solve the technical problems is：

A kind of distribution network planning method evolved based on tactful adaptive differential, the distribution network planning Method is comprised the following steps：

A1, according to substation locating and sizing and the result of future load forecast of distribution, create load on electric power GIS platform Point, transformer station and branch road to be selected, built branch road element information, and set up the topology rule in corresponding point-point, point-line, line-face Then；

A2, the element information to creating carry out pretreatment：Load point and transformer substation sequence are numbered, branch road to be selected and built Branch road serial number, and start of record peripheral node numbering；Relief area is generated for geographic barrier, buffer zone analysis is carried out, is looked for Go out to be not suitable for the line corridor of the construction of line；

A3, consider time value on assets, by single-candidate pay separately capital recovery calculating, with year invest and operation and maintenance cost most It is little for target, object function such as formula (1)：

In formula：S₁、S₂, S be respectively newly-built sets of lines, built sets of lines and total line collection；ω be year equivalence recovery coefficient, λ The percentage ratio of investment cost is accounted for for maintenance, depreciation；l_iIt is the length of circuit i, f (D_i) for line footpath be D_iCircuit unit length Cost, X_iFor the decision variable on i-th line road, the circuit is selected as feeder line branch road, then X_i1 is taken, otherwise takes 0；g(D_i) for line Footpath is D_iCircuit resistivity；P_iIt is circuit i by power, U_NFor rated voltage；τ_maxMaximum loss time in year, d are single Position electricity price；

In differential evolution algorithm, dimension D is branch road quantity N to be selected_line, population quantity NP=10*D, mutation operator F be with The string of binary characters that machine is produced, crossover operator CR take 0.1, and end condition is the optimum for obtaining after tabu search algorithm It is individual identical with the optimum individual fitness before tabu search algorithm is switched into；Switching condition into tabu search algorithm is Iteration exceedes some generations and optimum individual continuous some generations do not change；The condition for exiting tabu search algorithm exceedes for step-length In some steps and optimum individual continuous some generations, do not change；

A4, reading element information, generating algorithm initial population is evaluated to initial population, and makes first of population Individual is that global optimum is individual；

A5, judge whether to reach end condition, if so, then evolving terminates, using optimum individual as solution output, go to step A10；If it is not, then continue algorithm, into step A6；

A6, judge whether to need to update policy selection probability, if so, then update probability；

A7, a kind of strategy is randomly choosed according to the select probability of Different Strategies enter row variation and crossover operation, and tied Structure constraint checking, if individuality is unsatisfactory for constraint, proceeds to step A11；If meet the constraint, into step A8；

A8, carry out selection operation, if the new individual after variation intersects can replace it is old it is individual enter of future generation, corresponding strategy When former generation number of success is cumulative 1 time, otherwise the frequency of failure is cumulative 1 time；

A9, the current population of traversal, find out optimum individual, judge whether current optimum individual is individual better than global optimum, if It is to replace, otherwise retains, afterwards into step A5；

A10, the optimal solution that algorithm is obtained is decoded, the circuit number obtained according to decoding is on electric power GIS platform Draw out planning circuit；

A11, structure repair is carried out to individuality so as to meet Radial network structure, reparation is completed, and proceeds to step A8.

Further, in step A4, read load point positional information and payload, the positional information of transformer station, Branch road to be selected, built branch road.The random length that generates is N_lineString of binary characters, and guarantee the quantity of " 1 " in character string For N_node(number of nodes), the function for applying mechanically (1) formula calculate each individual target function value.

