CN109492797A - Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations - Google Patents
Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations Download PDFInfo
- Publication number
- CN109492797A CN109492797A CN201811200659.6A CN201811200659A CN109492797A CN 109492797 A CN109492797 A CN 109492797A CN 201811200659 A CN201811200659 A CN 201811200659A CN 109492797 A CN109492797 A CN 109492797A
- Authority
- CN
- China
- Prior art keywords
- population
- vector
- global
- local
- evolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 43
- 230000001568 sexual effect Effects 0.000 title claims abstract description 17
- 230000007246 mechanism Effects 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 230000000737 periodic effect Effects 0.000 claims abstract description 7
- 238000005457 optimization Methods 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 9
- 230000035772 mutation Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000008929 regeneration Effects 0.000 claims description 4
- 238000011069 regeneration method Methods 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 3
- 238000012546 transfer Methods 0.000 abstract description 19
- 238000004088 simulation Methods 0.000 abstract description 4
- 230000008859 change Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000018199 S phase Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06314—Calendaring for a resource
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Biophysics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Operations Research (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Physiology (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Genetics & Genomics (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of methods for leading to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations, it is related to Intelligent evolution to calculate and traffic scheduling field, by being optimized constantly to each originating for service route vehicle in periodic traffic network, to realize the target for the waiting time for minimizing passenger's transfer.Invention introduces a kind of parameter based on cooperation on multiple populations and operator controlling mechanisms, enhance efficiency and robustness that differential evolution algorithm solves periodic traffic timetable Problems of Optimal Dispatch, the ability that differential evolution algorithm jumps out local optimum is improved, and reduces differential evolution algorithm to the sensibility of parameter.Emulation testing is carried out by taking Subway Network in China and simulation railway network as an example, it was demonstrated that the method for invention is highly effective.
Description
Technical field
The present invention relates to Intelligent evolutions to calculate and traffic scheduling technical field, and in particular to a kind of poor with cooperation on multiple populations
The method for dividing the sexual intercourse of evolution algorithm optimizing cycle to lead to scheduling instance table.
Background technique
Periodic traffic timetable Problems of Optimal Dispatch has very important significance in reality.Currently, the moment in period
Table has been widely used in the systems such as railway, subway and public transport.In order to avoid conflicting between vehicle, it sets vehicle and arrives
At the time of standing and is outbound, and its periodicity is provided for passenger and is more fast serviced with convenient.In a complicated friendship
In open network, the waiting time for changing to different routes is the important performance indexes for assessing its service quality.Researcher proposes perhaps
The method of more setting cycle traffic timetables, for minimizing the waiting time of passenger's transfer.The currently used setting period hands over
The method of logical timetable has integer programming, branch and bound method and other nonlinear technologies.They can be divided into two classes, a kind of
It is traditional method, one kind is heuristic search.
Problem in view of this, such as integer programming, branch and bound method of traditional method need huge when solving extensive problem
Big memory space and calculation amount.And such as genetic algorithm, simulated annealing heuristic search, when solving this problem
Need to set dispatching a car the period for service route fixation.But in real life, the vehicle of each service route is dispatched a car often and is had
The characteristic of variable period, for example when peak on and off duty since passenger is excessive, the period of dispatching a car is often smaller than low peak period.On solving
State disadvantage, it would be highly desirable to give a kind of mathematical model and solution for constructing high quality variable cycle time-table, and solve
Certainly the differential evolution algorithm in scheme using cooperation on multiple populations is to solve the problem.Evolution algorithm is with Darwinian evolutionism think of
Based on thinking, pass through the self-organizing of the Solve problems of simulation biological evolution process and mechanism, adaptive artificial intelligence technology.It is raw
Object evolution is realized by breeding, variation, competition and selection；And evolution algorithm is then mainly by selecting, recombinating and make a variation this
The solution of optimization problem is realized in three kinds of operations.In numerous heuristic evolution algorithms, differential evolution (Differential
It EvolutonDE is) a kind of heuristic random searching algorithm based on population difference, by R.Storn and K.Price in nineteen ninety-five
It proposes.Differential evolution algorithm principle is simple, and controlled parameter is few, strong robustness.Optimize in constrained optimization calculating, cluster and calculates, is non-
Linear optimization control, Neural Network Optimization, filter design etc. are used widely.And periodic traffic moment list scheduling
Optimization problem is exactly the multi-modal optimization problem for containing a large amount of constraint conditions, has complexity between variable in this mathematical model
Incidence relation.For complicated multi-modal optimization problem, traditional differential evolution algorithm easily falls into local optimum words, and has
The disadvantages of extremely strong sensitivity to parameter.
