CN102411306B - Mixed flow assembly production scheduling control method based on bee society self-organization model - Google Patents
Mixed flow assembly production scheduling control method based on bee society self-organization model Download PDFInfo
- Publication number
- CN102411306B CN102411306B CN201110366396.8A CN201110366396A CN102411306B CN 102411306 B CN102411306 B CN 102411306B CN 201110366396 A CN201110366396 A CN 201110366396A CN 102411306 B CN102411306 B CN 102411306B
- Authority
- CN
- China
- Prior art keywords
- honeybee
- algorithm
- bee
- workpiece
- sequence
- 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.)
- Active
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a mixed flow assembly production scheduling control method based on a bee society self-organization model. The method comprises the following steps of: 1, setting algorithm parameters according to scale of mixed flow assembly production scheduling problems; 2, selecting a random +NEH method for an initialization function aiming at characteristics of the mixed flow assembly production scheduling problems; 3, dynamically adjusting a searching scale in neighboring areas; 4, performing an employment bee algorithm on each employment bee, and triggering a scout bee algorithm to replace the employment bee when the employment bee meets a limit parameter; 5, performing an observation bee algorithm on each observation bee, and triggering the scout bee algorithm to replace the observation bee when the observation bee meets the limit parameter; 6, performing an optimal control algorithm on the employment bee population and observation bee population; and 7, stopping the operation when iteration conditions are met, otherwise repeating the steps from 3 to 7. The method has high speed, high neighboring searching precision, high solution efficiency and good control effect, and local optimum is effectively avoided.
Description
Technical field
The present invention relates to a kind of Job-Shop method based on honeybee social self-organization model.
Background technology
Production scheduling is Ordering production run, the key issue that reasonable arrangement is produced.By production scheduling, by the operation resource, the logistic resources reasonable distribution that relate in production run, improve enterprises production efficiency and fluency, and reduce produce in production run time, space waste.To low margin age, the development of enterprise is significant.
It is a kind of production model that the discrete product of the technological processs such as current automobile industry, household electric appliances generally adopts that mixed flow process and assemble is produced, and is also developing production model.In many manufacturing enterprises based on assembling manufacturing, due to a variety of causes, enterprise not only will carry out product final assemble, also will carry out processing and the assembling of parts, meet general assembly line need of production.The production of preparation and assembly steelshop mainly comprises operation type production and assembly type produces two classes, and corresponding scheduling problem is also divided into operation job shop and flowing water making-up shop to dispatch two classes, covers the main process that process and assemble is produced.Wherein assembly line production is the important component part of mixed flow process and assemble.The parts that external coordination is produced or purchased by assembly line and the parts that processing line is produced are assembled into the final products meeting customer demand here by continuous print productive unit, as shown in Figure 1.Modern Assembling Production process, have employed the organizational form of mixes line production more under multi-varieties and small-batch is Influence of production.Mixes line production can improve the flexibility of organization of production, fast responding market and customer demand, reduces the input such as production line and mould, processing cost, transportation cost, effectively improves production capacity.But mixes line production, while raising system produces flexible and production efficiency, increases the difficulty of Technical innova-tion, the risk improve off-production, overstocking at goods.Therefore, effective mixed-model assembly production scheduling becomes the focus of modern management research.
Solving on complicated combinatorial optimization problem, Swarm Intelligent Algorithm (Population-basedIntelligent Optimization, PIO) is current study hotspot.It is characterized in that with the feasible solution representing problem individual in group, and according to ideal adaptation degree, progressively obtain satisfactory solution by the interactive optimizing between colony.It is evolution algorithm (the Evaluation Algorithm of representative that PIO is mainly divided into genetic algorithm (GA), evolutional programming (EP), evolution strategy (ES) etc., EA) and with ant group algorithm (ACO), Particle Swarm Optimization (PSO) etc. be the swarm intelligence algorithm (Swarm Intelligence, SI) of representative.Wherein evolution algorithm proposes the time early, and application is comparatively extensive, and particularly genetic algorithm, the combinatorial optimization problem such as the production schedule, scheduling is widely used.Swarm intelligence algorithm belongs to artificial intelligence category, is proposed the earliest by Beni, Hackwood and Wang in 1989, for describing the cooperating process of discrete natural or artificial self-organizing system.In 1999, Bonbeau, Dorigo and Theraulaz have expanded this concept: colony intelligence is not only for describing the cooperative behaviors between colony, and any cooperative behaviors of trooping by insect populations or other animals excites and the algorithm design of carrying out or distributed Resolving probiems Strategy Design process all belong to colony intelligence.
Swarm intelligence algorithm based on honeybee social self-organization model is the emerging optimized algorithm of a class, the social action model that algorithm utilizes bee colony to look for food, bee colony foraging behavior is mapped in combinatorial optimization problem, by bee colony self-organization (Stigimergy) coordination process, solving of problem of implementation.Seeley T.D. proposes the group behavior model of bee colony the earliest in nineteen ninety-five.Theraulaz G etc. to be looked for food working model by the offspring that internal-response threshold value and environmental stimulus signal carry out according to honeybee subsequently.Calendar year 2001
p and
d is in biological study and bee colony behavior model Research foundation, and first Application honeybee cooperative behaviors solves combinatorial optimization problem, and proposes bee colony optimization (Bee Colony Optimization, BCO).After this, a series of optimized algorithm based on bee colony model proposes in succession, studies the more artificial bee colony algorithm (ArtificialBee Colony, ABC) having Karaboga D to propose in 2005 for solving continuous optimization problems; Pham D.T. solves complicated continuous optimization problems in the honeybee algorithm (Bee Algorithm, BA) that 2006 propose.In addition, the virtual honeybee algorithm (Virtual Bee Algorithm, VBA) etc. proposed around honeycomb algorithm (BeeHive), Drias etc. that foraging areas concept proposes based on bee colony optimization (Bee Swarm Optimization) algorithm, Yang etc. that satisfiability problem proposes based on the MBO model, Wedde etc. of honeybee marriage principle also having Abbas to propose.