Further, in step A6, it is 10 to arrange statistical algebra LP, and it is 15 to update algebraically, i.e., often experiencing for 15 generations changes In generation, just count and work as former generation the G failure of each strategy in 10 generations of backstepping, number of success forward, and calculate the selection of corresponding strategy Probability, such as formula (2), wherein (3), S_k,GRepresent strategy k G for when the probability of success；ns_k,g, nf_k,gStrategy k g are represented respectively Generation success, the number of times of failure；ε is 0.01, is 0 to prevent the tactful probability of success；p_k,GRepresent that the selection in strategy k G generations is general Rate；

Further, in step A7, (4), (5), (6) 3 kinds of Mutation Strategies are set.

v_i,j=x_r1,j+F·(x_r2,j-x_r3,j) (3)

v_i,j=x_i,j+F·(x_best,j-x_i,j)+F·(x_r1,j-x_r2,j)+F·(x_r3,j-x_r4,j) (4)

v_i,j=x_i,j+F·(x_r2,j-x_r3,j)+F·(x_r4,j-x_r5,j) (5)

Mutation operation is to each target individual X_i,G, i=1,2 ..., NP produce variation according to probability selection strategy individual. Wherein, randomly selected sequence number r in strategy₁, r₂, r₃, r₄, r₅It is different, and r₁, r₂, r₃, r₄, r₅With target individual sequence number i Also it is different；X in tactful (5)_best,jFor optimum individual.During three kinds tactful, arithmetic operator is logical operator, i.e., "-" represents Logic is or, " " expression logic XOR, "+" represents logical AND.

Then carry out crossover operation according to formula (7), (8), (9), wherein randb (j) between [0,1] it is random generate the J estimated value, rnbr (i) are a randomly selected sequence.

U_i,G+1=(U_1i,G+1,U_2i,G+1,L,U_Di,G+1) (6)

(i=1,2, L, NP, j=1,2, L, D) (8)

Connectedness, radioactivity structural constraint verification are carried out to individuality, if constraint is unsatisfactory for proceeding to step A11 carries out structure Repair.

In step A13, the individuality first to being unsatisfactory for radioactivity structure carries out number of branches reparation, makes N_branch= N_node-1；Then extreme saturation is individual sets, and recording individual isolated portions, if individuality has closed loop, record closed loop branch road；Time Go through after completing, interrupt a branch road in closed loop at random, and individual isolated portions are coupled together.

Beneficial effects of the present invention are mainly manifested in：Present invention application strategy when distribution network planning is carried out is adaptive Solution procedure should be avoided and be absorbed in local optimum, and the analysis of network with reference to GIS, buffer finish blasting optimized algorithm process, make planning Process is more directly perceived, convenient that manually result is adjusted.

Description of the drawings

Fig. 1 is the flow chart of the distribution network planning method evolved based on tactful adaptive differential.

Fig. 2 is power transformation station location and load point distribution.

Fig. 3 is initial distribution network.

The optimization distribution line that Fig. 4 is obtained after being carried out algorithm.

Specific embodiment

The invention will be further described below in conjunction with the accompanying drawings.

With reference to Fig. 1～Fig. 4, a kind of distribution network planning method evolved based on tactful adaptive differential, such as Fig. 1 institutes Show, wherein comprising the steps of：

A1, according to substation locating and sizing and future load forecast of distribution result, create relevant factor in electric power GIS platform Information；

A2, the element information to creating carry out pretreatment；

A3, determine object function, the corresponding control parameter of algorithm and the specific strategy for being adopted are set；

A4, reading element information, generating algorithm initial population is evaluated to initial population, and makes first of population Individual is that global optimum is individual；

A5, judge whether to reach end condition, if so, then evolving terminates, using optimum individual as solution output, go to step A10；If it is not, then continue algorithm, into step A6；

A6, judge whether to need to update policy selection probability, if so, then update probability；

A7, a kind of strategy is randomly choosed according to the select probability of Different Strategies enter row variation and crossover operation, and tied Structure constraint checking, if individuality is unsatisfactory for constraint, proceeds to step A11；If meet the constraint, into step A8；

A8, carry out selection operation, if the new individual after variation intersects can replace it is old it is individual enter of future generation, corresponding strategy When former generation number of success is cumulative 1 time, otherwise the frequency of failure is cumulative 1 time；

A9, the current population of traversal, find out optimum individual, judge whether current optimum individual is individual better than global optimum, if It is to replace, otherwise retains, afterwards into step A5；

A10, the optimal solution that algorithm is obtained is decoded, the circuit number obtained according to decoding is on electric power GIS platform Draw out planning circuit.