Summary of the invention
The purpose of the present invention is to solve drawbacks described above in the prior art, provide a kind of with cooperation difference on multiple populations
The method of scheduling instance table is led in evolution algorithm optimizing cycle sexual intercourse.
The purpose of the present invention can be reached by adopting the following technical scheme that:
A method of leading to scheduling instance table, the side with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method includes:
Initialization population step originates moment range as boundary, according to apart from the lower bound moment using periodic traffic route
Time interval generate global population P at randomG(Global) and local population PL(Local), and according to valuation functions (target letter
Number) calculate population in all individuals assessed value, wherein global population PGWith local population PLAccording to represented by following formula:
WhereinWithRespectively global population PGWith local population PLI-th of individual, g indicates iterative evolution
Algebra, GN and LN respectively indicate global population PGWith local population PLPopulation number, population at individual (object vector) with as follows to
Amount form indicates:
Wherein D indicates the total number of traffic route,Indicate that the j-th strip route of i-th of individual in population originates the moment
With the interval time between the lower bound moment；
Global Evolution of Population step, for global population PGIn each object vector, according to the variation plan of global population
A variation vector is slightly generated, and calculates the assessed value of this variation vector according to valuation functions (objective function), by the vector that makes a variation
Assessed value be compared with the assessed value of corresponding initial target vector, if variation vector assessed value be parity with or superiority over correspondence
The variation vector is then replaced corresponding initial target vector, and enters the next of population by the assessed value of initial target vector
Generation；
Local Evolution of Population step, for local population PLIn each object vector, according to the variation plan of local population
A variation vector is slightly generated, and calculates the assessed value of this variation vector according to valuation functions (objective function), by the vector that makes a variation
Assessed value be compared with the assessed value of corresponding initial target vector, if variation vector assessed value be parity with or superiority over correspondence
The variation vector is then replaced corresponding initial target vector, and enters the next of population by the assessed value of initial target vector
Generation；
Population cooperation step, global population PGWith local population PLBy comparing optimal objective vector between simultaneously Population Regeneration
Population cooperation mode improves population diversity and search efficiency, if local population PLOptimal objective vector better than global kind
Group PGOptimal objective vector, then by local population PLOptimal objective vector replace global population PGOptimal objective vector；Instead
It, then by global population PGOptimal objective vector replace local population PLOptimal objective vector；Meanwhile in local population PL
Optimal objective vector be better than global population PGIn the case where optimal objective vector, if global population PGOptimal objective vector
Better than local population PLWorst object vector, then by global population PGOptimal objective vector replace local population PLIt is worst
Object vector；
Terminate judgment step, in the process of implementation if the number of iterations is more than defined maximum evolutionary generation or obtains at this time
Optimal solution meet certain error requirements (if error is less than 5%) if terminate optimization, otherwise repeat global Evolution of Population
Step, local Evolution of Population step and population cooperation step.
Further, in the global Evolution of Population step,
Zoom factor F in global population is set by the way of being randomly providedGWith crossover probability CRGValue, it may be assumed that
FG=rand (0,1)
CRG=rand (0,1).
Further, in the global Evolution of Population step, using DE/rand/1 mutation operator.
Further, in the global Evolution of Population step, the operation of a random variation is introduced, as global population PG
Execute crossover operation and then execute random variation operation, wherein the process of random variation as shown by the following formula:
WhereinIndicate global population PGI-th of individual jth dimension variable value, g indicates the algebra of iterative evolution, LBj
And UBjThe lower bound of the jth dimension variable of problem respectively to be solved and the upper bound, MAXEVALS and evals are respectively maximum adaptation degree
Number and current Fitness analysis number are assessed,The value that jth ties up variable is corresponded to for optimum individual, pm is one given in advance
Numerical value is operated random variation and is made a variation with the Probability p m of very little to the test vector of generation.
Further, in the global Evolution of Population step and local Evolution of Population step, it is all made of association on multiple populations
Make mode of evolution.
