CN106228243A - Metro depot Hui Ku station track arrangement method - Google Patents

Abstract

The invention discloses a kind of metro depot Hui Ku station track arrangement method.The method comprises the steps that storehouse the to be returned information of vehicles needed for first collection algorithm and station track information, and sets up the parameter set of practical problem；Secondly select suitable coded system that chromosome constructs and formulates initialization of population rule, generate Z chromosome and constitute initial population P；Then optimization object function and constraint are determined according to practical problem；Finally determine individuality selects rule and hybridization, the concrete operations mode of variation, calculates the adaptive value of each individuality according to fitness function and preserves the individuality that adaptive value is maximum, and repetitive operation is until maximum evolutionary generation.Train mission can rationally be accepted or rejected and find optimal station track to arrange scheme under resource constraint by the present invention, provides decision support for follow-up daily test, shunting service, is greatly improved rolling stock section's working performance.

Description

Metro depot Hui Ku station track arrangement method

Technical field

The invention belongs to technical field of rail traffic, particularly a kind of metro depot Hui Ku station track arrangement method.

Background technology

Metro depot as the main station section of subway production and transport, main is responsible for the parking of vehicle, is checked, reorganizes and outfit, transports With with the business such as repairing.Need to stop to rolling stock section to go out with Awaiting Overhaul operation or tomorrow after railcar terminates operation plan on the same day Car, it is significant to improving rolling stock section's production efficiency that scientific arrangement goes back to storing cycle station track, storehouse.Conventional truck Duan Huiku station track Arrange to be adjusted to adjust with inspection by row before each car Hui Ku, confer decision according to job requirements temporarily, pre-owing to not having overall arrangement to lack Opinion property, may result in the later period returns the job requirements of storehouse vehicle and cannot meet and cause delaying or unnecessary tune of operation Car and time-consuming, therefore consider Hui Ku station track is arranged to be optimized with at utmost support vehicles section operation the most efficiently.

Not yet there is the algorithm research arranged for rolling stock section Hui Ku station track at present, but reorganize and outfit district and passenger traffic about locomotive depot Studying of station track utilization of standing arranges algorithm design to have certain reference for rolling stock section Hui Ku station track.Ji Lixia is for machine The station track Arrangement Problem in district is reorganized and outfit in business section field personnel, it is proposed that self-adapted genetic algorithm based on N × M two-dimensional chromosome encoding, easily Operate and have good convergence precision；Zhang Yinggui uses the condition that should observe according to railway passenger station station track, utilizes modern times sequence Theory, builds station track application plan and automatically works out order models, be optimized the application plan of station track, passenger station.JW Chung Propose a kind of genetic algorithm based on mixed-integer programming model, solve under rolling stock section's capacity retrains with service ability Vehicle dispatching problem.Rolling stock section Hui Ku station track is arranged and inapplicable by algorithm above, reality should be arranged to carry out it in conjunction with station track Adjust and be applied.

Summary of the invention

It is an object of the invention to provide a kind of metro depot Hui Ku station track simple, efficient arrangement method.

The technical solution realizing the object of the invention is: a kind of metro depot Hui Ku station track arrangement method, including with Lower step:

Step 1, collects storehouse to be returned information of vehicles and station track information, and sets up the parameter set of practical problem；

Step 2, chromosome constructs: on the basis of analyzing Hui Ku station track Arrangement Problem coding requirement, selects coded system Chromosome is constructed；

Step 3, initializes population P: combine vehicle and return the actual formulation in storehouse initialization of population rule, and generate according to this rule Z chromosome constitutes initial population P；

Step 4, calculates fitness and also preserves optimum individual: determine optimization object function and constraint that Hui Ku station track arranges, Calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved；

Step 5, selects, hybridizes, makes a variation: for coded system, the initialization criterion of the arrangement of Hui Ku station track, to individual choosing The concrete operations mode selecting rule and hybridization, variation is determined；

Step 6, produces population P of future generation: according to the selection determined in step 5, hybridize, make a variation rule, choosing from population Go out Z individuality carry out hybridizing, mutation operation preserving to new population；

Step 7, end condition: repeat step 4 to step 6 until reaching maximum evolutionary generation, optimum now preserved Body is the optimal solution of the solved problem of algorithm, carries out picking up according to this optimal solution and can realize metro depot Hui Ku station track Optimum arrangements.

Further, storehouse the to be returned information of vehicles described in step 1 includes license number, whether Hui Ku, vehicle location, institute's sport car Secondary, return the storehouse time, return storehouse order, whether tomorrow runs morning peak, whether carwash, whether daily test, whether park station track type, tomorrow Reach the standard grade；Station track information includes station track title, station track type, takies situation and station track coding；The parameter set of practical problem includes: plant Group scale Z, probability of crossover p_c, mutation probability p_m, maximum evolutionary generation.

Further, the chromosome structure described in step 2, particularly as follows: using a gene as an available station track, will Train returns storehouse order and inserts to represent track occupied；Available Necessary Number of Tracks when every day, vehicle went back to storehouse is indefinite, so according to reality Chromosome is dynamically constructed by border situation, constructs chromosome with the coded system of one-dimensional matrix, each of chromosome all with A certain available station track numbering is corresponding.