A11, structure repair is carried out to individuality so as to meet Radial network structure, reparation is completed, and proceeds to step A8；

Wherein, DE algorithms adopt binary coding, i.e. character string on each " 1 " represent that the branch road of corresponding numbering is selected In, " 0 " represents that the branch road of corresponding numbering is not chosen.

Described method, wherein in step A1, according to substation locating and sizing and the result of future load forecast of distribution, Create load point, transformer station and branch road to be selected, built branch road element information on electric power GIS platform, and set up corresponding point- Point, point-line, the topology rule in line-face.

Described method, wherein in step A2, all nodes (including load point and transformer station) serial number owns Branch road (including branch road to be selected and built branch road) serial number, and start of record peripheral node numbering；Generate for geographic barrier Relief area, carries out buffer zone analysis, finds out the line corridor for being not suitable for the construction of line, reduces algorithm dimension space, reduces blindly Search.

Described method, wherein in step A3, Optimized model considers time value on assets, pays separately capital by single-candidate and returns Receive and calculate, invested with year and the minimum target of operation and maintenance cost, object function such as formula (6).

In formula：S₁、S₂, S be respectively newly-built sets of lines, built sets of lines and total line collection；ω be year equivalence recovery coefficient, λ The percentage ratio of investment cost is accounted for for maintenance, depreciation；l_iIt is the length of circuit i, f (D_i) for line footpath be D_iCircuit unit length Cost, X_iDecision variable for i-th line road (selects the circuit as feeder line branch road, then X_i1 is taken, otherwise is taken 0)；g(D_i) for line Footpath is D_iCircuit resistivity；P_iIt is circuit i by power, U_NFor rated voltage；τ_maxMaximum loss time in year, d are single Position electricity price.In DE (Differential Evolution, differential evolution) algorithm, dimension D is branch road quantity N to be selected_line, population Quantity NP=10*D, mutation operator F is the string of binary characters for randomly generating, and crossover operator CR takes 0.1, and end condition is iteration Reached for 200 generations；

Described method, wherein in step A4, reading positional information and payload, the position of transformer station of load point Information, branch road to be selected, built branch road.The random length that generates is N_lineString of binary characters, and guarantee " 1 " in character string Quantity is N_node(number of nodes), the function for applying mechanically (1) formula calculate each individual target function value.

In step A6, it is 10 to arrange statistical algebra LP, and it is 15 to update algebraically, i.e., often experience 15 generation iteration, is just counted When former generation the G failure of each strategy in 10 generations of backstepping, number of success forward, and the select probability of corresponding strategy is calculated, such as formula (10), (11), wherein S_k,GRepresent strategy k G for when the probability of success；ns_k,g, nf_k,gRepresent respectively strategy k g generation successfully, The number of times of failure；ε is 0.01, is 0 to prevent the tactful probability of success；p_k,GRepresent the select probability in strategy k G generations；

In step A7, (12), (13), (14) 3 kinds of Mutation Strategies are set.

v_i,j=x_r1,j+F·(x_r2,j-x_r3,j) (11)

v_i,j=x_i,j+F·(x_best,j-x_i,j)+F·(x_r1,j-x_r2,j)+F·(x_r3,j-x_r4,j) (12)

v_i,j=x_i,j+F·(x_r2,j-x_r3,j)+F·(x_r4,j-x_r5,j) (13)

Mutation operation is to each target individual X_i,G, i=1,2 ..., NP produce variation according to probability selection strategy individual. Wherein, randomly selected sequence number r in strategy₁, r₂, r₃, r₄, r₅It is different, and r₁, r₂, r₃, r₄, r₅With target individual sequence number i Also it is different；；X in tactful (13)_best,jFor optimum individual.During three kinds tactful, arithmetic operator is logical operator, i.e. "-" table Show logic or, " " expression logic XOR, "+" represents logical AND.

Then crossover operation is carried out according to formula (15), (16), (17), wherein randb (j) is to generate between [0,1] at random J-th estimated value, rnbr (i) is a randomly selected sequence.