Further, in the local Evolution of Population step, zoom factor FLValue be provided that
FL=rand (0,0.8).
Further, in the local Evolution of Population step, crossover probability CRLValue be provided that
Wherein, parameter ξ is set as 0.5 at the beginning.
Further, during evolution, if in parameter setting CRLIt is better than or is equal to part in the case where=0
The new individual of population optimum individual, then ξ will be increased to increase CRL=0 select probability, conversely, if in parameter setting CRL
In the case where=1, then ξ will decrease to increase CRL=1 select probability, the scaling process of parameter ξ as shown by the following formula:
Wherein, α ∈ [0,1] is learning rate.
Further, in the local Evolution of Population step, generated using DE/best/1 Variation mechanism variation to
Amount.
Further, from global population PGMiddle random selection individual generates the difference vector in DE/best/1, it may be assumed that
Wherein, r1, r2 ∈ [1, GN] are two mutually different integers,It is local population PLIn optimized individual,
FLFor the zoom factor in local population,WithRespectively global population PGR1 and the r2 individual.
The present invention has the following advantages and effects with respect to the prior art:
Differential evolution algorithm of the present invention based on cooperation on multiple populations has compared to traditional differential evolution algorithm
There are the stronger search capability for jumping out local optimum and search efficiency.Global population PGWith local population PLDifferent ginsengs is respectively adopted
The control of several and operator, and population diversity is improved by comparing the population cooperation mode of the optimal objective vector between Population Regeneration
With search efficiency.
Detailed description of the invention
Fig. 1 is disclosed by the invention with the logical scheduling instance table of cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method flow diagram.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment
Assuming that a transportation network has M traffic route altogether, it is expressed as set { r1,r2,r3,...,rM}.In the network of communication lines
In network, the crosspoint between route and route can be referred to as transfer point, i.e., can carry out between different routes in this website passenger
Transfer.It is assumed that sharing N number of transfer point in entire transportation network, it is expressed as set { p1,p2,...,pN}.Meanwhile for periodicity
Traffic, each traffic route need periodically to dispatch a car from the starting station.If in different times in section, dispatching a car the period always
It remains unchanged, is then properly termed as fixed cycle traffic, otherwise referred to as variable period traffic.Solution target of the invention is to be directed to change
Cycle traffic, at the time of optimizing all vehicles and leave each website, so that the average transfer stand-by period and minimizes of all passengers.
Traffic scheduling time-averaged pressure with variable period has the property that
1) the vehicle outbound time: route riOn train meet following rule by the time of continuous two websites:
Wherein, S (i, j) indicates route riIn j-th of website,Indicate n-th be periodically sent out from the starting station
At the time of vehicle leaves website S (i, j),Indicate traveling required for vehicle slave site S (i, j-1) to website S (i, j)
Time, US(i,j)It indicates to stop the time received guests when vehicle reaches website S (i, j).
2) variable period model: service route riService time Range-partition be multiple time interval [ti,1,ti,2],
[ti,2,ti,3] ..., and each time interval is associated with a cycle H of dispatching a cari,j(such as [ti,1,ti,2] it is associated with cycle H of dispatching a cari,1).Hair
Vehicle cycle Hi,jIt represents in time interval [ti,j,ti,j+1] in, route riOn period for being outputed from the starting station of train be Hi,j.This
Outside, also with the time interval and difference from a service route to transfer passenger's number of another service route.Usually exist
Transfer passenger's number of peak period (period such as on and off duty) will be more than low peak period.
It is assumed that train runs to time required for another website and train in the stop of a website from a website
Time is known.Moreover, in some time interval, from a service route to transfer passenger's number of another service route
It can pre-estimate out.So, there is fixed cycle time-table scheduling problem (i.e. Hi=Hi,j, j=1,2 ...)
