CN106250985A - A kind of improvement heredity film method solving distribution network structure multistage programming - Google Patents

Abstract

The invention discloses a kind of improvement heredity film method solving distribution network structure multistage programming, Revised genetic algorithum is introduced in the computation model of membranous system, for calculating the distribution network structure plan model containing distributed power source with the minimum target of overall life cycle cost value, have employed in solving the transmission of object and evolving intermembranous exchange, the operating mechanism such as self adaptation coordinated transposition and TSP question rewriting, and use Adjacent Matrix Method to produce the most feasible initial solution and the infeasible solution filtering in calculating.What the present invention was formed improves heredity film algorithm for optimizing the calculating distribution network structure plan model containing distributed power source with the minimum target of overall life cycle cost value containing investment, O&M, power failure, environmental protection and scraping expense；Use the operating mechanisms such as intermembranous exchange, self adaptation coordinated transposition and TSP question rewriting to go transmission and the evolution realizing solving object, thus improve computational efficiency and the ability of searching optimum of algorithm.

Description

A kind of improvement heredity film method solving distribution network structure multistage programming

Technical field

The invention belongs to electric information technical field, particularly relate to a kind of improvement solving distribution network structure multistage programming Heredity film method.

Background technology

Distribution network structure planning refers to, under conditions of meeting customer power supply demand and network operation constraint, seek one group Optimum decision variable (such as the erection path of feeder line) so that the various expense total value minimums of programme are such as invested, net Damage and Custom interruption cost expense etc..For medium-term and long-term distribution network planning, owing to its cycle is long, when being often divided into several by planning Between section carry out, i.e. multistage programming method.Multistage programming makes distribution network structure make dynamic adjustment with the change of load, its A kind of dynamic equipment investment or construction scheme can be realized, determine relevant device within planning horizon when most preferably putting into Between, it is ensured that program results is optimum within whole planning year.

The multi-stage optimization planning of distribution network structure, is discrete, a non-linear and multistage combinatorial optimization problem, Tradition optimized algorithm is easily absorbed in " dimension calamity ", has the most all done certain simplification, as the dynamic multistage is converted into static single phase, By linear for non-linear transfer and weaken reliability factor etc., although but can make solution procedure become simple finally obtain only It is the locally optimal solution on definite meaning, answers expenditure to be short of practical situation.In view of the defect of traditional algorithm, modern Heuritic approach (such as genetic algorithm) is increasingly being used for solving distribution network planning problem, and when using genetic algorithm, it is random Containing substantial amounts of infeasible solution in the initial solution produced, cause its tempo of evolution slow, be difficult to search globally optimal solution, so that Obtain whole computational efficiency relatively low.After considering existing power distribution network and comprising distributed power source, its plan model relates to Factor and variable will be more, and the complex nature of the problem strengthens further, solve difficulty and deepen further, thus to optimized algorithm Require also the highest.

Summary of the invention

It is an object of the invention to provide a kind of improvement heredity film method solving distribution network structure multistage programming, it is intended to Containing substantial amounts of infeasible in the initial solution that the multi-stage optimization planning solving distribution network structure uses genetic algorithm to randomly generate Solve, cause its tempo of evolution slow, be difficult to search globally optimal solution so that the problem that whole computational efficiency is relatively low.

The present invention be achieved in that a kind of improvement heredity film algorithm solving distribution network structure multistage programming include with Lower step:

1) mathematical model of distribution network structure multistage programming:

$\begin{array}{c}\min LCC=\underset{n=1}{\overset{N}{\mathrm{\Σ}}}(\underset{i=1}{\overset{n_i}{\mathrm{\Σ}}}{C}_{In,i}{P}_{V1}+(\underset{j=1}{\overset{n_j}{\mathrm{\Σ}}}(C_{On,j}+C_{Mn,j}+C_{Fn,j})+\underset{k=1}{\overset{n_k}{\mathrm{\Σ}}}{C}_{DGn,k}-\\ \underset{l=1}{\overset{n_l}{\mathrm{\Σ}}}{C}_{En,l})P_{Vsum}-\underset{m=1}{\overset{n_m}{\mathrm{\Σ}}}{C}_{Dn,m}{P}_{V2})\end{array}---\left(1\right)$