Solve job shop scheduling problems research based on honeybee social model now mainly contains, solving job shop scheduling problem aspect, Li Duanming etc. adopt ABC to solve different size workpiece unit (solve different size workpiece unit based on artificial bee colony algorithm and criticize scheduling problem, 2009) and criticize scheduling problem; Pham D.T. etc. adopt BA to solve single machine scheduling in (The Bees Algorithm, ANovel Toolfor Complex Optimisation Problems, 2006); Chin SoonChong etc. are at (Using a bee colony algorithm for neighborhood search in job shopscheduling problems, 2007 and A bee colony optimization algorithm to job shopscheduling, 2006) in, BCO is combined with other heuristic rules and has solved job-shop scheduling problem; Wong Li-Pei etc. will improve BCO application and the job-shop scheduling problem of neighborhood search in (Bee colony optimisation algorithm with big valley landscapeexploitation for job shop scheduling problems, 2010); Li Jun-qing etc. have employed HABC Algorithm for Solving Flexible Job-shop Scheduling Problems in (A Hybrid Artificial Bee ColonyAlgorithmfor Flexible Job Shop Scheduling Problems, 2011).
Flow Shop scheduling aspect, there is Pan Quan-ke etc. in (A discrete artificial bee colonyalgorithm for the lot-streaming flow shop scheduling problem, 2011), adopt ABC Algorithm for Solving batch fluvial incision.
Above-mentioned research generally has solved problem and solves effect preferably, demonstrates honeybee social model algorithm and compare the feature that other swarm intelligence algorithms are quick, ability of searching optimum is high in scheduling problem application.But optimization method spininess is to traditional scheduler problem, and BCO method is slow relative to ABC algorithm speed, easily be absorbed in local optimum, and also there is the shortcomings such as neighborhood search ability in ABC algorithm, single algorithm is difficult to solve complicated actual schedule problem, and it is not high to there is search precision, to the inefficient problem of extensive Scheduling Problem, and there is not systematic dispatching method in assembly line scheduling yet.
Summary of the invention
In order to the speed overcoming existing mixed-model assembly production scheduling control method is slow, easily be absorbed in the deficiency that local optimum, neighborhood search precision are not high, solution efficiency is lower, control effects is poor, the invention provides a kind of speed, effectively avoid being absorbed in local optimum, neighborhood search precision is high, solution efficiency is high, the mixed-model assembly production scheduling control method based on honeybee social self-organization model that control effects is good.
The technical solution adopted for the present invention to solve the technical problems is:
Based on a mixed-model assembly production scheduling control method for honeybee social self-organization model, described mixed-model assembly production scheduling control method comprises the following steps:
Step 1, parameter initialization: according to mixed-model assembly production scheduling problems scale set algorithm parameter, described parameter comprises population scale SN, overall controling parameters limit, optimal control parameter bestlimit and initial neighborhood parameter S;
Step 2, initialization of population: for the problem characteristic of mixed-model assembly line scheduling, initialization function selects BCO method initially to employ bee colony, random+NEH method initial inspection bee colony;
Step 3, neighborhood scale adjusts: according to initial scale and algorithm process, the search scale of dynamic conditioning neighborhood;
Step 4, employ honeybee algorithm: travel through all employ bee colony and employ honeybee to perform to each employ honeybee algorithm, when employing honeybee to meet limit parameter, triggering search bee algorithm and replacing and employ honeybee;
Step 5, observes honeybee algorithm: travel through all observation bee colonies and observe honeybee execution to each and observe honeybee algorithm, when observing honeybee and meeting limit parameter, trigger search bee algorithm and replace observation honeybee;
Step 6, optimal control algorithm, performs optimal control algorithm, the individuality repeated in displacement population to employing honeybee population and observing bee colony,
Wherein, the optimum control process of employing honeybee population and observing honeybee population is:
Step 6.1.1, calculates the quantity bestnum of optimum solution in population;
Step 6.1.2, if bestnum > bestlimit, then i=i+1 continuing, otherwise algorithm terminates;
Step 6.1.3, Stochastic choice two optimum solution bee1 and bee2, if bee1 and bee2 coding is not identical, continues step 6.1.4, otherwise skip to step 6.1.6;
Step 6.1.4, beecross operation is carried out to bee1 and bee2, four Index variablees are the random position produced, and bee1 and bee2 is exchanged by the subsequence of generation random in sequence and the other side and removed repeating part and obtains two new explanation finalbee1 and finalbee2;
Step 6.1.5, judges, if bee1 and bee2 is still optimum, then continues, otherwise skips to step 6.1.7;
Step 6.1.6, carries out search bee algorithm respectively to bee1 and bee2, namely adopts the initial method of random search to produce a RANDOM SOLUTION, and substitutes original individuality;
Step 6.1.7, if i is < (bestnum-bestlimit)/2, skips to step 6.1.3
Step 6.1.8, does simulated annealing operation to all optimum solutions of residue;
Step 7, judges whether to meet maxcycle stopping criterion for iteration, satisfied then stop, otherwise repeats step 3 ~ step 7.
Further, in described step 2, set up based on the random mixed method generation initial population of BCO and NEH+ respectively for employing bee colony and observing bee colony,
(2.1) employ honeybee initialization step as follows:
Step 2.1.1, step=step+1, and get one according to the order of sequence and employ honeybee, concentrate from unselected workpiece and select workpiece and renewal sequence coding;
Step 2.1.2, judges whether that traversal is all and employs honeybee, do not complete and then repeat step 2.1.1, otherwise continue;
Step 2.1.3, again get one according to the order of sequence in honeybee and employ honeybee employing, judge whether to meet probability, probability is obtained by formula 1, satisfied then continue step 2.1.4, otherwise skips to step 2.1.5;
Step 2.1.4, employs roulette selection one honeybee to separate from remainder and substitutes current solution;
Step 2.1.5, judges whether that traversal allly employs honeybee, satisfied then continuation step 2.1.6, otherwise jumps to step 2.1.3;
Step 2.1.6, judges whether step equals process number, satisfied then terminate, otherwise jumps to step 2.1.1;
Step 2.1.7, for employing bee colony initialization array limitarray1 (), array length SN/2, records and eachly employs honeybee not upgrade algebraically;
Wherein step is from 1 to SN/2 value, currently employs honeybee drop probability pd
caccording to formula 1 value, wherein fit
bestfor the best fitness of current population, fit
choneybee fitness is employed for current.