Further, the initialization population P described in step 3, particularly as follows:

(3.1) travel through storehouse to be returned vehicle, obtain back storehouse vehicle number and station track requires and returns the priority of storehouse time according to expectation Vehicle is numbered；

(3.2) determining available station track state: when same station track A-share, B stock all can use, 1. A-share state is, 2. B stock is； When A-share take, B stock can use time, 3. B strand state is；When B stock take, A-share can use time, 4. A-share state is；

(3.3) determine that vehicle returns storehouse numbering and inserts the rule of chromosome:

For there being the vehicle of daily test operation night, 1., 2., 4. or other have the stock of trench can park the state of station track is Road；

For there being the vehicle of carwash operation night, station track state can be parked for 1., 2., 3.；

For morning peak need to be arranged tomorrow or have the vehicle of appointment train number, the state of station track can be parked for 1., 4.；

For there being tomorrow system to repair the vehicle of operation, can park station track is the station track having protection network and trench；

Other are had and specifies the vehicle parking station track type, arrange the station track of corresponding types to park；

For having two and the vehicle of above job requirements, station track can be parked and to ask for what all individual event operation station tracks required Occur simultaneously, occur simultaneously for, time empty, upkeep operation requirement being ranked up by importance when station track can be parked, cancel part significance level low Job requirements or job content, job requirements importance is ranked up: daily test > system is repaiied and other station tracks types are specified > Carwash operation > morning peak or specify train number；

(3.4) select insert station track and available station track state is updated according to the rule in (3.3), be repeated up to institute Needing back storehouse vehicle all inserts chromosome；

(3.5) (3.4) are repeated until generating Z chromosome to constitute initial population P.

Further, calculate fitness described in step 4 and preserve optimum individual: determining optimization object function and constraint, Calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved, specific as follows:

(4.1) determining constraints, wherein constraints includes time interval constraint and metro depot service ability Constraint；

(4.2) determining object function: optimization aim is that task completeness is the highest, the process that picks up always is shunt least number of times, car Go out to put convenient degree in storage the highest；

(4.3) fitness function is determined；

(4.4) calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved.

Further, select described in step 5, hybridize, make a variation: for coded system, the initialization standard of the arrangement of Hui Ku station track Then, individuality select the concrete operations mode of rule and hybridization, variation be determined, specific as follows:

(5.1) determining the selection rule of individuality, employing roulette carrys out the individuality in selected population for crossover operation, if planting The adaptive value that in Qun, kth is individual is F_k, then the individual selected probability P of kth_kFor:

$P_k=\frac{F_k}{\underset{i=1}{\overset{Z}{\Σ}}{F}_i}$

(5.2) determining the crossover operation mode of individuality, crossover process comprises the following steps:

The first step: generation each to individuality to Z/2 one (0,1] random number, if random number is less than probability of crossover p_cThen enter Row crossover operation, on the contrary do not hybridize, directly as in new individual preservation to new population；

Second step: for needing the individuality of hybridization, randomly generate two [0, N_track) integer as hybridization location, to often Individual parent selectes hybridization portion；

3rd step: the hybridization portion of first parent and the hybridization portion of second parent are interchangeable；

4th step: remove all trains repeated from the generation original gene of offspring and return storehouse sequence number；

5th step: the gene producing offspring is carried out completion according to initialization of population principle, and will new individual preserve to newly In population；

6th step: repeat the first step～the 5th step until new population scale reaches Z；

(5.3) determining the mutation operation mode of individuality, mutation process comprises the following steps:

The first step: in new population every individual each produce one (0,1] random number, if random number is less than mutation probability p_m Then carry out mutation operation, otherwise the most do not carry out mutation operation；

Second step: for needing the individuality of variation, randomly generate two [0, N_track) integer as catastrophe point, and will be prominent Genic value in height swaps；

3rd step: perform second step～the 3rd step in all individualities of population.

Further, step (4.1) is described determines constraints, and wherein constraints includes time interval constraint and subway The constraint of rolling stock section's service ability, wherein:

Time interval is retrained, has two kinds of situations:

Situation a, first train elder generation Hui Ku is parked in A-share and carries out daily test operation, and operation is adjusted to B stock after completing, A-share empties Second train of follow-up continued access, first train is whole pick up during, the time-consuming 40min of daily test, shunt time-consuming 30min, therefore First is more than or equal to 70min with the storehouse time interval of returning of second train；When returning storehouse time interval less than 70min, B stock car Daily test task cannot complete, directly stop to B stock；

Situation b, first train first passes through washing track and docks to A-share from dextrad and carry out daily test operation, and operation is adjusted after completing To B stock, A-share empties second train of follow-up continued access, first train is whole pick up during, the time-consuming 15min of carwash, daily test Time-consuming 40min, shunt time-consuming 30min, therefore first is more than or equal to 85min with the storehouse time interval of returning of second train；When One with second train return storehouse time interval less than 85min and more than 70min time, abandon carwash task and only carry out daily test； When time storehouse time interval of first with second train is less than 70min, abandons daily test task and only carry out carwash；

For the constraint of metro depot service ability, number of tasks of shunting of same time is not more than 1, and every other day time same Inspection vehicle number is not more than 3, it may be assumed that

$\left\{\begin{array}{c}{N}_{di}\≤3\\ N_{dc}\≤1\end{array}\right.$

Wherein, N_diVehicle number for same time daily test；N_dcNumber of tasks of shunting for the same time.