In formula, LCC is distribution network structure programme life cycle management totle drilling cost present worth；N, n are respectively the stage of planning Number and the n-th planning stage；n_i, i is respectively i-th branch road that the newly-built branch road of n-th order section is total and newly-built；n_j, j is respectively n-th The j-th strip branch road that stage built branch road is total and built；n_k, k is respectively n-th order section operating distributed power source sum and the K distributed power source；n_l, l is respectively pollutant sum and the l kind pollutant that n-th order section coal fired power generation produces；n_m, m is respectively The m article branch road that branch road is total and split is removed for n-th order section；C_In,iCost of investment for n-th order section newly-built branch road i；C_On,jFor The operating cost of n-th order section branch road j；C_Mn,jMaintenance cost for n-th order section branch road j；C_Fn,jPower failure for n-th order section branch road j is damaged Mistake expense；C_DGn,kOperation and maintenance expenses for n-th order section distributed power source k are used；C_En,lThe environmental protection of l kind pollutant is reduced discharging for n-th order section Cost；C_Dn,mThe residual value of branch road m is removed latter stage for n-th order section；P_V1=1/ (1+r)^b(n)For the investment discount factor of n-th order section, b (n) be the time limit before n-th order section and；P_Vsum=(1+r)^g(n)-1/r(1+r)^g(n)+m(l-1)It is the n-th stage apparatus year to run dimension Protect power failure environmental protection present value of cost and, g (n) is the time limit that n-th order section comprises；For n-th order section and former rank thereof Total time limit that section comprises；P_V2=1/ (1+r)^e(n)For n-th order section end remanent value of equipment discount factor, e (n) is for from including n-th order section Time limit sum before and；R is social discount rate.

C_In:C_In,iFor the disposable Meteorological such as the purchase commodity of branch road i, installation fee.

C_On: C_On=C_pu×P_loss×τ_max, C_puFor electricity price；P_lossNetwork active loss for busy hour；τ_maxFor year Peak load loss hourage.

C_Mn: the maintenance expense of power supply unit is generally by a certain ratio that can be taken as initial outlay.

C_Fn:N_lCount for network load；EENS_iFor research period the i load point Expected loss of energy；α_iFor power supply enterprise in the power supply net profit of i-th load point；R_iIt it is the product electricity ratio of the i load point.

C_DGn:C_1,kUnit capacity year O&M cost for kth kind distributed power source；P_DG,kFor The rated capacity of kth kind distributed power source.

C_En:C_2l,C_3lIt is respectively discharge fee and the environment valency of l kind pollutant Value；B_n,lThe discharge intensity of l kind pollutant is produced for coal fired power generation.

C_Dn: do not want the equipment of premature removal to service life, scrap cost should contain its depreciation surplus value (depreciation by straight-line Method), the average residual value of distribution network structure equipment generally can be taken as a certain ratio of its initial outlay.

The constraints of distribution network structure multistage programming includes node voltage constraint, branch current constraint and distribution Net is connective and radioactivity retrains.

2) ultimate principle of heredity film algorithm is improved

Improve heredity film to calculate and construct the 4 tunic systems that degree is 3, the many tuples being expressed as:

Π=(V, T, μ, C, ω₁,ω₂,ω₃,G₁,G₂,G₃,(R₁,ρ₁),(R₂,ρ₂),(R₃,ρ₃)) (2)

In formula, V is object alphabet；T is output alphabet；μ is membrane structure；C is catalyst；ω_i(1≤i≤3) are district The object multiset character string of territory i；G_i(1≤i≤3) are the quantity of object in the i of region；R_i(1≤i≤3) are evolution in the i of region The finite aggregate of rule；ρ_i(1≤i≤3) are R_iIn partial ordering relation, also referred to as dominance relation.

1. coding rule

In underlying membrane calculates, object set generally uses character or string encoding Rule Expression.Herein in conjunction with distribution Net space truss project feature, uses binary coding rule to encode object, and 0 represents the construction of line, and 1 represents circuit not frame If item chromosome is a programme.

2. object set produces

In each film, object set is expressed as follows:

$\left\{\begin{array}{c}{\mathrm{\ω}}_i=radial(p_1...p_{G_i})\\ i=1,2,3\end{array}\right.---\left(3\right)$

The quantity that P system comprises object is:

$G=\underset{i=1}{\overset{3}{\Σ}}{G}_i---\left(4\right)$

3. evolutionary rule

Intermembranous exchange rule: this rule is in each iterative computation, by some object conveying high for this film endoadaptation angle value Go out and accept the respective numbers object that extraneous fitness value is high simultaneously, with the data message in the different film of exchange, improve algorithm complete Office's search capability, is expressed as follows shown:

$\left\{\begin{array}{c}{R}_{iCma}=R_{iCma1}\∪R_{iCma2}\\ R_{iCma1}:{\[p_{m1},p_{m2},...,p_{mn}\]}_i\→{\[\]}_i{p}_{m1},p_{m2},...,p_{mn}\\ R_{iCma2}:{\[\]}_i{p}_{m1}^{\′},p_{m2}^{\′},...,p_{mn}^{\′}\→{\[p_{m1}^{\′},p_{m2}^{\′},...,p_{mn}^{\′}\]}_i\\ i=1,2,3\end{array}\right.---\left(5\right)$

In formula, R_iCmaFor exchange rule；R_iCma1,R_iCma2For solving object in film to being transmitted across in film outer with film film Journey；p_m1,p_m2,…,p_mn；p′_m1,p′_m2,…,p′_mnN the object higher for film i endoadaptation angle value and film i exoadaptation angle value are relatively N high object.

Coordinated transposition rule: this rule in independent inheritance operation, first randomly chooses cross point and self adaptation is raw in film Become crossover probability, then the relevant position meeting two objects of crossover probability is exchanged with each other and obtains new object, be expressed as follows Shown in:

$\left\{\begin{array}{c}{R}_{i,Ctro}:{{\&lsqb;}_i(p_1,p_2)\&RightArrow;(p_3,p_4)\&rsqb;}_i\\ p_1=(x_1,x_2,...,x_{s1},...,x_{s2},...,x_l)\\ p_2=(y_1,y_2,...,y_{s1},...,y_{s2},...,y_l)\\ p_3=(x_1,x_2,...,y_{s1},...,y_{s2},...,x_l)\\ p_4=(y_1,y_2,...,x_{s1},...,x_{s2},...,y_l)\\ p_c=\left\{\begin{array}{cc}\frac{k_1(f_{\max }-f)}{f_{\max }-f_{avg}}& f\&GreaterEqual;f_{ave}\\ k_2& f<f_{ave}\end{array}\right.\end{array}\right.---\left(6\right)$

In formula, R_CrtoFor coordinated transposition rule；p₁,p₂For carrying out two objects of coordinated transposition in film i；p₃,p₄For in film i The new object produced after coordinated transposition；x_i(1≤i≤l),y_i(1≤i≤l) is object p_iI-th bit in (1≤i≤m)；p_c For coordinated transposition probability；f_maxMaximum adaptation degree for colony；f_avgAverage fitness for colony；F is coordinated transposition to be participated in Two objects in bigger fitness value；k₁,k₂For parameter.

Variation rewriting rule: for increase solve object multiformity, in solution procedure run variation rewriting rule, first with Machine selects change point adaptive generation mutation probability, then the relevant position gene meeting mutation probability object is changed to its etc. Position gene, to obtain new object, is expressed as follows shown:

$\left\{\begin{array}{c}{R}_{iMurw}:{{\&lsqb;}_i{p}_5\&RightArrow;p_6\&rsqb;}_i\\ p_5=(x_1^{\&prime;},x_2^{\&prime;},...,x_s^{\&prime;},...,x_l^{\&prime;})\\ p_6=(x_1^{\&prime;},x_2^{\&prime;},...,y_s^{\&prime;},...,x_l^{\&prime;})\\ p_m=\left\{\begin{array}{cc}\frac{k_3(f_{\max }-f^{\&prime;})}{f_{\max }-f_{avg}}& f^{\&prime;}\&GreaterEqual;f_{ave}\\ k_4& f^{\&prime;}<f_{ave}\end{array}\right.\end{array}\right.---\left(7\right)$

In formula, R_iMurwFor variation rewriting rule；p₅For carrying out the object that variation is rewritten in film i；p₆For after variation is rewritten The new object produced；p_mProbability is rewritten for variation；F' is the fitness value participating in the object that variation is rewritten；k₃,k₄For parameter.

The whole computation model improving heredity film algorithm is:

$\left\{\begin{array}{c}\mathrm{\&Pi;}=(V,T,\mathrm{\&mu;},C,{\mathrm{\&omega;}}_1,{\mathrm{\&omega;}}_2,{\mathrm{\&omega;}}_3,G_1,G_2,G_3,(R_1,{\mathrm{\&rho;}}_1),\\ (R_2,{\mathrm{\&rho;}}_2),(R_3,{\mathrm{\&rho;}}_3))\\ V=\left\{R\right\},T=\left\{R\right\}\\ \mathrm{\&mu;}={{\&lsqb;}_1{{\&lsqb;}_2{{\&lsqb;}_3\mathrm{\&rsqb;}}_3\mathrm{\&rsqb;}}_2{{\&lsqb;}_4\&rsqb;}_4\&rsqb;}_1\\ {\mathrm{\&omega;}}_i=p_1,...,p_{Gi}\\ R_i=\{R_{iCma},R_{i,Crto},R_{iMurw}\}\\ R_{iCma}=R_{iCma1}\&cup;R_{iCma2}\\ R_{iCma1}:{\&lsqb;p_{m1},p_{m2},...,p_{mn}\&rsqb;}_i\&RightArrow;{\&lsqb;\&rsqb;}_i{p}_{m1},p_{m2},...,p_{mn}\\ R_{iCma2}:{\&lsqb;\&rsqb;}_i{p}_{m1}^{\&prime;},p_{m2}^{\&prime;},...,p_{mn}^{\&prime;}\&RightArrow;{\&lsqb;p_{m1}^{\&prime;},p_{m2}^{\&prime;},...,p_{mn}^{\&prime;}\&rsqb;}_i\\ R_{i,Ctro}:{{\&lsqb;}_i(p_1,p_2)\&RightArrow;(p_3,p_4)\&rsqb;}_i\\ p_1=(x_1,x_2,...,x_{s1},...,x_{s2},...,x_l)\\ p_2=(y_1,y_2,...,y_{s1},...,y_{s2},...,y_l)\\ p_3=(x_1,x_2,...,y_{s1},...,y_{s2},...,x_l)\\ p_4=(y_1,y_2,...,x_{s1},...,x_{s2},...,y_l)\\ p_c=\left\{\begin{array}{cc}\frac{k_1(f_{\max }-f)}{f_{\max }-f_{avg}}& f\&GreaterEqual;f_{ave}\\ k_2& f<f_{ave}\end{array}\right.\\ R_{iMurw}:{{\&lsqb;}_i{p}_5\&RightArrow;p_6\&rsqb;}_i\\ p_5=(x_1^{\&prime;},x_2^{\&prime;},...,x_s^{\&prime;},...,x_l^{\&prime;})\\ p_6=(x_1^{\&prime;},x_2^{\&prime;},...,y_s^{\&prime;},...,x_l^{\&prime;})\\ p_m=\left\{\begin{array}{cc}\frac{k_3(f_{\max }-f^{\&prime;})}{f_{\max }-f_{avg}}& f^{\&prime;}\&GreaterEqual;f_{ave}\\ k_4& f^{\&prime;}<f_{ave}\end{array}\right.\\ {\mathrm{\&rho;}}_i=\{R_{iCma}>R_{i,Crto}>R_{iMurw}\}\\ i=1,2,3\end{array}\right.---\left(8\right)$

3) Adjacent Matrix Method judges the feasibility solved

Randomly generate initial solution and can there is great infeasibility, use Adjacent Matrix Method to filter meter for improving optimization efficiency To a feasibility solving object in heredity film, infeasible solution in calculation, judges that step is as follows:

A) obtain and solve object map (m is the node of network to the square formation that adjacency matrix A, A are m × m of real network Number), the value of its element is that " 0 " or " 1 " (without direct correlation relation between the former node, the latter represents straight between node Connect association)；

B) each row and each row to matrix A are sued for peace respectively, if having certain a line or certain string sum is 0, indicate node Directly not being connected with other any nodes in figure, i.e. there is acnode, this solution is infeasible；

C) matrix A is carried out short-circuit operation.First finding out element value in the first row of A is the node serial number of 1, by the first row and In A the row of reference numeral mutually or, simultaneously by first row and A reference numeral row phase or, then with phase or after result go more New matrix A, then delete the first row of A and its diagonal entry is set to 0 by first row simultaneously, then judges to be now 1 in matrix A Element number, if less than 2, illustrate node not with other node UNICOMs, this solution is the most infeasible, if more than 2, repeats this and walks Suddenly, until the dimension of adjacency matrix is 2 after updating, terminate to judge.