(2.2) observing the initialization of honeybee NEH+ random device: first all workpiece press total elapsed time sort descending, selecting the first two workpiece through comparing reservation deadline the shortest sequence; Secondly, by each position in the 3rd workpiece insetion sequence, retain deadline most short data records equally; Finally, residue workpiece is repeated work of drilling, obtain a feasible solution, this feasible solution be inserted into and employ in bee colony, other individualities of bee colony then adopt random device to generate, and concrete figure process is:
Step 2.2.1 calculates the total elapsed time of all workpiece, and temporally sort descending;
Step 2.2.2 selects the unit one coming sequence top to form coding, and Solution (i), wherein i=1 presentation code length, reject the workpiece of selection from workpiece sequence;
Step 2.2.3 selects the workpiece coming top from workpiece sequence, workpiece is inserted each position in Solution (i) and form i+1 solution, retain according to fitness and best be designated as solution (i+1), this workpiece is rejected from workpiece sequence;
Step 2.2.4 judges whether workpiece sequence is empty, if yes then enter step 2.2.5, otherwise jumps to step 2.2.4;
Step 2.2.5 stochastic generation (SN/2-1) individual solution;
Step 2.2.5 is initialization of population array limitarray2 (), array length SN/2, and record is observed honeybee individuality and do not upgraded algebraically.
Further again, in described step 4, described in employ the tool process of honeybee algorithm as follows:
Step 4.1, honeybee EB1 is employed in select progressively one, and judges whether it is optimum solution in group, satisfied then skip to step 4.3, otherwise continues;
Step 4.2, carries out neighborhood search based on EB1, obtains the individual new individuality of S ', and selects wherein optimum as EB2, skips to step 4.9;
Step 4.3, carries out simulated annealing operation to EB1, sets SA initial temperature, temperature and moves back warm coefficient;
Step 4.4, carries out neighborhood search to based on EB1 again, obtains the individual new individuality of S ', and selects wherein optimum as EB2;
Step 4.5, compares EB1 and EB2, if EB2 is inferior to EB1, continues step 4.6, otherwise skips to step 4.7
Step 4.6, compares generation (0 ~ 1) random number and compares with probability P, and P obeys formula 3, if random number is less than probability, continues step 7, otherwise skips to step 8;
P=exp (-(fit (EB2)-fit (EB1))) formula 3
Step 4.7, EB1=EB2;
Step 4.8, upgrades T and S ' if T < is Tend according to formula 4, formula 5, then continue step 4.9, otherwise repeats step 4.4-4.8;
T=k × Tend formula 4
S '=ceil (k × S ') formula 5
Step 4.9, greedy rule compares EB1 and EB2, if EB2 is not better than EB1, then limitarray1 (i)=limitarray1 (i)+1, i represents EB1 sequence number, otherwise skips to step 4.11;
Step 4.10, if limitarray1 (i)=limit, calls search bee algorithm and replaces EB1, and skip to step 4.12, otherwise jump directly to step 4.12;
Step 4.11, EB1=EB2, and limitarray1 (i)=0;
Whether step 4.12, travel through and allly employ honeybee, if i.e. i=SN/2, then algorithm terminates, otherwise skips to step 4.1.
Further, in described step 5, the tool process of described observation honeybee algorithm is as follows:
Step 5.1, honeybee OB1 is observed in select progressively one, employs honeybee EB1 according to fitness roulette selection;
Step 5.2, greedy rule compares OB1 and EB1, if EB1 is not better than OB1, then limitarray2 (j)=limitarray2 (j)+1, j represents OB1 sequence number, otherwise skips to step 5.4;
Step 5.3, if limitarray2 (j)=limit, calls search bee algorithm and replaces OB1, and skip to step 5, otherwise jump directly to step 5.5;
Step 5.4, OB1=EB1, and limitarray2 (i)=0;
Whether step 5.5, travel through all observation honeybees, if i.e. j=SN/2, then algorithm terminates, otherwise skips to step 5.1.
Technical conceive of the present invention is: the bee colony of occurring in nature, after searching food source, can get back in honeycomb and transmit food information by swing to other workers bee, and leads more multiplex (MUX) honeybee to get back to food source collection food.Karaboga D establishes the minimum model of bee colony foraging behavior according to this social behavior of honeybee, and proposes artificial bee colony algorithm on this basis.Model comprises food source, employs honeybee, observes honeybee and search bee four assemblies.Be mapped in optimization problem, food source represents the feasible solution of problem, the quality of the degree in plenty of food source then homographic solution.Bee colony is divided into employs honeybee and observes honeybee, and in the specific period, the some individuals in two kinds of bee colonies can change search bee into or transform mutually.At initial period, all honeybees are search bee, random search food source (feasible solution) near honeycomb.After getting back to honeycomb, each search bee becomes according to the quality choice holding feasible solution separately employs honeybee or observes honeybee.Employ honeybee to get back to former food source and continue search, observe honeybee then stay-at-home honeycomb.After employing honeybee again to get back to honeycomb, observe honeybee accepts to employ honeybee recruitment according to certain probability, select to follow to employ honeybee or replacement to employ honeybee.If employ honeybee after certain number of times is searched, fail to find better food source, then abandoning current foodstuff source becomes search bee, the food source that random search is new.Fig. 2 gives the groundwork process of bee colony.In Fig. 2, search bee SB1, SB2 get back to honeycomb and change into respectively and employ honeybee EB1 and observe honeybee UB1 after searching food source.Wherein, EB1 have found neighborhood food source B according to food source A, have selected B more afterwards, and gets back to honeycomb in dancing district and UB1 interchange information, UB1 has visited the neighborhood food source C of B, and compare the degree in plenty of B, C, if B is more excellent, EB1 can continue search according to B, and then find D, otherwise EB1 follows UB1, thus find E; EB2, repeatedly searching through early stage, after searching G, does not recruit to and observes honeybee, continues to get back to origin-location search; EB3 goes through in the search at I place and does not repeatedly improve, and abandons I and is converted into search bee SB3.