U_i,G+1=(U_1i,G+1,U_2i,G+1,L,U_Di,G+1) (14)

(i=1,2, L, NP, j=1,2, L, D) (16)

Connectedness, radioactivity structural constraint verification are carried out to individuality, if constraint is unsatisfactory for proceeding to step A11 carries out structure Repair.

In step A13, the individuality first to being unsatisfactory for radioactivity structure carries out number of branches reparation, makes N_branch= N_node-1；Then extreme saturation is individual sets, and recording individual isolated portions, if individuality has closed loop, record closed loop branch road；Time Go through after completing, interrupt a branch road in closed loop at random, and individual isolated portions are coupled together.

Embodiment

The present embodiment is a real system with 25 nodes and 42 10kv power distribution networks that can extend branch road, electric power GIS platform manages GIS information systeies using electric lines of force, and the system is on probation in the ground such as Shandong Yantai input, and operation reflection is good.

Node data as shown in table 1, capacity 0 for transformer station；Branch data is as shown in table 2.Voltage landing constraint is arranged 0.105, λ is taken for 5%, ω and takes 0.05, τ_max10000h is taken, d takes 0.5.

Node key element such as Fig. 2 is created on electric lines of force management GIS information system platforms, solid dot is load, and hollow dots are Transformer station, creates circuit key element such as Fig. 3.Carry out after computing using algorithm, obtain the distribution line cabling scenario such as Fig. 4.Table 1 is Node data, 2 feeder line data of table.

Table 1

Branch road number	Starting point	Terminal	Resistance (Ω)	Reactance (Ω)	Cost (unit)	Capacity (A)	Length (km)
								1	1	2	2.52	0.84	106500	90	2.1
2	1	3	1.98	0.66	99750	90	1.65

3	1	42	2.64	0.88	10800	90	2.2
								4	2	5	4.2	0.8	30000	125	2
5	2	6	3.15	0.6	22500	125	1.5
								6	3	6	3.68	0.7	26250	125	1.75
7	3	7	3.68	0.7	26250	125	1.75
								8	4	7	3.68	0.7	26250	125	1.75
9	4	8	2.1	0.4	15000	125	1
								10	4	12	2.1	0.4	15000	125	1
11	5	9	2.63	0.5	18750	125	1.25
								12	6	9	3.15	0.6	22500	125	1.5
13	6	10	3.68	0.7	26250	125	1.75
								14	7	10	4.2	0.8	30000	125	2
15	7	11	4.2	0.8	30000	125	2
								16	7	8	3.68	0.7	26250	125	1.75
17	9	15	2.63	0.5	18750	125	1.25
								18	9	10	3.68	0.7	26250	125	1.75
19	10	14	3.68	0.7	26250	125	1.75
								20	10	13	5.78	1.1	41250	125	2.75
21	11	13	3.68	0.7	26250	125	1.75
								22	1	16	1.8	0.6	97500	90	1.5
23	2	16	2.21	0.42	15750	125	1.05
								24	16	17	1.58	0.3	11250	125	0.75
25	2	17	2.21	0.42	15750	125	1.05
								26	5	17	2.1	0.4	15000	125	1
27	17	18	3.15	0.6	22500	125	1.5
								28	5	18	1.58	0.3	11250	125	0.75
29	15	18	2.63	0.5	18750	125	1.25
								30	1	19	1.86	0.62	98250	90	1.55
31	4	19	2.1	0.4	15000	125	1
								32	19	20	1.58	0.3	11250	125	0.75
33	12	20	1.58	0.3	11250	125	0.75
								34	12	21	1.05	0.2	7500	125	0.5
35	21	22	1.05	0.2	7500	125	0.5
								36	8	23	2.21	0.42	15750	125	1.05
37	11	23	1.05	0.2	7500	125	0.5
								38	8	22	1.37	0.26	9750	125	0.65
39	3	24	1.58	0.3	11250	125	0.75
								40	9	25	0.95	0.18	6750	125	0.45
41	14	25	1.05	0.2	7500	125	0.5
								42	4	24	0.84	0.16	6000	125	0.4

Table 2

Every loop line road be can be seen that from wiring result the supply district of clear and non-overlapping copies, mounts per loop line road Load capacity also meet the restriction of circuit maximum capacity, it is seen that its result is rational, and meets engineering reality.The present embodiment Showing, method of the present invention is a feasible cabling scenario more effectively to be calculated under certain calculation scale, And intuitively reflect result on electric power GIS platform, so as to improve planning personnel's work efficiency.