The improvement heredity film method solving distribution network structure multistage programming that the present invention provides, draws improved adaptive GA-IAGA Enter in the computation model of membranous system, form improvement heredity film algorithm and be used for optimizing calculating with containing investment, O&M, power failure, environmental protection And the distribution network structure plan model containing distributed power source of the minimum target of overall life cycle cost value of scraping expense；Use The operating mechanisms such as intermembranous exchange, self adaptation coordinated transposition and TSP question rewriting go transmission and the evolution realizing solving object, Thus improve computational efficiency and the ability of searching optimum of algorithm.And use Adjacent Matrix Method to produce the most feasible initial solution With the infeasible solution filtered in calculating, provide for discrete, non-linear and multistage solving of distribution network structure planning problem A kind of efficient method.It is multistage for solving the distribution network structure containing distributed power source that the present invention proposes improvement heredity film algorithm Section planning problem, is different from genetic algorithm in solution procedure and directly randomly generates initial population, and have employed Adjacent Matrix Method Generate initial solution, it is ensured that all initial solutions are the most feasible thus reduce algorithm search scope, carried out elite reservation simultaneously The genetic manipulations such as strategy, self adaptation coordinated transposition and TSP question rewriting, it is possible to guide the effective evolution solving object in sub-film, Increasing the multiformity solved, this external each son is intermembranous has used ac operation mechanism, it is achieved that the information solving object in different films is handed over Change, improve the ability of searching optimum of algorithm to a greater degree.Improvement heredity film algorithm and genetic algorithm that the present invention proposes are divided Do not solve planning problem, can be seen that the optimal solution that improvement heredity film algorithm is sought compares genetic algorithm from two kinds of convergence of algorithm figures The optimal solution sought is more preferable, i.e. improves heredity film algorithm and has higher ability of searching optimum, and genetic algorithm changes at nearly 7th time For time be the most easily absorbed in locally optimal solution, and be difficult to jump out locally optimal solution, and in convergence rate, improve heredity film algorithm Also faster than genetic algorithm.

Accompanying drawing explanation

Fig. 1 is the membrane structure schematic diagram improving heredity film algorithm that the embodiment of the present invention provides.

Fig. 2 is the improvement heredity film Algorithm for Solving distribution network structure multistage programming flow chart that the embodiment of the present invention provides.

Fig. 3 is the 10 node initial network structural representations that the embodiment of the present invention provides.

Fig. 4 is the single phase planning network structural representation that the embodiment of the present invention provides.

Fig. 5 is the multistage programming schematic network structure that the embodiment of the present invention provides.

Fig. 6 is the genetic algorithm film algorithmic statement curve comparison diagram hereditary with improvement that the embodiment of the present invention provides.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with embodiment, to the present invention It is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not used to Limit the present invention.

Below in conjunction with the accompanying drawings the application principle of the present invention is further described.

As it is shown in figure 1, after analyzing the feature of planning problem, determine that the membrane structure improving heredity film algorithm includes The film of 4 different levels, wherein No. 2, No. 3 and No. 4 films are contained within object multiset and evolutionary rule, and this example is in these 3 films The genetic manipulation improved, simultaneously in these 3 intermembranous exchange mechanisms of intermembranous operation, finally exports in the region of No. 1 film Excellent solution object.Containing 4 tunics of different levels in membrane structure, wherein, it is any that the inside of No. 3 and No. 4 films no longer comprises other Film, they are underlying membrane；No. 1 film is positioned at the outermost layer of system, and it is top layer film, and it can keep apart membranous system and external environment condition； Each tunic has its specific region, and for underlying membrane, region is the space that its own comprises, for non-underlying membrane Speech, region is the space between the film that this film directly comprises self；Because the relation between film and its region is one a pair Answer, so the labelling of film also illustrates that the labelling of its corresponding region；Object multiset and evolutionary rule in membranous system are deposited in In the region of film, it is possible to instruct the properly functioning of this film.

As in figure 2 it is shown, 1, start to calculate, the initial data of input planning, 2, coding generate each film initial solution object, 3, front Push back the trend calculating power distribution network network for method, 4, calculate the fitness value of all solution objects in each film, 5, carry out the film of sub-film Interior genetic manipulation (include elite retain selection, self adaptation coordinated transposition and TSP question rewriting operation), 6, judge be The no constraints meeting the network operation, 7, penalty term is added fitness function, 8, calculate a new generation and solve the fitness of object Value, 9, carry out the intermembranous alternating current operation that each son is intermembranous, 10, judge whether to meet stopping criterion for iteration, 11, output optimum programming side Case, terminates to calculate.First input the initial data of planning, include the node of distribution network and branch parameters, year planning horizon number, The data such as initial solution object number and iterations in discount rate, sub-film；Use binary coding strategy to solving object coding And use Adjacent Matrix Method to generate initial solution the most feasible in each film；Forward-backward sweep method is used to carry out the Load flow calculation of network, Obtain voltage and the network loss value of network of egress；The fitness value of object is solved in calculating each film according to formula (1)；According to fitness value Height carries out selecting operation to solving object, uses elite retention strategy directly to preserve the object that this generation is optimum to lower generation simultaneously； According to formula (6), the object selected is carried out self adaptation coordinated transposition, and wherein coordinated transposition probability is the fitness according to object Value determines, high its coordinated transposition probability of fitness value is big, otherwise low its probability of fitness value is less, thus realizes excellent Retaining and the evolution of object inferior of elegant object；Object after coordinated transposition is carried out TSP question rewriting according to formula (7)；Sentence Breaking and whether meet constraints, if being unsatisfactory for, penalty term being added fitness function；Calculate the adaptation solving object of a new generation Several objects preferable in this film are given adjacent membranes and accept the outstanding object that adjacent membranes is sent by angle value；Judge whether to meet Stopping criterion for iteration；Optimal object in output membranous system, is the programme of optimum.