Beneficial effect of the present invention is mainly manifested in: adopt basic artificial bee colony algorithm as algorithm main framework, the Population Initialization algorithm established based on BCO and NEH improves algorithm initial population quality, establish the Neighborhood-region-search algorithm becoming neighborhood and improve algorithm search precision, establish optimal control algorithm and improve population diversity, honeybee algorithm is employed based on Simulated Anneal Algorithm Optimize, simplify and optimize and observe honeybee algorithm, add solution efficiency and the accuracy of algorithm, a kind of systematic optimization method is provided to mixed-model assembly Workshop Production.
Accompanying drawing explanation
Fig. 1 is the schematic diagram of assembly line production run.
Fig. 2 is bee colony working model figure.
Fig. 3 improves artificial bee colony algorithm overview flow chart.
Fig. 4 is BCO method flow diagram.
Fig. 5 employs honeybee algorithm flow chart.
Fig. 6 is optimal control algorithm process flow diagram.
Fig. 7 is the schematic diagram of Beecross process.
Embodiment
Below in conjunction with accompanying drawing, the invention will be further described.
With reference to Fig. 1 ~ Fig. 7, a kind of mixed-model assembly production scheduling control method based on honeybee social self-organization model, HBCA algorithm is divided into initialization algorithm (NEH initial algorithm, BCO initial algorithm) on the whole, employs honeybee algorithm, observes honeybee algorithm, search bee algorithm and optimal control algorithm five subalgorithms, comprise 5 parameters, population scale SN respectively, iterations maxcycle, the global regulation parameter limit, optimal control parameter bestlimit, initial neighborhood scale S, with reference to Fig. 3, its technical step is generally as follows:
Step 1, parameter initialization, according to problem scale set algorithm parameter.Algorithm major parameter is population scale SN, overall controling parameters limit, optimal control parameter bestlimit and initial neighborhood parameter S.By the example experimental verification of standard benchmark storehouse, more excellent for small-scale problem population number desirable 50-100, limit value 6; For in extensive problem, desirable 100, the limit values 8 of population scale are more excellent.The equal value of Bestlimit is that pop/6 is more excellent, and the equal value n/2 of initial neighborhood S is more excellent.
Step 2, initialization of population, for the problem characteristic of mixed-model assembly line scheduling, initialization function selects random+NEH method.Concrete grammar is shown in detailed design in 5.3.2.
Step 3, neighborhood scale adjusts, according to initial scale and algorithm process, the search scale of dynamic conditioning neighborhood;
Step 4, employs honeybee algorithm, travel through all employ bee colony and employ honeybee to perform to each employ honeybee algorithm, when employing honeybee to meet limit parameter, triggering search bee algorithm and replacing and employ honeybee, employ honeybee and search bee concrete steps to see explanation in 5.3.3 and 5.3.6.
Step 5, observes honeybee algorithm, travels through all observation bee colonies and observe honeybee execution to each to observe honeybee algorithm, when observing honeybee and meeting limit parameter, triggers search bee algorithm and replaces observation honeybee.
Step 6, optimal control algorithm, perform optimal control algorithm to employing honeybee population and observing bee colony, the individuality repeated in displacement population, improves population diversity.
Step 7, judges iterated conditional, judges whether to meet maxcycle stopping criterion for iteration, satisfied then stop, otherwise repeats step 3-7.
Population encoding and decoding: for mixed flow preparation and assembly steelshop, MBCA adopts the serial code mode of job sequence, and the job sequence as 7 products is, and { 5,7,3,2,4,1,6}, product assembles by number successively.Decode phase, can obtain putting in order of product operation according to numerical order corresponding in sequence.
Initialization of population process: in the initialization of population of mixed-model assembly scheduling, sets up respectively based on the random mixed method generation initial population of BCO and NEH+ for employing bee colony and observing bee colony, improves initial population quality and diversity.
(2.1) wherein employ honeybee initialization step as follows:
Step 2.1.1, step=step+1, and get one according to the order of sequence and employ honeybee, concentrate from unselected workpiece and select workpiece and renewal sequence coding;
Step 2.1.2, judges whether that traversal is all and employs honeybee, do not complete and then repeat step 2.1.1, otherwise continue;
Step 2.1.3, again get one according to the order of sequence in honeybee and employ honeybee employing, judge whether to meet probability, probability is obtained by formula 1, satisfied then continue step 2.1.4, otherwise skips to step 2.1.5;
Step 2.1.4, employs roulette selection one honeybee to separate from remainder and substitutes current solution;
Step 2.1.5, judges whether that traversal allly employs honeybee, satisfied then continuation step 2.1.6, otherwise jumps to step 2.1.3;
Step 2.1.6, judges whether step equals process number, satisfied then terminate, otherwise jumps to step 2.1.1;
Step 2.1.7, for employing bee colony initialization array limitarray1 (), array length SN/2, records and eachly employs honeybee not upgrade algebraically;
Fig. 5 describes BCO method flow diagram, and wherein step is from 1 to SN/2 value, currently employs honeybee drop probability pd
caccording to formula 1 value, wherein fit
bestfor the best fitness of current population, fit
choneybee fitness is employed for current.Its concrete steps are:
(2.2) honeybee NEH+ random device initialization step is observed as follows:
In NEH method, first all workpiece being pressed total elapsed time sort descending, selecting the first two workpiece to retain deadline the shortest sequence through comparing; Secondly, by each position in the 3rd workpiece insetion sequence, retain deadline most short data records equally; Finally, residue workpiece is repeated work of drilling, obtain a feasible solution, this feasible solution be inserted into and employ in bee colony, other individualities of bee colony then adopt random device to generate.