Further, step (4.2) is described determines object function: optimization aim is that task completeness is the highest, picks up process Always shunting least number of times, it is the highest that vehicle goes out to put in storage convenient degree, and wherein task completeness is the highest has the first importance, shunts time Number minimum has the second importance, goes out to put in storage that convenient degree is the highest has the 3rd importance；For target z_j(j=1,2,3) reach To priority and each target need the sub-goal that reaches, the expression formula of object function is as follows:

$lex\min \{z_1=\underset{i=1}{\overset{m_0}{\Σ}}(w_{1i}^{+}{d}_i^{+}+w_{1i}^{-}{d}_i^{-}),z_2=\underset{i=1}{\overset{m_0}{\Σ}}(w_{2i}^{+}{d}_i^{+}+w_{2i}^{-}{d}_i^{-}),z_3=\underset{i=1}{\overset{m_0}{\Σ}}(w_{3i}^{+}{d}_i^{+}+w_{3i}^{-}{d}_i^{-})\}$

Wherein, lexmin represents and minimizes target according to lexicographic order；m₀Represent sub-goal quantity；Represent i-th mesh Mark exceedes intended overgauge variable；Represent i-th target and exceed intended minus deviation variable；Representative is distributed to's Positive weights；Representative is distributed toNegative weight；

It is the highest for task completeness, if object function is z₁, z₁There are 4 sub-goals: daily test completion rate of the plan is the highest, car Park that station track type and its operation station track require that matching degree is the highest, carwash completion rate of the plan is the highest, morning peak station track requires Degree of joining is the highest, the object function z that task completeness is the highest₁For:

$z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X$

Wherein, T_iuIt is not fully complete number of tasks for i-th sub-goal；X is penalty coefficient；For distributing to each sub-goal Negative weight；

For least number of times of shunting, if object function is z₂, A, B two strands of and if only if same station track is both needed to park and treats Return storehouse vehicle and B stock vehicle need and can daily test time, need to carry out shunting service, the object function z of least number of times of shunting₂ For:

z₂=T_dc

Wherein, T_dcFor actual number of times of shunting；

It is the highest for going out to put in storage convenient degree, if object function is z₃, z₃Have 2 sub-goals: warehouse-in minimum with And outbound is minimum, go out to put in storage the object function z that convenient degree is the highest₃For:

z₃=T_rk+T_ck

Wherein, T_rcFor time-consumingly counting outside warehouse-in total value；T_ckFor time-consumingly counting outside outbound total value；

In sum, the general objective function I of Hui Ku station track Arrangement Problem is:

$I=lex\min \{z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Work as z₁=0, z₂=0, z₃When=0, the station track corresponding to I arranges scheme to be required optimal solution.

Further, step (4.3) is described determines fitness function, comprises the following steps:

The first step: calculate desired value, there are 3 object functions, i.e. in each individuality

$lex\min \{z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Second step: individuality is ranked up according to the first priority target functional value；If some individuality has identical Target function value, then compare, by that analogy the second priority target function；If some individuality finally cannot sort, The most randomly ordered；The so individual order arrangement raised according to target function value；

3rd step: for each individual x_kDistribute adaptive value based on sequence, if r_kIt is individual x_kPut in order, according to Parameter a ∈ (0,1) the definition adaptive value function based on sequence that family gives is as follows:

$eval\left(x_k\right)=a{(1-a)}^{r_k-1}$

Compared with prior art, its remarkable advantage is the present invention: (1) meets rolling stock section Hui Ku station track and arranges reality, operation Simply, quickly station track can be distributed for storehouse to be returned vehicle；(2) rationally train mission can be accepted or rejected, and find out resource constraint bar Vehicle Hui Ku station track scheme optimum under part, is either all substantially better than artificial arrangement in efficiency or in reasonability.

Accompanying drawing explanation

Fig. 1 is the flow chart of metro depot Hui Ku station track of the present invention arrangement method.

Fig. 2 is available station track State Encoding Style schematic diagram.

Fig. 3 is the scheme that the picks up schematic diagram that need to consider back storehouse time interval.

Fig. 4 is certain depot's conspectus.

Fig. 5 is proxy target function trendgram in embodiment 1.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is described in further detail.

In conjunction with Fig. 1～3, metro depot Hui Ku station track of the present invention arrangement method, comprise the following steps:

Step 1, storehouse the to be returned information of vehicles needed for collection algorithm and station track information, and set up the parameter set of practical problem, Storehouse to be returned information of vehicles includes license number, whether Hui Ku, vehicle location, institute's sport car, returns the storehouse time, returns storehouse order, tomorrow Whether run morning peak, whether carwash, whether daily test, park station track type, whether tomorrow reaches the standard grade, station track information includes station track name Title, station track type, taking situation and station track coding etc., the parameter set of practical problem includes: population scale Z, probability of crossover p_c, become Different Probability p_m, maximum evolutionary generation etc..