It can model as follows.It is assumed that the time that originates of all M service routes is b1,b2,...,bM, then formula (1) can be with table
It reaches are as follows:
When the waiting time of running time and each website between website is fixed, vehicle each website the outbound time by
Originate the moment and and the route period decision of dispatching a car, shown in specific mathematical relationship such as formula (2).According to this conclusion, if
Given dispatching a car the period for all routes, then the timetable of optimization whole cycle transportation network is equivalent to optimize all routes
Originate the time.Assuming that route riWith route rjThere is common website pk, be expressed as in respective route S (i, f) and S (j,
D), then from route riOn n-th vehicle pass through website pkIt changes to route rjWaiting time are as follows:
Wherein q indicates route rjOn the q vehicle, O (s (i, f), ri,rj) indicate passenger in website S (i, f) by route ri
Change to route rjTraveling time.From formula (3) it can be concluded that, it is all in transfer website pkFrom route riIt changes to route rj
The total waiting time of passenger are as follows:
Wherein miIndicate route riThe index of last upper vehicle,Indicate that working as the time isWhen in transfer
Website pkFrom route riIt changes to route rjTotal passenger, e indicate route riOn the e vehicle.
According to formula (4), total passenger changes to the waiting time in entire transportation network are as follows:
Total transfer passenger's number in entire transportation network are as follows:
According to formula (5) and (6), the average passenger of entire railway network changes to the waiting time are as follows:
The above method can extend the average passenger for calculating the railway network with variable period and change to the waiting time.Wherein
Crucial problem is departure time of the determining all trains in each website.It thus can directly pass through formula (3) calculating to multiply
Visitor is from route riOn the transfer of n-th vehicle to route rjTime, and average passenger's transfer time can then be counted by formula (7)
It calculates.All trains are determined in the departure time of each website, one of them important problem is how to determine train in neighbour
Connect the departure interval of time interval.A kind of common method is using Mean Time Between Replacement method.This method utilizes adjacent time zone
Between period of dispatching a car average value as the departure interval.According to Mean Time Between Replacement method, the time of departure of the vehicle in first stop
It can be indicated by formula (8) and (9), wherein i is the arbitrary value of section [1, M]:
The headway of i-th route t moment are as follows:
WhereinI-th the predefined of route t moment is returned to dispatch a car the period.I-th route vehicle goes out in variable period
It stands the time are as follows:
All trains are calculated in the frequency of each website with formula (8) to formula (11).If giving all services
Route originates the time, and the flat of the period time-table scheduling problem with variable period is calculated according to formula (3) to formula (9)
Equal passenger's transfer time.
The flow chart of inventive algorithm optimizing cycle transit schedules scheduling problem is given in attached drawing 1.It just flows below
The content substep of journey figure describes the specific embodiment of entire algorithm:
1, initialization of population.
In the present invention, the individual of population is encoded into the real vector of D dimension:
Wherein, g is the algebra when evolution, and i is the index of individual, and D is the dimension of problem, i.e. the number of service route,Originating the moment and originating the time interval at moment earliest for j-th strip service route in the solution that i-th of individual includes is represent,
The unit of time interval is the second.Time interval is indicated with real number rather than with integer the reason of is as follows: 1) using real coding
Time interval can be more accurately described, actual application is convenient for.It can only be needed using real coding to Fitness analysis letter
Number, which carries out subtle modification, can meet the needs of different precision.Such as it needs using 0.001 second as minimum time list
Position, then the integer part of real number and the front three fractional part of the real number can be intercepted for Fitness analysis；2) it uses
Real coding can make algorithm be more easily implemented.This is because DE algorithm itself is to be designed to solve continuous optimization problems
's.In most cases, the mutation operator of DE can only generate the test vector containing real number element.In an experiment, with the second
As the smallest chronomere.Therefore, the integer part for the real vector for only being included with individual assesses individual fitness,
And fractional part is then ignored.That is, fitness evaluating function meets following property:
The purpose of initialization procedure is that respectively two sub- populations generate an initial population.Remember of the two sub- populations
Body is shown in formula (14) to (17):
WhereinWithRespectively global population PGWith local population PLI-th individual, GN and LN are respectively two
The size of population, D are the dimension of problem to be solved,WithRespectively PGAnd PLI-th of individual jth dimension variable value,WithIn section [LBj,UBj] interior random initializtion obtains, wherein LBjAnd UBjThe jth of problem respectively to be solved, which is tieed up, to be become
The lower bound of amount and the upper bound.
2, global Evolution of Population.