Step 2.2.1 calculates the total elapsed time of all workpiece, and temporally sort descending;
Step 2.2.2 selects the unit one coming sequence top to form coding, and Solution (i), wherein i=1 presentation code length, reject the workpiece of selection from workpiece sequence;
Step 2.2.3 selects the workpiece coming top from workpiece sequence, workpiece is inserted each position in Solution (i) and form i+1 solution, retain according to fitness and best be designated as solution (i+1), this workpiece is rejected from workpiece sequence;
Step 2.2.4 judges whether workpiece sequence is empty, if yes then enter step 2.2.5, otherwise jumps to step 2.2.4;
Step 2.2.5 stochastic generation (SN/2-1) individual solution;
Step 2.2.5 is that initialization of population array limitarray2 (0, array length SN/2, observe honeybee individuality and do not upgrade algebraically by record.
Honeybee algorithmic procedure is employed to be: to employ honeybee to assume responsibility for the search mission of outstanding solution.In MBCA, establish a kind of variable neighborhood search algorithm with iterative process change region of search scope, employ honeybee to carry out search and Population Regeneration at given interior scope neighborhood.In the setting of region of search, first set a hunting zone S, then the contiguous range that current iteration number of times is corresponding obeys formula 2:
S '=ceil (S × (1-cycle/ (literation+1))) formula 2
Literation represents iterations, and cycle represents current iteration algebraically, and ceil represents and rounds downwards content in bracket.For ensureing effective contiguous range in iteration latter stage, setting S ' is not less than 5.Be different from the fixing search scope of determining neighborhood search, the neighborhood scale becoming neighborhood search constantly diminishes with iterative process on the basis of initial value S, and namely more to the later stage of algorithm implementation, hunting zone is narrower.Under the support becoming neighborhood, the algorithm initial stage can improve the global optimizing ability of algorithm, and the algorithm later stage can accelerate convergence of algorithm speed, improves the execution efficiency of algorithm on the whole.Neighborhood solution production method in neighborhood search adopts cross and variation strategy (Swap), and for mixed-model assembly problem, in random selecting coded sequence, two positions also exchange corresponding encoded radio and obtain new individuality.Employ the idiographic flow of honeybee algorithm as Fig. 6, step is described below:
Step 4.1, honeybee EB1 is employed in select progressively one, and judges whether it is optimum solution in group, satisfied then skip to step 4.3, otherwise continues;
Step 4.2, carries out neighborhood search based on EB1, obtains the individual new individuality of S ', and selects wherein optimum as EB2, skips to step 4.9;
Step 4.3, carries out simulated annealing operation to EB1, sets SA initial temperature, temperature and moves back warm coefficient;
Step 4.4, carries out neighborhood search to based on EB1 again, obtains the individual new individuality of S ', and selects wherein optimum as EB2;
Step 4.5, compares EB1 and EB2, if EB2 is inferior to EB1, continues step 4.6, otherwise skips to step 4.7
Step 4.6, compares generation (0 ~ 1) random number and compares with probability P, and P obeys formula 3, if random number is less than probability, continues step 7, otherwise skips to step 8;
P=exp (-(fit (EB2)-fit (EB1))) formula 3
Step 4.7, EB1=EB2;
Step 4.8, upgrades T and S ' if T < is Tend according to formula 4, formula 5, then continue step 4.9, otherwise repeats step 4.4-4.8;
T=k × Tend formula 4
S '=ceil (k × S ') formula 5
Step 4.9, greedy rule compares EB1 and EB2, if EB2 is not better than EB1, then limitarray1 (i)=limitarray1 (i)+1, i represents EB1 sequence number, otherwise skips to step 4.11;
Step 4.10, if limitarray1 (i)=limit, calls search bee algorithm and replaces EB1, and skip to step 4.12, otherwise jump directly to step 4.12;
Step 4.11, EB1=EB2, and limitarray1 (i)=0;
Whether step 4.12, travel through and allly employ honeybee, if i.e. i=SN/2, then algorithm terminates, otherwise skips to step 4.1.
The process of observing honeybee algorithm is: MBCA employs in honeybee algorithm, because change neighborhood search has added employ bee colony search range, simulated annealing adds the search depth employing the outstanding solution of bee colony, effectively improves the optimizing ability of algorithm.Therefore, observe honeybee choose employ honeybee after no longer carry out binary search, only upgrade according to greedy rule, keep the excellent individual of population; Meanwhile, in observation bee colony, introduce array limitarray2, record each observation bee colony and do not upgrade algebraically continuously.When observation honeybee corresponding numerical value in limitarray2 meets limit, trigger search bee algorithm and replace this observation honeybee.Its concrete steps are as follows:
Step 5.1, honeybee OB1 is observed in select progressively one, employs honeybee EB1 according to fitness roulette selection;
Step 5.2, greedy rule compares OB1 and EB1, if EB1 is not better than OB1, then limitarray2 (j)=limitarray2 (j)+1, j represents OB1 sequence number, otherwise skips to step 5.4;
Step 5.3, if limitarray2 (j)=limit, calls search bee algorithm and replaces OB1, and skip to step 5, otherwise jump directly to step 5.5;
Step 5.4, OB1=EB1, and limitarray2 (i)=0;
Whether step 5.5, travel through all observation honeybees, if i.e. j=SN/2, then algorithm terminates, otherwise skips to step 5.1.