Step 2, chromosome constructs: on the basis of analyzing Hui Ku station track Arrangement Problem coding requirement, selects suitable coding Chromosome is constructed by mode, particularly as follows: using a gene as an available station track, train returns storehouse order and inserts with table Show track occupied；Available Necessary Number of Tracks when every day, vehicle went back to storehouse is indefinite, so dynamically entering chromosome according to practical situation Row structure, constructs chromosome with the coded system of one-dimensional matrix, and each of chromosome is all relative with a certain available station track numbering Should.

Step 3, initializes population P: combine vehicle and return the actual formulation in storehouse initialization of population rule, and generate according to this rule Z chromosome constitutes initial population P, specifically comprises the following steps that

(3.1) travel through storehouse to be returned vehicle, obtain back storehouse vehicle number and station track requires and returns the priority of storehouse time according to expectation Vehicle is numbered；

(3.2) available station track state is determined.When same station track A-share, B stock all can use, 1. A-share state is, 2. B stock is； When A-share take, B stock can use time, 3. B strand state is；When B stock take, A-share can use time, 4. A-share state is；

(3.3) combine Fig. 2, determine that vehicle returns storehouse numbering and inserts the rule of chromosome:

For there being the vehicle of daily test operation night, 1., 2., 4. or other have the stock of trench can park the state of station track is Road；

For there being the vehicle of carwash operation night, station track state can be parked for 1., 2., 3.；

For morning peak need to be arranged tomorrow or have the vehicle of appointment train number, the state of station track can be parked for 1., 4.；

For there being tomorrow system to repair the vehicle of operation, can park station track is the station track having protection network and trench；

Other are had and specifies the vehicle parking station track type, the station track of corresponding types should be arranged as far as possible to park；

For having two and the vehicle of above job requirements, station track can be parked and to ask for what all individual event operation station tracks required Occur simultaneously.Occuring simultaneously for, time empty, upkeep operation requirement need to being ranked up by importance when station track can be parked, cancelling part significance level Low job requirements or job content.Job requirements importance is ranked up: daily test system is repaiied and other station track types are specified > carwash operation > and morning peak or specify train number.

(3.4) select to insert station track and available station track state is updated according to the rule in (3.3), be repeated up to Being needed back storehouse vehicle all inserts chromosome.

(3.5) (3.4) are repeated until generating Z chromosome to constitute initial population P.

Step 4, calculates fitness and also preserves optimum individual: determine optimization object function and constraint that Hui Ku station track arranges, Calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved, specific as follows:

(4.1) determining constraints, wherein constraints includes time interval constraint and metro depot service ability Constraint, wherein:

In conjunction with Fig. 3, time interval is retrained, has two kinds of situations:

Situation a, first train elder generation Hui Ku is parked in A-share and carries out daily test operation, and operation is adjusted to B stock after completing, A-share empties Second train of follow-up continued access, first train is whole pick up during, the time-consuming 40min of daily test, shunt time-consuming 30min, therefore First is more than or equal to 70min with the storehouse time interval of returning of second train；When returning storehouse time interval less than 70min, B stock car Daily test task cannot complete, directly stop to B stock；

Situation b, first train first passes through washing track and docks to A-share from dextrad and carry out daily test operation, and operation is adjusted after completing To B stock, A-share empties second train of follow-up continued access, first train is whole pick up during, the time-consuming 15min of carwash, daily test Time-consuming 40min, shunt time-consuming 30min, therefore first is more than or equal to 85min with the storehouse time interval of returning of second train；When One with second train return storehouse time interval less than 85min and more than 70min time, abandon carwash task and only carry out daily test； When time storehouse time interval of first with second train is less than 70min, abandons daily test task and only carry out carwash；

Employing formula is expressed as follows:

Wherein TimeA_jThe storehouse time is estimated back for A-share train；TimeB_jThe storehouse time is estimated back for B stock train.

For the constraint of metro depot service ability, number of tasks of shunting of same time is not more than 1, and every other day time same Inspection vehicle number is not more than 3, it may be assumed that

$\left\{\begin{array}{c}{N}_{di}\≤3\\ N_{dc}\≤1\end{array}\right.$

Wherein, N_diVehicle number for same time daily test；N_dcNumber of tasks of shunting for the same time.

(4.2) object function is determined: arranging scheme for obtaining the most efficient Hui Ku station track, optimization aim is considered as following Some: task completeness is the highest；The process that picks up always is shunt least number of times；Vehicle goes out to put convenient degree in storage the highest.