Simulation calculation is carried out for 10 node power distribution networks shown in Fig. 3.

This distribution network has 1 power supply point, 2 existing branch roads (in Fig. 3, heavy line represents), treats that frame branch road is (in Fig. 3 for 14 Dotted line represents), a photovoltaic plant (in Fig. 3, Red Star type represents), electric pressure is 10kV, its branch road and node data such as table 1, Shown in 2.

Table 1

Table 2

Planning initial data:

Electricity price is 0.4 yuan/kWh；The peak load loss time of each load point is 3000h；Planning horizon is 10 years；Discount Rate is 0.04；Line upkeep cost is taken as the 3% of initial outlay；The average residual value of circuit is taken as the 2% of its initial outlay；Assume The existing photovoltaic plant (rated power 200kW, power factor 0.9) of load point 2, operation and maintenance cost takes 0.02 ten thousand yuan/kW；Combustion The NO that coal power generation produces_x、SO₂And CO₂Discharge intensity be respectively 1.634,4.445 and 1008.788kg/MWh；NO_x、SO₂With CO₂Discharge fee be respectively 2,1.26 and 0.765 yuan/kg, NO_x、SO₂And CO₂Environmental value be respectively 8,6 and 0.023 Unit/kg.

Single phase planning application:

According to above-mentioned parameter, distribution network structure is carried out the result of single phase planning as shown in Figure 4, wherein thin in Fig. 4 Solid line represents newly-built circuit.

Multistage programming is analyzed:

According to above-mentioned parameter, distribution network structure is carried out multistage programming, it is assumed that load point 9,10 is newly-increased load, planning Being divided into two stages, every stages period is 5 years.The result of multistage programming as it is shown in figure 5, wherein red fine line represent second Stage newly-built circuit.

Compare as shown in table 3 by single phase, containing the DG multistage with without DG multistage programming result.

Table 3

Multistage programming scheme totle drilling cost value on the whole cycle of planned project as can be seen from Table 3: 1) (217.591 ten thousand yuan), less than the totle drilling cost value (257.938 ten thousand yuan) of planning of direct single phase, illustrate to use multistage programming method The planning carrying out rack not only can meet the dynamic change of load, and can arrange the construction of equipment on the suitable time, closes The investment of reason distributing equipment, has more preferable motility and economy.2) on interruption cost (CF), single phase planning side Case is 74.025 ten thousand yuan, and multistage programming scheme is 66.439 ten thousand yuan, and the loss of outage of the latter is the least i.e. illustrates multistage programming The distribution network shelf structure obtained has higher reliability.3) although containing in the scheme of DG and adding its operation and maintenance expenses use, but draw After entering DG, network operation cost decreases, and environmental benefit adds, and its overall life cycle cost total value (217.591 ten thousand yuan) is less than Without the scheme (269.768 ten thousand yuan) of DG, embody the power distribution network containing DG and there is less network loss, higher environmental benefit and more Excellent economy.

Using genetic algorithm and improvement heredity film algorithm to solve example respectively, acquired results is to such as shown in table 4 and Fig. 6.

Table 4

By table 4 and Fig. 6 it is found that improve optimal solution that the heredity optimal solution sought of film algorithm seeks than genetic algorithm more Good, genetic algorithm is the most easily absorbed in locally optimal solution when nearly 7th iteration, and is difficult to jump out locally optimal solution, and improves something lost Pass film algorithm and substantially there is higher ability of searching optimum, and convergence rate is also fast than genetic algorithm.

The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all essences in the present invention Any amendment, equivalent and the improvement etc. made within god and principle, should be included within the scope of the present invention.