The process of optimal control algorithm is: groupy phase is seemingly spent conference and caused precocious phenomenon, although introduce limit in algorithm to control population at individual, but after two population cooperate optimization, population optimum solution expansion rate is fast, limits the ability of searching optimum of algorithm.Therefore, optimal control algorithm is devised for reducing the multiplicity of optimum solution in two populations.To employ bee colony algorithm flow as Fig. 7, detailed process is:
Step 6.1.1 calculates the quantity bestnum of optimum solution in population;
If step 6.1.2 bestnum > bestlimit, then i=i+1 continuing, otherwise algorithm terminates;
Step 6.1.3 Stochastic choice two optimum solution bee1 and bee2, if bee1 and bee2 coding is not identical, continues step 6.1.4, otherwise skip to step 6.1.6;
Step 6.1.4 carries out beecross operation to bee1 and bee2, four Index variablees are the random position produced, and bee1 and bee2 is exchanged by the subsequence of generation random in sequence and the other side and removed repeating part and obtains two new explanation finalbee1 and finalbee2;
Step 6.1.5 judges, if bee1 and bee2 is still optimum, then continues, otherwise skips to step 6.1.7;
Step 6.1.6 carries out search bee algorithm respectively to bee1 and bee2;
If step 6.1.7 i < (bestnum-bestlimit)/2, skips to step 6.1.3
Step 6.1.8 does simulated annealing operation to all optimum solutions of residue;
Observe honeybee population optimum control process to adopt and the operation steps of employing honeybee same.
After optimal control algorithm operation, the optimum individual of colony must control within bestlimit.For improving the search capability of outstanding bee colony, adopt Simulated Anneal Algorithm Optimize to the residue individuality not being selected into interlace operation, be balance search efficiency simultaneously, the method that in simulated annealing, neighborhood generation rule adopts single to intersect herein.Through optimal control algorithm, control optimum solution quantity in population, under the prerequisite of optimum solution is not lost in guarantee, add population diversity, improve the global search performance of algorithm.
The process of search bee algorithm is: search bee algorithm scout () adopts the initial method of random search to produce a RANDOM SOLUTION, and substitutes original individuality.
Example: certain motor making-up shop general assembly line C adopts mixed production system, this is traditional thread binding joins 5 kinds of products, and containing 10 workstations, daily planning output is respectively A product 120, B product 140, C product 40, D product 60, E product 80, and switching time is 5.Minimum production circulation can be obtained for 6A, 7B, 2C, 3D, 4E, totally 22 workpiece.Each product corresponding station process time is as table 1.
Table 1 work pieces process timetable
Parameter population scale 50, iterations 200, limit is got 6, bestlimit and is got pop/6, initial neighborhood scale S=11.More whether embodiment for target, comprises the problem considering switching time with the shortest completion date respectively.Wherein comprising one of gained minimum production cyclic sequence switching time is { AAAABEBBCDEBBDDCEBBEAA}, only consider that one of minimum production cyclic sequence process time is that { AAAABBEBCEBDBDDEBCBEAA} processes 20 circulations according to this Sequentially continuous and can complete a day processing tasks.
Adopt patented method can solve the scheduling problem in mixed-model assembly workshop smoothly, and on solution efficiency, have lifting in various degree, system, consistent method are also convenient to perform and management.To the management planning that mixed-model assembly is produced, there is obvious help.
Claims (3)
1. based on a mixed-model assembly production scheduling control method for honeybee social self-organization model, it is characterized in that: described mixed-model assembly production scheduling control method comprises the following steps:
Step 1, parameter initialization: according to mixed-model assembly production scheduling problems scale set algorithm parameter, described parameter comprises population scale SN, overall controling parameters limit, optimal control parameter bestlimit and initial neighborhood parameter S;
Step 2, initialization of population: for the problem characteristic of mixed-model assembly line scheduling, initialization function selects BCO method initially to employ bee colony, + NEH method initial inspection bee colony at random, wherein random+NEH method is: first adopt NEH method to produce one and observe honeybee, then adopts random device to produce residue and observes honeybee;
Initial population is produced for the mixed method of employing bee colony and observation bee colony to set up respectively based on BCO and random+NEH,
(2.1) employ honeybee initialization step as follows:
Step 2.1.1, step=step+1;
Step 2.1.2, gets one according to the order of sequence and employs honeybee, concentrates select workpiece and renewal sequence coding from unselected workpiece;
Step 2.1.3, judges whether that traversal is all and employs honeybee, do not complete and then repeat step 2.1.2, otherwise continue;
Step 2.1.4, again get one according to the order of sequence in honeybee and employ honeybee employing, judge whether to meet probability, probability is obtained by formula 1, satisfied then continue step 2.1.5, otherwise skips to step 2.1.6;
Step 2.1.5, employs roulette selection one honeybee to separate from remainder and substitutes current solution;
Step 2.1.6, judges whether that traversal allly employs honeybee, satisfied then continuation step 2.1.7, otherwise jumps to step 2.1.4;
Step 2.1.7, judges whether step equals process number, satisfied then continue step 2.1.8, otherwise jumps to step 2.1.1;
Wherein step is from 1 to SN/2 value, currently employs honeybee drop probability pd
caccording to formula 1 value, wherein fit
bestfor the best fitness of current population, fit
choneybee fitness is employed for current;