Owing to three optimization aim have obvious priority level level, therefore it is considered as priority goal programming and solves Hui Ku station track decision-making problem of multi-objective, wherein task completeness is the highest has the first importance, obtains first and gives priority to；Shunt Least number of times has the second importance, obtains second and gives priority to；Go out to put in storage that convenient degree is the highest has the 3rd importance, obtain 3rd gives priority to.For target z_jPriority that (j=1,2,3) reaches and each target need the sub-goal reached, target The expression formula of function is as follows:

$lex\min \{z_1=\underset{i=1}{\overset{m_0}{\Σ}}(w_{1i}^{+}{d}_i^{+}+w_{1i}^{-}{d}_i^{-}),z_2=\underset{i=1}{\overset{m_0}{\Σ}}(w_{2i}^{+}{d}_i^{+}+w_{2i}^{-}{d}_i^{-}),z_3=\underset{i=1}{\overset{m_0}{\Σ}}(w_{3i}^{+}{d}_i^{+}+w_{3i}^{-}{d}_i^{-})\}$

Wherein, lexmin represents and minimizes target according to lexicographic order；m₀Represent sub-goal quantity；Represent i-th mesh Mark exceedes intended overgauge variable；Represent i-th target and exceed intended minus deviation variable；Representative is distributed to's Positive weights；Representative is distributed toNegative weight；

It is the highest for task completeness, if object function is z₁, it has 4 sub-goals: daily test completion rate of the plan is the highest, car Park that station track type and its operation station track require that matching degree is the highest, carwash completion rate of the plan is the highest, morning peak station track requires Degree of joining is the highest, optimally, it is desirable to each sub-goal all reaches the completion rate of 100%, gives by sub-goal importance The weight that each target is different, owing to the rate that actually accomplishes of sub-goal is not over 100%, therefore only need toAssignment, It is respectively 1000,100,10,1.Therefore the object function z that task completeness is the highest₁For:

$z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X$

Wherein, T_iuIt is not fully complete number of tasks for i-th sub-goal；X is penalty coefficient；For distributing to each sub-goal Negative weight.

When the daily test vehicle number in the same time period or number of tasks of shunting are more than constraint, station track arranges scheme infeasible, Now penalty coefficient X is taken bigger value, makes X=10000, make the fitness of the program diminish, then the program follow-up enter The probability being eliminated during change will become big.

For least number of times of shunting, if object function is z₂, A, B two strands of and if only if same station track is both needed to park and treats Return storehouse vehicle and B stock vehicle need and can daily test time, need to carry out shunting service, the most ideally, it is desirable to whole Returning number of times of shunting in the operation of storehouse is 0, the object function z of least number of times of shunting₂For:

z₂=T_dc

Wherein, T_dcFor actual number of times of shunting；

It should be noted that when A, B two strands of same station track is both needed to park storehouse to be returned vehicle and B stock vehicle comprises daily test During operation, need when B stock daily test operation that and if only if can complete to shunt.

It is the highest for going out to put in storage convenient degree, if object function is z₃, it has 2 sub-goals: warehouse-in is minimum And outbound is minimum, the most ideally, it is desirable to warehouse-in is extra time-consuming and outbound is the most time-consumingly 0, two Sub-goal is of equal importance, therefore goes out to put in storage the object function z that convenient degree is the highest₃For:

z₃=T_rk+T_ck

Wherein, T_rcFor time-consumingly counting outside warehouse-in total value；T_ckFor time-consumingly counting outside outbound total value.

During vehicle returns it should be noted that and if only if storehouse from right side warehouse-in and vehicle without carwash task time, put total value in storage Outer time-consuming number adds 1；The A-share of and if only if certain station track is unavailable or the second day that can't park cars car and B stock parks cars second When day needs car, outside outbound total value, time-consuming number adds 1.

In sum, the general objective function I of Hui Ku station track Arrangement Problem is:

$I=lex\min \{z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Work as z₁=0, z₂=0, z₃When=0, the station track corresponding to I arranges scheme to be required optimal solution.

For preferably representing the trend of evolution of each generation population, a proxy target function I'=100z is set₁+10z₂+z₃, The least representative of desired value is closer to globally optimal solution.

(4.3) fitness function is determined: adaptive value calculates and comprises the following steps:

The first step: calculate desired value.3 object functions of each individual existence, i.e.

$lex\min \{z_1=\underset{i=1}{\overset{4}{\Σ}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Second step: individuality is ranked up according to the first priority target functional value.If some individuality has identical Target function value, then compare, by that analogy its second priority target function.If some individuality finally cannot be arranged Sequence, then allow them randomly ordered.The so individual order just raised according to target function value is lined up.

3rd step: for each individual x_kDistribute adaptive value based on sequence.If r_kIt is individual x_kPut in order.According to Parameter a ∈ (0,1) the definition adaptive value function based on sequence that family gives is as follows:

$eval\left(x_k\right)=a{(1-a)}^{r_k-1}$

(4.4) calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved.