After having initialized two sub- populations, algorithm is arranged first using the parameter and operator for being conducive to global search come more
New PGIndividual.Specifically, improving P in terms of fourGAbility of searching optimum.It on the one hand is arranged using random fashion
Zoom factor FGWith crossover probability CRGValue.Due to optimal FGAnd CRGBe arranged it is related with specific problem to be solved, and in reality
In the application on border, the characteristic of problem to be solved is not usually known.Therefore, for the ease of the robust of actual application and raising algorithm
Property, F is set by the way of being randomly providedGAnd CRGValue, it may be assumed that
FG=rand (0,1) (18)
CRG=rand (0,1) (19)
The second aspect is using the DE/rand/1 mutation operator with stronger ability of searching optimum.Select DE/rand/1
Mutation operator is considered based on following: 1) DE/rand/1 mutation operator selects difference vector using random manner, therefore not
There can be Preference to some specific direction.This is conducive to the diversity for keeping population；2) DE/rand/1 operator is numerous
Good ability of searching optimum is shown in, is well suited for solving multiple peak problem.
It is the operation for introducing a random variation in terms of third.Work as PGIt is executed using based on above-mentioned parameter and operator setting
Complete crossover operation and then execute an additional random variation operation.Random variation operation is with the Probability p m of very little to production
Raw test vector makes a variation.The process of random variation can be described as shown in formula (20):
Wherein,Indicate global population PGI-th of individual jth dimension variable value, g indicates the algebra of iterative evolution,
LBjAnd UBjThe lower bound of the jth dimension variable of problem respectively to be solved and the upper bound, MAXEVALS and evals are respectively maximum adaptation
Degree assessment number and current Fitness analysis number,It is optimum individual to the value of dependent variable.The introducing of random variation operator
So that algorithm has an opportunity to traverse entire search space, it is ensured that algorithm may finally find the globally optimal solution of problem.
4th aspect is to use cooperative coevolution mode on multiple populations, enhances the diversity of population, accelerates algorithm
Convergence rate, and prevent algorithm from falling into suboptimization.
3, local Evolution of Population.
After algorithm is updated global population, algorithm is then set using the parameter and operator that are conducive to local search
It sets to update the individual of local population.Firstly, zoom factor FLValue be provided that
FL=rand (0,0.8) (21)
In this way, can achieve the purpose of quickening algorithm the convergence speed by reducing zoom factor.According to Ronkkonen et al.
Research think, if problem to be solved is separable problem, crossover probability CRLValue should the person of being set as close to 0, otherwise answer
When being disposed proximate in 1.Based on above-mentioned consideration, the simple adaptive control mechanism as shown in formula (22) is devised CR is arrangedL's
Value:
Wherein parameter ξ is set as 0.5 when algorithm is initial, so that algorithm be made to have identical parameter probability valuing CRL=0 and CRL=
1.During evolution, if being set as CR in parameterLIt is better than optimal of (or being equal to) local population in the case where=0
The new individual of body, then ξ will be increased to increase CRL=0 select probability.Conversely, if CRL=1 the case where, then ξ will be subtracted
It is small to increase CRL=1 select probability.Shown in specific scaling process such as formula (23):
Wherein α ∈ [0,1] is learning rate (e.g., α=0.1).
It is arranged based on above-mentioned parameter, variation vector is generated using DE/best/1 Variation mechanism.It is using the Variation mechanism
Because it will always make a variation, vector is drawn close toward the preferably solution being currently found, thus can accelerate the convergence rate of population.Using DE/
Another reason for best/1 Variation mechanism is because DE/best/1 Variation mechanism usually shows in solving many practical problems
Very good performance out.According to the two-way migration mechanism that can be described to below, if PLIn optimum individual remain unchanged, that
PLIn individual will become quickly identical.Although this accelerates convergence rate, P can be madeLPart is fallen into too early most
It is excellent.In order to overcome this problem, from PGMiddle random selection individual generates the difference vector in DE/best/1, it may be assumed that
Wherein r1, r2 ∈ [1, GN] are two mutually different integers,It is local population PLIn optimized individual, FL
For the zoom factor in local population,WithRespectively global population PGR1 and the r2 individual.
4, population cooperates.