Claims (3)

1. it is a kind of based on tactful adaptive differential evolve distribution network planning method, it is characterised in that：The power distribution network Network Expansion Planning method is comprised the following steps：

A1, according to substation locating and sizing and the result of future load forecast of distribution, on electric power GIS platform create load point, Transformer station and branch road to be selected, built branch road element information, and set up corresponding point-point, point-line, the topology rule in line-face；

A2, the element information to creating carry out pretreatment：By load point and transformer substation sequence numbering, branch road to be selected and built branch road Serial number, and start of record peripheral node numbering；Relief area is generated for geographic barrier, buffer zone analysis is carried out, is found out not The line corridor of the suitable construction of line；

A3, consideration time value on assets, pay separately capital recovery calculating by single-candidate, invested with year and operation and maintenance cost is minimum Target, object function such as formula (1)：

$\min F=\underset{i\∈S_1}{\Σ}(\mathrm{\ω}+\mathrm{\λ})l_{i}f\left(D_i\right)X_i+\underset{i\∈S_2}{\Σ}{\mathrm{\λl}}_{i}f\left(D_i\right)+\underset{i\∈S}{\Σ}{l}_{i}g\left(D_i\right)\frac{P_i^2}{U_N^2}{\mathrm{\τ}}_{max}d---\left(1\right)$

In formula：S₁、S₂, S be respectively newly-built sets of lines, built sets of lines and total line collection；ω is year equivalence recovery coefficient, and λ is dimension Repair, depreciation accounts for the percentage ratio of investment cost；l_iIt is the length of circuit i, f (D_i) for line footpath be D_iCircuit unit length cost, X_iFor the decision variable on i-th line road, the circuit is selected as feeder line branch road, then X_i1 is taken, otherwise takes 0；g(D_i) for line footpath be D_i Circuit resistivity；P_iIt is circuit i by power, U_NFor rated voltage；τ_maxFor the maximum loss time in year, d is unit electricity Valency；

In differential evolution algorithm, dimension D is branch road quantity N to be selected_line, population quantity NP=10*D, mutation operator F are random product Raw string of binary characters, crossover operator CR take 0.1, and end condition is the optimum individual for obtaining after tabu search algorithm It is identical with the optimum individual fitness before tabu search algorithm is switched into；Switching condition into tabu search algorithm is iteration More than some generations and optimum individual continuous some generations do not change；The condition for exiting tabu search algorithm is step-length more than some Walk and optimum individual continuous some generations do not change；

A4, reading element information, generating algorithm initial population is evaluated to initial population, and makes an individual for population It is individual for global optimum；

A5, judge whether to reach end condition, if so, then evolving terminates, using optimum individual as solution output, go to step A10；If It is no, then continue algorithm, into step A6；

A6, judge whether to need to update policy selection probability, if so, then update probability；

It is 10 to arrange statistical algebra LP, it is 15 to update algebraically, i.e., often experience 15 generation iteration, and just statistics is when former generation G backsteppings forward The failure of each strategy in 10 generations, number of success, and the select probability of corresponding strategy is calculated, such as formula (2), wherein (3), S_k,GTable Show tactful k G for when the probability of success；ns_k,g, nf_k,gStrategy k g generation successes, the number of times of failure are represented respectively；ε is 0.01, It is 0 to prevent the tactful probability of success；p_k,GRepresent the select probability in strategy k G generations；