Step 5, selects, hybridizes, makes a variation: for coded system, the initialization criterion of the arrangement of Hui Ku station track, to individual choosing The concrete operations mode selecting rule and hybridization, variation is determined, specific as follows:

(5.1) the selection rule of individuality is determined.This algorithm use roulette come in selected population for crossover operation Body, if the adaptive value that in population, kth is individual is F_k, then the individual selected probability P of kth_kFor:

$P_k=\frac{F_k}{\underset{i=1}{\overset{Z}{\Σ}}{F}_i}$

(5.2) determining the crossover operation mode of individuality, crossover process comprises the following steps:

The first step: generation each to individuality to Z/2 one (0,1] random number, if random number is less than probability of crossover p_cThen enter Row crossover operation, on the contrary do not hybridize, directly as in new individual preservation to new population；

Second step: for needing the individuality of hybridization, randomly generate two [0, N_track) integer as hybridization location, to often Individual parent selectes hybridization portion；

3rd step: the hybridization portion of first parent and the hybridization portion of second parent are interchangeable；

4th step: remove all trains repeated from the generation original gene of offspring and return storehouse sequence number；

5th step: the gene producing offspring is carried out completion according to initialization of population principle, and will new individual preserve to newly In population；

6th step: repeat the first step to step the five step until new population scale reaches Z.

(5.3) determining the mutation operation mode of individuality, mutation process comprises the following steps:

The first step: in new population every individual each produce one (0,1] random number, if random number is less than mutation probability p_m Then carry out mutation operation, otherwise the most do not carry out mutation operation；

Second step: for needing the individuality of variation, randomly generate two [0, N_track) integer as catastrophe point, and will be prominent Genic value in height swaps；

3rd step: perform second step～the 3rd step in all individualities of population.

Step 6, produces population P of future generation: according to the selection determined in step 5, hybridize, make a variation rule, choosing from population Go out Z individuality carry out hybridizing, mutation operation preserving to new population；

Step 7, end condition: repeat step 4 to step 6 until reaching maximum evolutionary generation, optimum now preserved Body is the optimal solution of the solved problem of algorithm, carries out picking up according to this optimal solution and can realize metro depot Hui Ku station track Optimum arrangements.

Embodiment 1

Certain metro depot line arrangement schematic diagram as shown in Figure 4, vehicle from left entrance rolling stock section, night main diamond Territory is for using storehouse (4 station tracks are to 16 station tracks) and maintenance storehouse (17 station tracks are to 19 station tracks), and wherein A-share road band trench, B station track is without ground Ditch, 17 to 19 station tracks are provided with trench and protection network, vehicle can from the left or right both direction go out warehouse-in, wherein dextrad go out warehouse-in need through Walk line.

The effectiveness of verification algorithm, sets up metro depot Hui Ku station track and arranges emulation platform.Rule of thumb algorithm is joined Number be configured, if population scale Z be 40, probability of crossover p_cBe 0.8, mutation probability p_mIt is 0.1, maximum evolutionary generation is set is 20, fitness function variable a is 0.5.State when first picking up information of vehicles and station track is added up, and has 15 and treats back Storehouse vehicle, and reserved 15 empty station tracks supply back storehouse storing cycle.It is then based on multiple target theory application genetic algorithm to returning storehouse stock Road Arrangement Problem is optimized, and represents the trend of evolution of population with proxy target functional value, as it is shown in figure 5, algorithm is from the 8th In generation, starts to converge to optimal solution.

Table 1 emulation contrasts scheme with manual approach

Arranging scheme to compare the station track manually obtained with emulation, be shown in Table 1, simulating scheme is the most excellent in each index Arrange in artificial.In summary, under extreme conditions, train mission can rationally be accepted or rejected by this algorithm, and finds out resource Vehicle Hui Ku station track scheme optimum under constraints, is either all substantially better than artificial peace in efficiency or in reasonability Row.

Claims (9)

1. a metro depot Hui Ku station track arrangement method, it is characterised in that comprise the following steps:

Step 1, collects storehouse to be returned information of vehicles and station track information, and sets up the parameter set of practical problem；

Step 2, chromosome constructs: on the basis of analyzing Hui Ku station track Arrangement Problem coding requirement, selects coded system to dye Colour solid constructs；

Step 3, initializes population P: combine vehicle and return the actual formulation in storehouse initialization of population rule, and generate Z according to this rule Chromosome constitutes initial population P；

Step 4, calculates fitness and also preserves optimum individual: determine optimization object function and constraint that Hui Ku station track arranges, according to Fitness function calculates the adaptive value of each individuality, and preserves the individuality that adaptive value is maximum；

Step 5, selects, hybridizes, makes a variation: for coded system, the initialization criterion of the arrangement of Hui Ku station track, to individual selection rule Then and hybridization, variation concrete operations mode be determined；

Step 6, produces population P of future generation: according to the selection determined in step 5, hybridize, make a variation rule, selects Z individual from population Individuality carries out hybridizing, mutation operation preserving to new population；

Step 7, end condition: repeating step 4 to step 6 until reaching maximum evolutionary generation, the optimum individual now preserved is i.e. By the optimal solution of the solved problem of algorithm, carry out picking up the optimum that can realize metro depot Hui Ku station track according to this optimal solution Arrange.

Metro depot Hui Ku station track the most according to claim 1 arrangement method, it is characterised in that described in step 1 Storehouse to be returned information of vehicles includes license number, whether Hui Ku, vehicle location, institute's sport car, returns the storehouse time, whether returns storehouse order, tomorrow Run morning peak, whether carwash, whether daily test, park station track type, whether tomorrow reaches the standard grade；Station track information includes station track title, stock Road type, take situation and station track coding；The parameter set of practical problem includes: population scale Z, probability of crossover p_c, mutation probability p_m, maximum evolutionary generation.