Global population PGWith local population PLPopulation cooperation mode by comparing optimal objective vector between simultaneously Population Regeneration comes
Improve population diversity and search efficiency.Cooperation mode on multiple populations proposed by the present invention is come in recon population from both direction
Individual.The cooperation mode in the two directions is respectively with global population PGUpdate local population PLWith with local population PLIt updates
Global population PG.Remember PGOptimized individual, PLOptimized individual and PLIt is worst individual be respectively Mbest,NbestAnd Nworst, then more
New PLThe mode in stage is as follows:
And update PGThe mode in stage is as follows:
if f(Mbest) > f (Nbest),then Mbest=Nbest (26)
In this way, updating PLStage ensure that PLCurrent optimal solution is obtained in time, and accelerates convergence rate；Meanwhile updating PG
Stage then ensure that PGIt timely updates and optimal individual and keeps the diversity of population, to improve the search efficiency of algorithm.
After having executed the evolution of population cooperation, algorithm, which returns, executes global Evolution of Population, local Evolution of Population and population
Cooperative operation.It is certain so to recycle the optimal solution satisfaction for being more than defined maximum evolutionary generation until the number of iterations or obtaining at this time
Error requirements (such as error is less than 5%).
In order to test and assess the performance of algorithm of the invention, carried out by taking Subway Network in China timetable optimizing scheduling as an example
Emulation testing.For example Shenzhen Metro includes 5 main travel routes and 13 transfer websites, Guangzhou Underground network includes 16 masters
The service route wanted and 12 transfer websites.During experiment simulation, following hypothesis is done:
(a) all service routes originate the moment known to time interval；
(b) train in each service route is from originating known to website to the running time of each website on the way；
(c) train is at the time of each website stops and known to the time of passenger's transfer；
(d) known to all service routes are divided time interval and corresponding period；
(e) transfer passenger's number is known to peak period, centre forward's phase and low peak period.
The parameter setting of algorithm of the invention are as follows: GN=10, LN=10, pm=0.005, p=0.1.Final result is aobvious
Show, algorithm of the invention does not fall into suboptimization in multiple emulation testing, and average effect of optimization is better than tradition
Heuristic feasible operator, genetic algorithm and some famous evolution algorithms being recently proposed.This explanation solves week using the present invention
Phase property transit schedules Problems of Optimal Dispatch is highly effective.
In conclusion enhancing difference invention introduces a kind of parameter based on cooperation on multiple populations and operator controlling mechanism
The efficiency and robustness for dividing evolution algorithm to solve periodic traffic timetable Problems of Optimal Dispatch improve differential evolution algorithm jump
The ability of local optimum out, and differential evolution algorithm is reduced to the sensibility of parameter.With Subway Network in China and simulation railway
Emulation testing is carried out for network, it was demonstrated that the method for invention is highly effective.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (10)
1. a kind of method for leading to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations, feature exist
In the method includes:
Initialization population step originates moment range as boundary using periodic traffic route, according to apart from the lower bound moment when
Between be spaced and random generate global population PGWith local population PL, and according to valuation functions calculate population in all individuals assessed value,
Wherein, global population PGWith local population PLAccording to represented by following formula:
WhereinWithRespectively global population PGWith local population PLI-th individual, g indicate iterative evolution algebra,
GN and LN respectively indicate global population PGWith local population PLPopulation number, population at individual indicates with following vector form:
Wherein D indicates the total number of traffic route,Indicate that the j-th strip route of i-th of individual in population originates moment and lower bound
Interval time between moment；
Global Evolution of Population step, for global population PGIn each object vector, according to global population Mutation Strategy generate
One variation vector, and according to valuation functions calculate this variation vector assessed value, by the assessed value for the vector that makes a variation with it is corresponding at the beginning of
The assessed value of beginning object vector is compared, if the assessed value of variation vector is parity with or superiority over commenting for corresponding initial target vector
The variation vector is then replaced corresponding initial target vector, and enters the next generation of population by valuation；
Local Evolution of Population step, for local population PLIn each object vector, according to local population Mutation Strategy generate
One variation vector, and according to valuation functions calculate this variation vector assessed value, by the assessed value for the vector that makes a variation with it is corresponding at the beginning of
The assessed value of beginning object vector is compared, if the assessed value of variation vector is parity with or superiority over commenting for corresponding initial target vector
The variation vector is then replaced corresponding initial target vector, and enters the next generation of population by valuation；
Population cooperation step, global population PGWith local population PLBy comparing the population of optimal objective vector between simultaneously Population Regeneration
Cooperation mode improves population diversity and search efficiency, if local population PLOptimal objective vector be better than global population PG
Optimal objective vector, then by local population PLOptimal objective vector replace global population PGOptimal objective vector；Conversely,
Then by global population PGOptimal objective vector replace local population PLOptimal objective vector；Meanwhile in local population PLMost
Excellent object vector is better than global population PGIn the case where optimal objective vector, if global population PGOptimal objective vector be better than
Local population PLWorst object vector, then by global population PGOptimal objective vector replace local population PLWorst target
Vector；
Terminate judgment step, in the process of implementation if the number of iterations is more than defined maximum evolutionary generation or obtains at this time most
Excellent solution reaches defined error requirements and then terminates optimization, otherwise repeats global Evolution of Population step, local Evolution of Population step
Rapid and population cooperation step.
2. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the global Evolution of Population step,
Zoom factor F in global population is set by the way of being randomly providedGWith crossover probability CRGValue, it may be assumed that
FG=rand (0,1)
CRG=rand (0,1).
3. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the global Evolution of Population step, using DE/rand/1 mutation operator.
4. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the global Evolution of Population step, the operation of a random variation is introduced, as global population PG
Execute crossover operation and then execute random variation operation, wherein the process of random variation as shown by the following formula:
WhereinIndicate global population PGI-th of individual jth dimension variable value, g indicates the algebra of iterative evolution, LBjAnd UBj
The lower bound of the jth dimension variable of problem respectively to be solved and the upper bound, MAXEVALS and evals are respectively the assessment time of maximum adaptation degree
Several and current Fitness analysis number,The value that jth ties up variable is corresponded to for optimum individual, pm is a prior given numerical value, is made
Random variation operation makes a variation to the test vector of generation with the Probability p m of very little.
5. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the global Evolution of Population step and local Evolution of Population step, be all made of cooperation on multiple populations
Mode of evolution.
6. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the local Evolution of Population step, zoom factor FLValue be provided that
FL=rand (0,0.8).
7. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the local Evolution of Population step, crossover probability CRLValue be provided that
Wherein, parameter ξ is set as 0.5 at the beginning.
8. according to claim 7 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that during evolution, if in parameter setting CRLIt is better than in the case where=0 or equal to local kind
The new individual of group's optimum individual, then ξ will be increased to increase CRL=0 select probability, conversely, if in parameter setting CRL=1
In the case where, then ξ will decrease to increase CRL=1 select probability, the scaling process of parameter ξ as shown by the following formula:
Wherein, α ∈ [0,1] is learning rate.
9. according to claim 1 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that in the local Evolution of Population step, variation vector is generated using DE/best/1 Variation mechanism.
10. according to claim 9 lead to scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations
Method, which is characterized in that from global population PGMiddle random selection individual generates the difference vector in DE/best/1, it may be assumed that
Wherein, r1, r2 ∈ [1, GN] are two mutually different integers,It is local population PLIn optimized individual, FLFor office
Zoom factor in portion population,WithRespectively global population PGR1 and the r2 individual.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811200659.6A CN109492797A (en) | 2018-10-16 | 2018-10-16 | Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811200659.6A CN109492797A (en) | 2018-10-16 | 2018-10-16 | Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109492797A true CN109492797A (en) | 2019-03-19 |
Family
ID=65689528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811200659.6A Pending CN109492797A (en) | 2018-10-16 | 2018-10-16 | Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109492797A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110134007A (en) * | 2019-05-22 | 2019-08-16 | 南昌航空大学 | Multiple no-manned plane cooperates with target assignment method |
| CN110986960A (en) * | 2019-12-31 | 2020-04-10 | 哈尔滨工业大学 | Unmanned aerial vehicle track planning method based on improved clustering algorithm |
| CN113794638A (en) * | 2021-08-24 | 2021-12-14 | 内蒙古农业大学 | SDN data center network elephant flow scheduling method based on differential evolution algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103246937A (en) * | 2013-04-25 | 2013-08-14 | 中山大学 | Dual population differential evolution algorithm-based optimization method for periodic train schedule dispatching |
| CN104239484A (en) * | 2014-09-05 | 2014-12-24 | 浙江工业大学 | Multi-mode bus combinatorial dispatch-based schedule making method |
| CN106875031A (en) * | 2016-12-21 | 2017-06-20 | 深圳大学 | A kind of many Job Shop Scheduling method and devices |
-
2018