$S_{k,G}=\frac{{\mathrm{\Σ}}_{g=G-LP}^{G-1}{\mathrm{ns}}_{k,g}}{{\mathrm{\Σ}}_{g=G-LP}^{G-1}{\mathrm{ns}}_{k,g}+{\mathrm{\Σ}}_{g=G-LP}^{G-1}{\mathrm{nf}}_{k,g}}+\mathrm{\ϵ}---\left(2\right)$

$p_{k,G}=\frac{S_{k,G}}{{\mathrm{\Σ}}_{k=1}^K{S}_{k,G}}---\left(3\right);$

A7, a kind of strategy is randomly choosed according to the select probability of Different Strategies enter row variation and crossover operation, and carry out structure about Beam is verified, if individuality is unsatisfactory for constraint, proceeds to step A11；If meet the constraint, into step A8；

(4), (5), (6) 3 kinds of Mutation Strategies are set,

v_i,j=x_r1,j+F·(x_r2,j-x_r3,j) (4)

v_i,j=x_i,j+F·(x_best,j-x_i,j)+F·(x_r1,j-x_r2,j)+F·(x_r3,j-x_r4,j) (5)

v_i,j=x_i,j+F·(x_r2,j-x_r3,j)+F·(x_r4,j-x_r5,j) (6)

Mutation operation is to each target individual X_i,G, i=1,2 ..., NP, according to probability selection strategy generation variation individuality, wherein, Randomly selected sequence number r in strategy₁, r₂, r₃, r₄, r₅It is different, and r₁, r₂, r₃, r₄, r₅With target individual sequence number i also not Together；X in tactful (5)_best,jFor optimum individual, during three kinds tactful, arithmetic operator is logical operator, i.e., "-" represents logic Or, " " represents logic XOR, "+" represents logical AND；

Then crossover operation is carried out according to formula (7), (8), (9), wherein randb (j) is random j-th for generating between [0,1] Estimated value, rnbr (i) are a randomly selected sequence

U_{I, G+1=}(U_1i,G+1,U_2i,G+1,…,U_Di,G+1) (7)

$U_{ji,G+1}=\left\{\begin{array}{cc}{V}_{ji,G+1}& if\left(randb\right(j)\≤CR)orj=rnbr\left(i\right)\\ X_{ji,G+1}& if\left(randb\right(j)>CR)andj\≠rnbr\left(i\right)\end{array}\right.---\left(8\right)$

(i=1,2 ..., NP, j=1,2 ..., D) (9)

Connectedness, radioactivity structural constraint verification are carried out to individuality, if constraint is unsatisfactory for step A11 is proceeded to and is carried out structure and repair It is multiple；

A8, carry out selection operation, if the new individual after variation intersects can replace it is old it is individual enter of future generation, corresponding strategy is current Cumulative 1 time for number of success, otherwise the frequency of failure is cumulative 1 time；

A9, the current population of traversal, find out optimum individual, judge whether current optimum individual is individual better than global optimum, if then Replace, otherwise retain, afterwards into step A5；

A10, the optimal solution that algorithm is obtained is decoded, drawn on electric power GIS platform according to the circuit number that decoding is obtained Go out to plan circuit；

A11, structure repair is carried out to individuality so as to meet Radial network structure, reparation is completed, and proceeds to step A8.

2. a kind of distribution network planning method evolved based on tactful adaptive differential as claimed in claim 1, which is special Levy and be：In step A4, read load point positional information and payload, the positional information of transformer station, branch road to be selected, Built branch road, the random length that generates is N_lineString of binary characters, and guarantee that the quantity of " 1 " in character string is N_node, apply mechanically (1) function of formula calculates each individual target function value.

3. a kind of distribution network planning method evolved based on tactful adaptive differential as claimed in claim 2, which is special Levy and be：In step A11, the individuality first to being unsatisfactory for radioactivity structure carries out number of branches reparation, makes N_branch= N_node-1；Then extreme saturation is individual sets, and recording individual isolated portions, if individuality has closed loop, record closed loop branch road；Time Go through after completing, interrupt a branch road in closed loop at random, and individual isolated portions are coupled together.