Metro depot Hui Ku station track the most according to claim 1 arrangement method, it is characterised in that described in step 2 Chromosome constructs, particularly as follows: using a gene as an available station track, train returns storehouse order and inserts to represent that station track accounts for With；Available Necessary Number of Tracks when every day, vehicle went back to storehouse is indefinite, so dynamically chromosome being constructed according to practical situation, with The coded system structure chromosome of one-dimensional matrix, each of chromosome is all corresponding with a certain available station track numbering.

Metro depot Hui Ku station track the most according to claim 1 arrangement method, it is characterised in that described in step 3 Initialize population P, particularly as follows:

(3.1) travel through storehouse to be returned vehicle, obtain back storehouse vehicle number and station track requires and returns the priority of storehouse time according to expectation to car It is numbered；

(3.2) determining available station track state: when same station track A-share, B stock all can use, 1. A-share state is, 2. B stock is；Work as A-share Take, B stock is when can use, and 3. B strand state is；When B stock take, A-share can use time, 4. A-share state is；

(3.3) determine that vehicle returns storehouse numbering and inserts the rule of chromosome:

For there being the vehicle of daily test operation night, 1., 2., 4. or other have the station track of trench can park the state of station track is；

For there being the vehicle of carwash operation night, station track state can be parked for 1., 2., 3.；

For morning peak need to be arranged tomorrow or have the vehicle of appointment train number, the state of station track can be parked for 1., 4.；

For there being tomorrow system to repair the vehicle of operation, can park station track is the station track having protection network and trench；

Other are had and specifies the vehicle parking station track type, arrange the station track of corresponding types to park；

For having two and the vehicle of above job requirements, station track can be parked and to ask for the friendship that all individual event operation station tracks require Collection, occurs simultaneously for, time empty, being ranked up upkeep operation requirement by importance when parking station track, cancels part significance level low Job requirements or job content, be ranked up job requirements importance: daily test > system is repaiied and other station track types are specified > wash Car operation > morning peak or specify train number；

(3.4) select insert station track and available station track state is updated according to the rule in (3.3), be repeated up to be needed Return storehouse vehicle and all insert chromosome；

(3.5) (3.4) are repeated until generating Z chromosome to constitute initial population P.

Metro depot Hui Ku station track the most according to claim 1 arrangement method, it is characterised in that count described in step 4 Calculate fitness and preserve optimum individual: determining optimization object function and constraint, calculating the adaptation of each individuality according to fitness function Value, and the individuality that adaptive value is maximum is preserved, specific as follows:

(4.1) determining constraints, wherein constraints includes the constraint of time interval constraint and metro depot service ability；

(4.2) determining object function: optimization aim is that task completeness is the highest, the process that picks up always is shunt least number of times, and vehicle goes out Put convenient degree in storage the highest；

(4.3) fitness function is determined；

(4.4) calculate the adaptive value of each individuality according to fitness function, and the individuality that adaptive value is maximum is preserved.

Metro depot Hui Ku station track the most according to claim 1 arrangement method, it is characterised in that select described in step 5, Hybridization, variation: for coded system, the initialization criterion of the arrangement of Hui Ku station track, to individual selection rule and hybridization, make a variation Concrete operations mode be determined, specific as follows:

(5.1) determining the selection rule of individuality, employing roulette carrys out the individuality in selected population for crossover operation, if in population The adaptive value of kth individuality is F_k, then the individual selected probability P of kth_kFor:

$P_k=\frac{F_k}{\underset{i=1}{\overset{Z}{\Σ}}{F}_i}$

(5.2) determining the crossover operation mode of individuality, crossover process comprises the following steps:

The first step: generation each to individuality to Z/2 one (0,1] random number, if random number is less than probability of crossover p_cThen hybridize Operation, on the contrary do not hybridize, directly as in new individual preservation to new population；

Second step: for needing the individuality of hybridization, randomly generate two [0, N_track) integer as hybridization location, to each father Generation selected hybridization portion；

3rd step: the hybridization portion of first parent and the hybridization portion of second parent are interchangeable；

4th step: remove all trains repeated from the generation original gene of offspring and return storehouse sequence number；

5th step: the gene producing offspring is carried out completion according to initialization of population principle, and will new individual preserve to new population In；

6th step: repeat the first step～the 5th step until new population scale reaches Z；

(5.3) determining the mutation operation mode of individuality, mutation process comprises the following steps:

The first step: in new population every individual each produce one (0,1] random number, if random number is less than mutation probability p_mThen enter Row variation operates, otherwise does not the most carry out mutation operation；

Second step: for needing the individuality of variation, randomly generate two [0, N_track) integer as catastrophe point, and by catastrophe point On genic value swap；

3rd step: perform second step～the 3rd step in all individualities of population.