Claims (4)

1. the improvement heredity film method solving distribution network structure multistage programming, it is characterised in that described in solve power distribution network The heredity film method that improves of rack multistage programming is in the computation model that Revised genetic algorithum introduces membranous system, is formed new The heredity film algorithm that improves of type is used for calculating the power distribution network containing distributed power source with the minimum target of overall life cycle cost value Space truss project model, have employed in solving the transmission of object and evolving intermembranous exchange, self adaptation coordinated transposition and adaptive strain The operating mechanisms such as different rewriting, and use Adjacent Matrix Method produce the most feasible initial solution and filter in calculating infeasible Solve；

Revised genetic algorithum one degree of structure is the 4 tunic systems of 3, the many tuples being expressed as:

Π=(V, T, μ, C, ω₁,ω₂,ω₃,G₁,G₂,G₃,(R₁,ρ₁),(R₂,ρ₂),(R₃,ρ₃))；In formula, V is object letter Table；T is output alphabet；μ is membrane structure；C is catalyst；ω_i(1≤i≤3) are the object multiset character string of region i；G_i (1≤i≤3) are the quantity of object in the i of region；R_i(1≤i≤3) are the finite aggregate of evolutionary rule in the i of region；ρ_i(1≤i≤3) For R_iIn partial ordering relation.

Solve the improvement heredity film method of distribution network structure multistage programming the most as claimed in claim 1, it is characterised in that institute State and solve the improvement heredity film method of distribution network structure multistage programming and include:

After initially setting up the mathematical model of distribution network structure multistage programming containing distributed power source, according to the spy of variable to be optimized Point is determined the membrane structure of heredity film and solves the coding strategy of object；

Then each straton film in membrane structure carries out independent improvement genetic manipulation and include that elite retention strategy, self adaptation are handed over Fork transposition and TSP question are rewritten；The a new generation producing each sub-film solves object and carries out trap queuing, by preferable for each sub-film Solution object carry out exchange transmission；

The last the highest solution object of fitness value that exports from whole membranous system after meeting the end condition of calculating is as asking The optimum programming scheme gone out.

Solve the improvement heredity film method of distribution network structure multistage programming the most as claimed in claim 1, it is characterised in that institute State the mathematical model of distribution network structure multistage programming:

$\begin{array}{c}\min LCC=\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}(\underset{i=1}{\overset{n_i}{\mathrm{\&Sigma;}}}{C}_{In,i}{P}_{V1}+(\underset{j=1}{\overset{n_j}{\mathrm{\&Sigma;}}}\left(C_{On,j}+C_{Mn,j}+C_{Fn,j}\right)+\underset{k=1}{\overset{n_k}{\mathrm{\&Sigma;}}}{C}_{DGn,k}-\\ \underset{l=1}{\overset{n_l}{\mathrm{\&Sigma;}}}{C}_{En,l})P_{Vsum}-\underset{m=1}{\overset{n_m}{\mathrm{\&Sigma;}}}{C}_{Dn,m}{P}_{V2})\end{array};$

In formula, LCC is distribution network structure programme life cycle management totle drilling cost present worth；N, n be respectively planning number of stages and N-th planning stage；n_i, i is respectively i-th branch road that the newly-built branch road of n-th order section is total and newly-built；n_j, j is respectively n-th order section The j-th strip branch road that built branch road is total and built；n_k, k is respectively n-th order section operating distributed power source sum and kth Distributed power source；n_l, l is respectively pollutant sum and the l kind pollutant that n-th order section coal fired power generation produces；n_m, m is respectively The n stage removes the m article branch road that branch road is total and split；C_In,iCost of investment for n-th order section newly-built branch road i；C_On,jIt is n-th The operating cost of stage branch road j；C_Mn,jMaintenance cost for n-th order section branch road j；C_Fn,jLoss of outage for n-th order section branch road j Expense；C_{DGn, k}Operation and maintenance expenses for n-th order section distributed power source k are used；C_En,lThe environmental protection reducing discharging l kind pollutant for n-th order section becomes This；C_Dn,mThe residual value of branch road m is removed latter stage for n-th order section；P_V1=1/ (1+r)^b(n)For the investment discount factor of n-th order section, b (n) be the time limit before n-th order section and；P_Vsum=(1+r)^g(n)-1/r(1+r)^g(n)+m(l-1)It is the n-th stage apparatus year to run dimension Protect power failure environmental protection present value of cost and, g (n) is the time limit that n-th order section comprises；For n-th order section and former rank thereof Total time limit that section comprises；P_V2=1/ (1+r)^e(n)For n-th order section end remanent value of equipment discount factor, e (n) is for from including n-th order section Time limit sum before and；R is social discount rate；