- 2018-10-16 CN CN201811200659.6A patent/CN109492797A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103246937A (en) * | 2013-04-25 | 2013-08-14 | 中山大学 | Dual population differential evolution algorithm-based optimization method for periodic train schedule dispatching |
| CN104239484A (en) * | 2014-09-05 | 2014-12-24 | 浙江工业大学 | Multi-mode bus combinatorial dispatch-based schedule making method |
| CN106875031A (en) * | 2016-12-21 | 2017-06-20 | 深圳大学 | A kind of many Job Shop Scheduling method and devices |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110134007A (en) * | 2019-05-22 | 2019-08-16 | 南昌航空大学 | Multiple no-manned plane cooperates with target assignment method |
| CN110986960A (en) * | 2019-12-31 | 2020-04-10 | 哈尔滨工业大学 | Unmanned aerial vehicle track planning method based on improved clustering algorithm |
| CN110986960B (en) * | 2019-12-31 | 2022-10-28 | 哈尔滨工业大学 | Unmanned aerial vehicle flight path planning method based on improved clustering algorithm |
| CN113794638A (en) * | 2021-08-24 | 2021-12-14 | 内蒙古农业大学 | SDN data center network elephant flow scheduling method based on differential evolution algorithm |
| CN113794638B (en) * | 2021-08-24 | 2022-10-14 | 内蒙古农业大学 | SDN data center network elephant flow scheduling method based on differential evolution algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021248607A1 (en) | Deep reinforcement learning-based taxi dispatching method and system | |
| Noaeen et al. | Reinforcement learning in urban network traffic signal control: A systematic literature review | |
| El-Tantawy et al. | Multiagent reinforcement learning for integrated network of adaptive traffic signal controllers (MARLIN-ATSC): methodology and large-scale application on downtown Toronto | |
| Zhu et al. | A bi-level model for single-line rail timetable design with consideration of demand and capacity | |
| CN109492797A (en) | Lead to the method for scheduling instance table with cooperation differential evolution algorithm optimizing cycle sexual intercourse on multiple populations | |
| Eaton et al. | Ant colony optimization for simulated dynamic multi-objective railway junction rescheduling | |
| Hernández et al. | Analysis of real-time control strategies in a corridor with multiple bus services | |
| Yang et al. | A bi-objective timetable optimization model incorporating energy allocation and passenger assignment in an energy-regenerative metro system | |
| Carvajal-Carreño et al. | Optimal design of energy-efficient ATO CBTC driving for metro lines based on NSGA-II with fuzzy parameters | |
| CN109508751B (en) | Deep neural network model modeling method for high-speed railway train late time prediction | |
| CN103246937A (en) | Dual population differential evolution algorithm-based optimization method for periodic train schedule dispatching | |
| Han et al. | Train timetabling in rail transit network under uncertain and dynamic demand using advanced and adaptive NSGA-II | |
| Tang et al. | Scheduling local and express trains in suburban rail transit lines: Mixed–integer nonlinear programming and adaptive genetic algorithm | |
| CN103489042A (en) | Method for optimizing disaster emergency decision system path | |
| CN106652434A (en) | Bus dispatching method based on rail transit coordination | |
| Shui et al. | A clonal selection algorithm for urban bus vehicle scheduling | |
| Hu | Multi-objective genetic algorithm for berth allocation problem considering daytime preference | |
| Chiou et al. | Sustainable consumption, production and infrastructure construction for operating and planning intercity passenger transport systems | |
| Núñez et al. | Multiobjective model predictive control for dynamic pickup and delivery problems | |
| Jiang et al. | A distributed multi-agent reinforcement learning with graph decomposition approach for large-scale adaptive traffic signal control | |
| Yang et al. | Optimized two-stage fuzzy control for urban traffic signals at isolated intersection and Paramics simulation | |
| CN110211379A (en) | A kind of public transport method for optimizing scheduling based on machine learning | |
| Ma et al. | A decentralized model predictive traffic signal control method with fixed phase sequence for urban networks | |
| Korecki et al. | Analytically guided reinforcement learning for green it and fluent traffic | |
| Silva-Soto et al. | Timetabling with flexible frequencies to synchronise groups of bus lines at common stops |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190319 |
|
| RJ01 | Rejection of invention patent application after publication |