Metro depot Hui Ku station track the most according to claim 5 arrangement method, it is characterised in that step (4.1) is described Determining constraints, wherein constraints includes the constraint of time interval constraint and metro depot service ability, wherein:

Time interval is retrained, has two kinds of situations:

Situation a, first train elder generation Hui Ku is parked in A-share and carries out daily test operation, and operation is adjusted to B stock after completing, A-share empties follow-up Second train of continued access, first train is whole pick up during, the time-consuming 40min of daily test, shunt time-consuming 30min, therefore first With second train return storehouse time interval more than or equal to 70min；When returning storehouse time interval less than 70min, B stock vehicle Daily test task cannot complete, and directly stops to B stock；

Situation b, first train first passes through washing track and docks to A-share from dextrad and carry out daily test operation, and operation is adjusted to B after completing Stock, A-share empties second train of follow-up continued access, first train is whole pick up during, the time-consuming 15min of carwash, daily test consumes Time 40min, shunt time-consuming 30min, thus first with second train return storehouse time interval more than or equal to 85min；When first With second train return storehouse time interval less than 85min and more than 70min time, abandon carwash task and only carry out daily test；When First with second train return storehouse time interval less than 70min time, abandon daily test task and only carry out carwash；

For the constraint of metro depot service ability, number of tasks of shunting of same time is not more than 1, and same time daily test car Number is not more than 3, it may be assumed that

$\left\{\begin{array}{c}{N}_{di}\≤3\\ N_{dc}\≤1\end{array}\right.$

Wherein, N_diVehicle number for same time daily test；N_dcNumber of tasks of shunting for the same time.

Metro depot Hui Ku station track the most according to claim 5 arrangement method, it is characterised in that step (4.2) is described Determining object function: optimization aim is that task completeness is the highest, the process that picks up always is shunt least number of times, and vehicle goes out puts convenient journey in storage Spending the highest, wherein task completeness is the highest has the first importance, and least number of times of shunting has the second importance, and it is convenient to go out to put in storage Degree is the highest has the 3rd importance；For target z_j(j=1,2,3) priority reached and each target need the son reached Target, the expression formula of object function is as follows:

$lexmin\{z_1=\underset{i=1}{\overset{m_0}{\Σ}}(w_{1i}^{+}{d}_i^{+}+w_{1i}^{-}{d}_i^{-}),z_2=\underset{i=1}{\overset{m_0}{\Σ}}(w_{2i}^{+}{d}_i^{+}+w_{2i}^{-}{d}_i^{-}),z_3=\underset{i=1}{\overset{m_0}{\Σ}}(w_{3i}^{+}{d}_i^{+}+w_{3i}^{-}{d}_i^{-})\}$

Wherein, lexmin represents and minimizes target according to lexicographic order；m₀Represent sub-goal quantity；Represent i-th target to exceed Intended overgauge variable；Represent i-th target and exceed intended minus deviation variable；Representative is distributed toPositive weights；Representative is distributed toNegative weight；

It is the highest for task completeness, if object function is z₁, z₁There are 4 sub-goals: daily test completion rate of the plan is the highest, vehicle stops Put station track type and require that matching degree is the highest, carwash completion rate of the plan is the highest, morning peak station track requires matching degree with its operation station track The highest, that task completeness is the highest object function z₁For:

$z_1=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{w}_{1i}^{-}\×T_{iu}+X$

Wherein, T_iuIt is not fully complete number of tasks for i-th sub-goal；X is penalty coefficient；For distributing to the negative power of each sub-goal Weight；

For least number of times of shunting, if object function is z₂, A, B two strands of and if only if same station track is both needed to park and treats Hui Ku Vehicle and B stock vehicle need and can daily test time, need to carry out shunting service, the object function z of least number of times of shunting₂For:

z₂=T_dc

Wherein, T_dcFor actual number of times of shunting；

It is the highest for going out to put in storage convenient degree, if object function is z₃, z₃There are 2 sub-goals: warehouse-in is minimum and goes out Storehouse is minimum, goes out to put in storage the object function z that convenient degree is the highest₃For:

z₃=T_rk+T_ck

Wherein, T_rcFor time-consumingly counting outside warehouse-in total value；T_ckFor time-consumingly counting outside outbound total value；

In sum, the general objective function I of Hui Ku station track Arrangement Problem is:

$I=lexmin\{z_1=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Work as z₁=0, z₂=0, z₃When=0, the station track corresponding to I arranges scheme to be required optimal solution.

Metro depot Hui Ku station track the most according to claim 5 arrangement method, it is characterised in that step (4.3) is described Determine fitness function, comprise the following steps:

The first step: calculate desired value, there are 3 object functions, i.e. in each individuality

$lexmin\{z_1=\underset{i=1}{\overset{4}{\mathrm{\Σ}}}{w}_{1i}^{-}\×T_{iu}+X,z_2=T_{dc},z_3=T_{rk}+T_{ck}\}$

Second step: individuality is ranked up according to the first priority target functional value；If some individuality has identical target Functional value, then compare, by that analogy the second priority target function；If some individuality finally cannot sort, then with Machine sorts；The so individual order arrangement raised according to target function value；

3rd step: for each individual x_kDistribute adaptive value based on sequence, if r_kIt is individual x_kPut in order, give according to user Fixed parameter a ∈ (0,1) definition adaptive value function based on sequence is as follows:

$eval\left(x_k\right)=a{(1-a)}^{r_k-1}.$