C_In:C_In,iDisposable Meteorological for branch road i；

C_On: C_On=C_pu×P_loss×τ_max, C_puFor electricity price；P_lossNetwork active loss for busy hour；τ_maxMaximum for year Load loss hourage；

C_Mn: the maintenance expense of power supply unit is taken as a certain ratio of initial outlay；

C_Fn:N_lCount for network load；EENS_iIt it is the i load point expected energy not supplied Value；α_iFor power supply enterprise in the power supply net profit of i-th load point；R_iIt it is the product electricity ratio of the i load point；

C_DGn:C_1,kUnit capacity year O&M cost for kth kind distributed power source；P_DG,kFor kth kind The rated capacity of distributed power source；

C_En:C_2,l,C_3,lIt is respectively discharge fee and the environmental value of l kind pollutant； B_n,lThe discharge intensity of l kind pollutant is produced for coal fired power generation.

C_Dn: do not want the equipment of premature removal to service life, scrap cost should contain its depreciation surplus value.

Solve the improvement heredity film method of distribution network structure multistage programming the most as claimed in claim 1, it is characterised in that institute The computation model stating improvement heredity film algorithm is:

$\left\{\begin{array}{c}\mathrm{\&Pi;}=(V,T,\mathrm{\&mu;},C,{\mathrm{\&omega;}}_1,{\mathrm{\&omega;}}_2,{\mathrm{\&omega;}}_3,G_1,G_2,G_3,\left(R_1,{\mathrm{\&rho;}}_1\right),\\ \left(R_2,{\mathrm{\&rho;}}_2\right),\left(R_3,{\mathrm{\&rho;}}_3\right))\\ V=\left\{R\right\},T=\left\{R\right\}\\ \mathrm{\&mu;}={{\&lsqb;}_1{{\&lsqb;}_2{{\&lsqb;}_3\&rsqb;}_3\&rsqb;}_2{{\&lsqb;}_4\&rsqb;}_4\&rsqb;}_1\\ {\mathrm{\&omega;}}_i=p_1,...,p_{Gi}\\ R_i=\left\{R_{iCma},R_{i,Crto},R_{iMurw}\right\}\\ R_{iCma}=R_{iCma1}\&cup;R_{iCma2}\\ R_{iCma1}:{\&lsqb;p_{m1},p_{m2},...,p_{mm}\&rsqb;}_i\&RightArrow;{\&lsqb;\&rsqb;}_i{p}_{m1},p_{m2},...,p_{mm}\\ R_{iCma2}:{\&lsqb;\&rsqb;}_i{p}_{m1}^{\&prime;},p_{m2}^{\&prime;},...,p_{mm}^{\&prime;}\&RightArrow;{\&lsqb;p_{m1}^{\&prime;},p_{m2}^{\&prime;},...,p_{mm}^{\&prime;}\&rsqb;}_i\\ R_{i,Crto}:{{\&lsqb;}_i\left(p_1,p_2\right)\&RightArrow;\left(p_3,p_4\right)\&rsqb;}_i\\ p_1=\left(x_1,x_2,...,x_{s1},...,x_{s2},...,x_l\right)\\ p_2=\left(y_1,y_2,...,y_{s1},...,y_{s2},...,y_l\right)\\ p_3=\left(x_1,x_2,...,y_{s1},...,y_{s2},...,x_l\right)\\ p_4=\left(y_1,y_2,...,x_{s1},...,x_{s2},...,y_l\right)\\ p_c=\left\{\begin{array}{cc}\frac{k_1\left(f_{\max }-f\right)}{f_{\max }-f_{avg}}& f\&GreaterEqual;f_{avg}\\ k_2& f<f_{avg}\end{array}\right.\\ R_{iMurw}:{{\&lsqb;}_i{p}_5\&RightArrow;p_6\&rsqb;}_i\\ p_5=\left(x_1^{\&prime;},x_2^{\&prime;},...,x_s^{\&prime;},...,x_l^{\&prime;}\right)\\ p_6=\left(y_1^{\&prime;},y_2^{\&prime;},...,y_s^{\&prime;},...,y_l^{\&prime;}\right)\\ p_m=\left\{\begin{array}{cc}\frac{k_3\left(f_{\max }-f^{\&prime;}\right)}{f_{\max }-f_{avg}}& f^{\&prime;}\&GreaterEqual;f_{avg}\\ k_4& f^{\&prime;}<f_{avg}\end{array}\right.\\ {\mathrm{\&rho;}}_i=\left\{R_{iCma}>R_{i,Crto}>R_{iMurw}\right\}\\ i=1,2,3\end{array}\right..$