Step 2.1.8, for the bee colony of employing formed sets up the array limitarray1 () of length SN/2, employs each individuality of honeybee to continue the algebraically do not upgraded for recording;
(2.2) observing the random+NEH method initialization of honeybee: first all workpiece press total elapsed time sort descending, selecting the first two workpiece through comparing reservation deadline the shortest sequence; Secondly, by each position in the 3rd workpiece insetion sequence, retain deadline most short data records equally; Finally, residue workpiece is repeated work of drilling, obtain a feasible solution, this feasible solution be inserted into and employ in bee colony, other individualities of bee colony then adopt random device to generate, and detailed process is:
Step 2.2.1 calculates the total elapsed time of all workpiece, and temporally sort descending;
Step 2.2.2 selects the unit one coming sequence top to form coding, be designated as Solution (i), wherein i=1 represents sequence length, then Solution (i) represents that sequence length is the coding of i, is then rejected from workpiece sequence by the workpiece of selection;
Step 2.2.3 selects the workpiece coming top from workpiece sequence, workpiece is inserted each position in Solution (i) and form i+1 coding, retain according to fitness and best be designated as solution (i+1), this workpiece is rejected from workpiece sequence;
Step 2.2.4 judges whether workpiece sequence is empty, if yes then enter step 2.2.5, otherwise jumps to step 2.2.3;
Step 2.2.5 stochastic generation (SN/2-1) individual coding;
Step 2.2.6 is the array limitarray2 () that the observation bee colony formed sets up length SN/2, for recording the algebraically observed each individuality of honeybee and continue not upgrade;
Step 3, neighborhood scale adjusts: according to initial scale and algorithm process, the search scale of dynamic conditioning neighborhood;
Step 4, employ honeybee algorithm: travel through all employ bee colony and employ honeybee to perform to each employ honeybee algorithm, when employing honeybee to meet limit parameter, triggering search bee algorithm and replacing and employ honeybee;
Step 5, observes honeybee algorithm: travel through all observation bee colonies and observe honeybee execution to each and observe honeybee algorithm, when observing honeybee and meeting limit parameter, trigger search bee algorithm and replace observation honeybee;
Step 6, optimal control algorithm, performs optimal control algorithm, the individuality repeated in displacement population to employing bee colony and observing bee colony,
Wherein, employ bee colony and observe bee colony optimum control process and be:
Step 6.1.1, calculates the quantity bestnum of optimum solution in population;
Step 6.1.2, if bestnum>bestlimit, then i=i+1 continuing, otherwise algorithm terminates;
Step 6.1.3, Stochastic choice two optimum solution bee1 and bee2, if bee1 and bee2 coding is not identical, continues step 6.1.4, otherwise skip to step 6.1.6;
Step 6.1.4, beecross operation is carried out to bee1 and bee2, four Index variablees are the random position produced, and bee1 and bee2 is exchanged by the subsequence of generation random in sequence and the other side and removed repeating part and obtains two new explanation finalbee1 and finalbee2;
Step 6.1.5, judges, if bee1 and bee2 is still optimum, then continues, otherwise skips to step 6.1.7;
Step 6.1.6, carries out search bee algorithm respectively to bee1 and bee2, namely adopts the initial method of random search to produce a RANDOM SOLUTION, and substitutes original individuality;
Step 6.1.7, if i< (bestnum-bestlimit)/2, skips to step 6.1.3;
Step 6.1.8, does simulated annealing operation to all optimum solutions of residue;
Step 7, judges whether the end condition meeting iterations maxcycle, satisfied then stop, otherwise repeats step 3 ~ step 7.
2., as claimed in claim 1 based on the mixed-model assembly production scheduling control method of honeybee social self-organization model, it is characterized in that: in described step 4, described in employ the detailed process of honeybee algorithm as follows:
Step 4.1, honeybee EB1 is employed in select progressively one, and judges whether it is optimum solution in group, satisfied then skip to step 4.3, otherwise continues;
Step 4.2, carries out neighborhood search based on EB1, obtains the individual new individuality of S ', and selects wherein optimum as EB2, skips to step 4.9;
Step 4.3, carries out simulated annealing operation to EB1, sets initial temperature, the Mo Wen in simulated annealing (SA) and moves back warm coefficient;
Step 4.4, carries out neighborhood search to based on EB1 again, obtains the individual new individuality of S ', and selects wherein optimum as EB2;
Step 4.5, compares EB1 and EB2, if EB2 is inferior to EB1, continues step 4.6, otherwise skips to step 4.7;
Step 4.6, random number in 0 ~ 1 scope of generation also compares with probability P, and P obeys formula 3, and wherein fit (EB1) and fit (EB2) represent the fitness value of employing honeybee EB1 and EB2 respectively, if random number is less than probability, continue step 4.7, otherwise skip to step 4.8;
P=exp (-(fit (EB2)-fit (EB1))) formula 3
Step 4.7, EB1=EB2;
Step 4.8, according to formula 4, formula 5 upgrades T and S ' if T<Tend, then continue step 4.9, otherwise repeats step 4.4-4.8, and wherein T represents Current Temperatures, and Tend represents end temperature, and ceil represents and rounds downwards numerical value in bracket;
T=k × Tend formula 4
S'=ceil (k × S') formula 5
Step 4.9, greedy rule compares EB1 and EB2, if EB2 is not better than EB1, then limitarray1 (i)=limitarray1 (i)+1, i represents EB1 sequence number, otherwise skips to step 4.11;
Step 4.10, if limitarray1 (i)=limit, calls search bee algorithm and replaces EB1, and skip to step 4.12, otherwise jump directly to step 4.12;
Step 4.11, EB1=EB2, and limitarray1 (i)=0;
Whether step 4.12, travel through and allly employ honeybee, if i.e. i=SN/2, then algorithm terminates, otherwise skips to step 4.1.
3. as claimed in claim 2 based on the mixed-model assembly production scheduling control method of honeybee social self-organization model, it is characterized in that: in described step 5, the detailed process of described observation honeybee algorithm is as follows:
Step 5.1, honeybee OB1 is observed in select progressively one, employs honeybee EB1 according to fitness roulette selection;
Step 5.2, greedy rule compares OB1 and EB1, if EB1 is not better than OB1, then limitarray2 (j)=limitarray2 (j)+1, j represents OB1 sequence number, otherwise skips to step 5.4;
Step 5.3, if limitarray2 (j)=limit, calls search bee algorithm and replaces OB1, and skip to step 5.4, otherwise jump directly to step 5.5;
Step 5.4, OB1=EB1, and limitarray2 (i)=0;
Whether step 5.5, travel through all observation honeybees, if i.e. j=SN/2, then algorithm terminates, otherwise skips to step 5.1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110366396.8A CN102411306B (en) | 2011-11-17 | 2011-11-17 | Mixed flow assembly production scheduling control method based on bee society self-organization model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110366396.8A CN102411306B (en) | 2011-11-17 | 2011-11-17 | Mixed flow assembly production scheduling control method based on bee society self-organization model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102411306A CN102411306A (en) | 2012-04-11 |
| CN102411306B true CN102411306B (en) | 2014-12-17 |
Family
ID=45913428
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110366396.8A Active CN102411306B (en) | 2011-11-17 | 2011-11-17 | Mixed flow assembly production scheduling control method based on bee society self-organization model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102411306B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104571007B (en) * | 2014-11-20 | 2018-11-27 | 昆明理工大学 | The Optimization Scheduling of general assembly line production assembling process in a kind of automotive manufacture |
| CN107437121B (en) * | 2016-05-25 | 2020-07-10 | 华中科技大学 | Production process control method suitable for simultaneously processing single workpiece by multiple machines |
| CN106909992B (en) * | 2017-03-02 | 2021-04-30 | 河南科技大学 | Split charging production method and device of multi-variety small-batch mixed flow assembly line |
| CN107247447A (en) * | 2017-05-23 | 2017-10-13 | 长安大学 | A kind of mixed-model assembly dispatch control method based on hybrid algorithm framework |
| CN107730065B (en) * | 2017-06-12 | 2018-07-24 | 合肥工业大学 | Based on the production scheduling method and system for improving variable neighborhood search algorithm |
| CN108647859A (en) * | 2018-04-17 | 2018-10-12 | 东莞理工学院 | The double population collaborative learning strategies of displacement assembly line and optimization method of Knowledge driving |
| CN109740817B (en) * | 2018-12-30 | 2020-07-31 | 江苏金陵智造研究院有限公司 | Workshop facility layout optimization method based on improved artificial bee colony algorithm |
| CN110084545B (en) * | 2019-03-05 | 2021-07-27 | 浙江工业大学 | Integrated scheduling method of multi-lane automatic stereoscopic warehouse based on mixed integer programming model |
| CN110488271A (en) * | 2019-08-01 | 2019-11-22 | 中国人民解放军陆军工程大学 | A kind of one-dimensional range profile speed compensation method |
| CN110516978A (en) * | 2019-09-02 | 2019-11-29 | 上海无线电设备研究所 | A kind of electronic product commissioning production line mixed flow scheduled production method |
| CN111798120A (en) * | 2020-06-30 | 2020-10-20 | 南京信息工程大学 | Flexible job shop scheduling method based on improved artificial bee colony algorithm |
| CN113313283B (en) * | 2021-04-14 | 2022-08-19 | 兰州理工大学 | Knowledge-driven method-based distributed assembly blocking flow workshop integrated scheduling method |
| CN113095572A (en) * | 2021-04-14 | 2021-07-09 | 兰州理工大学 | Distributed assembly zero-waiting flow shop scheduling problem optimization system |
-
2011
- 2011-11-17 CN CN201110366396.8A patent/CN102411306B/en active Active
Non-Patent Citations (4)
| Title |
|---|
| Teodorovic, Dusan et al..Bee Colony Optimization: Principles and Applications.《Proceedinds of the 8th seminar on Neural Network Applications in Electrical Engineering,IEEE》.2006,第151-156页. * |
| 基于粒子群优化和模拟退火的混合调度算法;潘全科 等;《中国机械工程》;20060531;第17卷(第10期);第1044-1046,1064页 * |
| 求解批量流水线调度问题的离散蜂群算法;桑红燕 等;《中国机械工程》;20110930;第22卷(第18期);第2195-2202页 * |
| 混合蜂群算法求解柔性作业车间调度问题;李修琳 等;《计算机集成制造系统》;20110731;第17卷(第7期);第1495-1500页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102411306A (en) | 2012-04-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102411306B (en) | Mixed flow assembly production scheduling control method based on bee society self-organization model | |
| Zaman et al. | Evolutionary algorithms for dynamic economic dispatch problems | |
| CN109190857B (en) | Optimization algorithm based on multi-target resource limited project scheduling model | |
| Son et al. | Multiheuristic approach for resource leveling problem in construction engineering: Hybrid approach | |
| CN101901425A (en) | Flexible job shop scheduling method based on multi-species coevolution | |
| CN107168267A (en) | Based on the production scheduling method and system for improving population and heuristic strategies | |
| Liu et al. | Aggregate production planning for shipbuilding with variation-inventory trade-offs | |
| CN114548660A (en) | Production scheduling method, device, equipment and storage medium for household electrical appliance | |
| CN113139710B (en) | Multi-resource parallel task advanced plan scheduling method based on genetic algorithm | |
| CN103310279B (en) | Multi-objective optimization product configuration method based on hybrid ant colony | |
| CN109472362A (en) | A kind of AGV dynamic dispatching method and device based on variable task window | |
| Maghsoudlou et al. | A framework for preemptive multi-skilled project scheduling problem with time-of-use energy tariffs | |
| CN108279647A (en) | A kind of two benches assembly flow-shop scheduling | |
| CN110135752A (en) | A kind of dispatching method of the whole-set order with switching time | |
| Pimentel et al. | Optimization of vertical farms energy efficiency via multiperiodic graph-theoretical approach | |
| Fumagalli et al. | A novel scheduling framework: Integrating genetic algorithms and discrete event simulation | |
| Shastri et al. | A novel decomposition and distributed computing approach for the solution of large scale optimization models | |
| Kumar et al. | Stochastic make-to-stock inventory deployment problem: an endosymbiotic psychoclonal algorithm based approach | |
| Fumagalli et al. | Simulation-supported framework for job shop scheduling with genetic algorithm | |
| CN116774657A (en) | Dynamic scheduling method for remanufacturing workshop based on robust optimization | |
| Sihlongonyane et al. | The fourth industrial revolution: Synopses and implications for STI policy development | |
| Wang et al. | Modeling and analysis for multi-period, multi-product and multi-resource production scheduling | |
| CN110321208A (en) | A kind of evolutionary computation method solving cloud task schedule | |
| Chen et al. | A multi-population genetic algorithm based on chaotic migration strategy and its application to inventory programming | |
| ÖZKAYA | Industry 4.0 Journey and Key Resources |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |