CN105956689A - Transportation and production coordinated scheduling method based on improved particle swarm optimization - Google Patents

Abstract

The invention discloses a transportation and production coordinated scheduling method based on improved particle swarm optimization. The transportation and production coordinated scheduling method is characterized in that 1, work pieces are tested in a grouped way; 2, algorithm parameters are set; 3, an initial population is generated; 4, an adaption degree value is calculated; 5, whether a termination condition is satisfied is determined, and when the termination condition is satisfied, a global optimal solution is output, and when the termination condition is not satisfied, the process is continued; 6, a local optimal solution and the global optimal solution are updated; 7, positions and speeds of particles are updated; 8, the intersection operations of the particles are carried out; 9, the immigration operations of the particles are carried out; 10, the particle positions are adjusted, and the process returns to the step 4. The transportation and production coordinated scheduling method is advantageous in that optimization of enterprise total economic benefits is realized, and therefore enterprise costs are reduced, and enterprise service level is improved.

Description

A kind of transport based on Modified particle swarm optimization and procreative collaboration dispatching method

Technical field

The invention belongs to supply chain field, a kind of transport based on Modified particle swarm optimization and procreative collaboration Dispatching method, in the Production scheduling management of enterprise.

Background technology

Under Vehicles Collected from Market environment, the core competitiveness of manufacturing enterprise is no longer simple manufacturing capacity, but supply Chain collaboration capabilities.The aggravation of the market competition, makes the cooperation between enterprise more and more closer.In order to technology, resource etc. are concentrated on The key link produced, some operation is often contracted out to multiple manufacturer by enterprise.Different manufacturers is often positioned in different Geographical position, which results in and there is different haulage times between enterprise and different manufacturer.Enterprise only manufactures with other Enterprise is collaborative together carries out overall situation control by the link such as transport, production, production system and logistics transportation system is carried out further Combined optimization, could meet the timely supply of consumer product to greatest extent, thus obtain the maximization of overall economic benefit, promote The competitiveness of enterprise.

Cooperative scheduling is the class optimization method towards supply chain, uses the mode of accurately scheduling, each in design supply chain The cooperative scheduling scheme of link, it is achieved the optimization of enterprise-wide economic benefit, thus the service level of enterprise.

At present, the research to cooperative scheduling problem is concentrated mainly under traditional mode of production pattern, in this kind of production model, and one Platform equipment can process an operation or the operation of a collection of fixed qty simultaneously, but often do not account for multiple manufacturer and divide Cloth is in the situation of diverse geographic location.In reality industry, this kind of production model is to exist.And traditional dispatching method is often Haulage time is ignored or abstract for identical, therefore can not adapt to Production requirement instantly.

Summary of the invention

The present invention is the weak point in order to overcome prior art to exist, it is provided that a kind of production based on particle group optimizing is joined Send coordinated dispatching method, to the optimization of overall economic benefit can be realized, it is thus possible to reduce production cost, promote work effect Rate.

The present invention solves that technical problem adopts the following technical scheme that

The feature of a kind of transport based on Modified particle swarm optimization of the present invention and procreative collaboration dispatching method is to be in work After n workpiece at part set carries out batch processing, it is delivered at m platform equipment carry out production and processing by haulage vehicle；Described n The workpiece set that individual workpiece is constituted is designated as J={J₁,J₂,…,J_j,…,J_n, J_jRepresent jth workpiece, 1≤j≤n；By jth Workpiece J_jSize be designated as s_j, process time be designated as p_j；Described m platform equipment is designated as M={M₁,M₂,…,M_i,…,M_m, M_iRepresent the I platform equipment, 1≤i≤m；By i-th equipment M_iThe speed of processing workpiece is designated as v_i, i-th will be arrived at described workpiece set Equipment M_iHaulage time be designated as t_i；The volume of the volume of haulage vehicle and processing workpiece device is all designated as C；

Described transport and procreative collaboration dispatching method are to carry out as follows:

Step 1, the workpiece in described workpiece set J is ranked up by the order of non-increasing process time, if process time Identical, it is ranked up by the order of the non-increasing of workpiece size, thus obtains the workpiece set J ' after sequence；

Step 2, the 1st unappropriated workpiece in the workpiece set J ' after described sequence is put into can accommodate the described 1st The size of individual unallocated workpiece and remaining space minimum batch in, batch remaining space be that volume C owns to putting in corresponding batch The difference of workpiece size sum；If can not accommodate the size of the 1st unallocated workpiece in the remaining space criticized, then generating volume is New batch of C, and the 1st unappropriated workpiece is added in new batch, until all workpiece in described workpiece set J ' all distribute In corresponding batch；

Step 3, obtain in step 2 all batches are arranged by batch processing time non-increasing, obtain batch processing set B= {b₁,b₂,…,b_q,…,b_l, b_qRepresenting that q-th is criticized, process time q-th criticized is designated as p^(q), q-th criticizes b_qAdd man-hour Between p^(q)It is to be criticized b by q-th_qThe workpiece decision that middle process time is the longest；Lot count is designated as l, Table Show the smallest positive integral not less than x；

Step 4, the parameters of initialization particle cluster algorithm, including: total number of particles N, iterations L, greatest iteration time Number L_max, crossover probability P_c1And P_c2, 1≤L≤L_max；And initialize L=1；

Step 5, generation initial population, it is thus achieved that the initial position of the kth particle in L generation And initial velocity WithRepresent that in L generation, kth particle is searched in d dimension respectively Rope position spatially and speed, wherein, 1≤d≤l；1≤k≤N；

Step 6, calculate the fitness of kth particle in L generationThus obtain the local of kth particle in L generation Optimal solutionWherein,Represent that in L generation, kth particle ties up search volume at d On optimal location；

Step 7, repetition step 6, it is thus achieved that the locally optimal solution of N number of particle in L generation, and therefrom select maximum adaptation angle value Corresponding optimal solution, as the globally optimal solution in L generation, is designated asWhereinRepresent Whole particle colony optimal location on d dimension search volume in L generation；

Step 8, according to the position of kth particle in L generationAnd speedCalculate the kth in L+1 generation respectively The position of particleAnd speedThus obtain L+1 for the position of particle N number of in population and speed；

Step 9, adjust the described L+1 position for particle N number of in population, thus obtain the N in the L+1 generation after adjustment The position of individual particle；

Step 10, calculate L+1 for kth particle in populationFitnessAnd with L generation in kth The fitness of particleCompare, using particle position corresponding for bigger fitness value as kth particle in L+1 generation Optimal solution

Step 11, the step 10 that repeats, thus obtain the optimal solution of N particle in L+1 generation and therefrom select maximum adaptation degree The optimal solution of value correspondence is as the globally optimal solution in L+1 generation

Step 12, L+1 is assigned to L, it is judged that L ＜ L_maxWhether setting up, if setting up, then performing step 8；Otherwise, represented Become L_maxSecondary iteration, and obtain globally optimal solutionWith globally optimal solutionCorresponding scheduling scheme is adjusted as optimum Degree scheme.

The feature of production scheduling method based on particle group optimizing of the present invention lies also in,

During described step 5 creates initial population, N-1 particle is to be produced by random fashion, remaining one Particle is to produce as follows:

Step 5.1, orderRepresent i-th equipment M_iUpper q-th criticizes b_qCompletion date, n_iRepresent i-th equipment M_iOn The quantity criticized of processing, A_iIt is i-th equipment M_iFree time；Initializen_i=1, q=1, A_i=0；

Step 5.2, judge whether q ＞ l sets up, if setting up, then it represents that generate a particle；Otherwise, step 5.3 is performed；

Step 5.3, utilize formula (1) obtain q-th criticize b_qCompletion date on i-th equipmentThus obtain q Individual crowd of b_qCompletion date on m platform equipment:

$C_{b_q}\[i\]=max\{t_i\×n_i,A_i\}+\frac{p^{\left(q\right)}}{v_i}---\left(1\right)$

In formula (1), max{x, y} represent the greater taking in x and y；

Step 5.4, criticize b from q-th_qCompletion date on m platform equipment selects setting corresponding to minimum makespan Standby, it is designated as equipment min, by q-th batch b_qTransport is processed on equipment min；Again by n_min+ 1 is assigned to n_min,It is assigned to A_min, after q+1 is assigned to q, perform step 5.2.

Described step 8 is to carry out as follows:

Step 8.1, by the initial position of the kth particle in L generationIt is assigned toWherein,For withThere is the variable of identical meanings, compareWithThe value of middle corresponding element is the most identical, If identical, then willThe value of middle corresponding element is set to 0, if it is different, then willMiddle corresponding element value is set toIn The value of corresponding element；Thus obtain renewalUpdateIn the value of each element represent the sequence number of equipment；And it is every Individual element and each batch of one_to_one corresponding in batch processing set B；

Step 8.2, by updateCriticizing corresponding to the element of middle non-zero value is dispatched to setting representated by non-zero value successively It is processed on Bei, and calculates the completion date of relevant device；Thus obtain equipment completion date set C '=C ' [1] ..., C ' [i] ..., C ' [m] } and batch size set n '={ n '₁,…,n′_i,…,n′_m}；C ' [i] represents i-th equipment M_iComplete Between man-hour, n '_iRepresent i-th equipment M_iThe quantity criticized of upper processing；

Step 8.3, by n '_i+ 1 is assigned to n '_i, C ' [i] is assigned to A '_i, wherein, variables A '_iWith A_iThere is identical meanings；

Step 8.4, by updateMiddle all values is the batch corresponding to the element of 0, by batch processing time non-increasing Arrangement obtains crowd set B '={ b '₁,…,b′_q′,…,b′_l′, wherein, l ' represents renewalIntermediate value is the total of the element of 0 Number, the most unappropriated batch total；

Step 8.5, make q '=1；

If step 8.6 q ' is ＞ l ', represent renewalAgain update, and perform step 8.9；No Then, step 8.7 is performed；

Step 8.7, formula (2) is utilized to obtain q ' individual batch b '_q′Completion date on i-th equipmentThus obtain Obtain q ' individual batch b '_q′Completion date on m platform equipment:

$C_{b_{q^{\′}}^{\′}}\[i\]=max\{t_i\×n_i^{\′},A_i^{\′}\}+\frac{p^{\left(q^{\′}\right)}}{v_i}---\left(2\right)$

Step 8.8, from q ' individual batch b '_q′Completion date on m platform equipment selects corresponding to minimum makespan Equipment, be designated as equipment min ', by q ' individual batch b '_q′Transport is processed on equipment min '；Again by n '_min′+ 1 is assigned to n′_min′,It is assigned to A '_min′, q '+1 is assigned to q ', performs step 8.6；

Step 8.9, by bernoulli distribution obtain l dimension group R being made up of 0 and 1₁, array R₁In each element With again updateIn each element one_to_one corresponding；If R₁In element be 1, the most again updateMiddle corresponding element The value of element keeps constant, if R₁In element be 0, the most again updateThe value of middle corresponding element is set to 0；

Step 8.10, by again updateAs updateAnd substitute into execution in step 8.2-step 8.8, from And after obtaining third time renewal

Step 8.11, generalAsWillAsAnd substitute into execution in step 8.1-step 8.10, from And after obtaining third time renewal

Step 8.12, will third time update afterAfter updating with third timeWith probability P_c1Carry out bernoulli friendship Fork, the position of the kth particle in the L generation after being intersected

Step 8.13, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the kth particle in L+1 generation Speed

Step 8.14, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the kth particle in L+1 generation Position

Step 8.15, generalWith globally optimal solutionWith probability P_c2Carry out bernoulli intersection, it is thus achieved that after two intersect Particle, relatively prechiasmal particleWith intersect after the fitness of two particles, using particle big for fitness as The kth particle in L+1 generationPosition；

Step 8.16, calculate the L+1 fitness for particle each in population, fitness value is minimumThe individual particle particle randomly generated replaces,Represent the maximum integer less than x；Thus update L+1 Position for population.

Described step 9 is to be adjusted as follows:

Step 9.1, defined variable f and h, the greatest length of definition Local Search is f_max；Make f=1, i=1, h=m；

Step 9.2, judge f ＜ f_maxWhether setting up, if setting up, then m platform equipment being arranged by the non-increasing of completion date Row, perform step 9.3；Otherwise, L+1 is completed for particle position N number of in populationAdjustment；

Step 9.3, i-th equipment M of selection_i；Select h platform equipment M_h；

Step 9.4, judging whether h ＞ 1 sets up, if setting up, then performing step 9.5；Otherwise, L+1 is completed for population In N number of particle positionAdjustment；

Step 9.5, find any α crowd of b_α, b_α∈M_i；And any β crowd b_β, b_β∈M_h；If α crowd b_α With β crowd b_βMeet formula (3), then exchange α crowd b_αWith β crowd b_β, and perform step 9.6；Otherwise, step is performed 9.7:

$\left\{\begin{array}{c}0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{c_h}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_h}<C\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\\ 0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{v_i}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_i}<C\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\end{array}\right.---\left(3\right)$

In formula (3), p^(α)Represent α crowd b_αProcess time, C [i] represents i-th equipment M_iCompletion date, p^(β)Table Show β crowd b_βProcess time, C [h] represents h platform equipment M_hCompletion date,

Step 9.6, by be assigned on same equipment batch by non-increasing arrangement batch process time, and calculate individual device Completion date, is assigned to f by f+1, performs step 9.2；

Step 9.7, h-1 is assigned to h after, perform step 9.4.

Compared with the prior art, the present invention has the beneficial effect that:

1, the present invention is under typical difference is produced in batches pattern, and transport and the production two benches of research manufacturing enterprise are worked in coordination with Scheduling problem, by using modified particle swarm optiziation, is then based on haulage time in batches first against difference workpiece, criticizes The process velocity of process time and equipment proposes corresponding scheduling strategy, draws position and the speed of particle；Recycling particle Current location and speed more new regulation, update particle position and speed, it is achieved that successive ignition, finally obtain optimal solution；Grain Swarm optimization, on the degree of optimization of available time and result, is that a kind of performance preferably optimizes the approximation calculation manufacturing span Method；Solve job batching transport and the combined optimization problem of production in reality industry, it is achieved that enterprise-wide economic benefit Optimize, reduce energy consumption, provide cost savings, improve the service level of enterprise.

2, the present invention is producing during initial population based on haulage time, batch processed time and the processing of equipment The corresponding strategy of speed proposition, is assigned to make on batch equipment completed at first by batch, thus in generation initial population one by one Body, produces other individualities in population, both ensure that the quality of initial population, and also ensured that in conjunction with random population producing method The multiformity of initial population.

3, the present invention is updated by the optimal solution obtaining iteration each time, too fast during solving algorithm search Focus on locally optimal solution problem；In location updating rule, add the crossover operator in genetic algorithm and one adaptive The immigrant's operator answered, while ensureing to solve quality, also maintains the multiformity of population.

4, present invention process velocity based on batch processing time and equipment devises the adjustable strategies of particle position；To set For being ranked up by the non-increasing of completion date, then by criticizing on equipment shorter to equipment longer for completion date and completion date Secondary it is adjusted by constraints, thus realizes the position of the particle of iteration each time is adjusted, improve the quality understood.

Accompanying drawing explanation

Fig. 1 is that the present invention uses particle cluster algorithm to enter the method flow diagram of transport production cooperative scheduling；

Fig. 2 is that the present invention transports and flow chart.

Detailed description of the invention

In this embodiment, a kind of transport based on particle group optimizing and production scheduling method, its flow process as it is shown in figure 1, Be variant for workpiece volume and production time, produce that device rate is variant and on different transit route there is haulage time The transport of difference, procreative collaboration scheduling problem are modeled, and are then solved by a kind of particle cluster algorithm that improves, thus The prioritization scheme dispatched to a set of transport production, is substantially reduced total operating cost of Target Enterprise with this, improves corporate operation effect Rate.Specifically, it is will to be in after n workpiece at workpiece set carry out batch processing, be delivered to m platform by haulage vehicle and set Standby place carries out production and processing；The workpiece set of n workpiece composition is designated as J={J₁,J₂,…,J_j,…,J_n, J_jRepresent jth work Part, 1≤j≤n；By jth workpiece J_jSize be designated as s_j, process time be designated as p_j；M platform equipment is designated as M={M₁,M₂,…, M_i,…,M_m, M_iRepresent i-th equipment, 1≤i≤m；By i-th equipment M_iThe speed of processing workpiece is designated as v_i, will be from workpiece collection I-th equipment M is arrived at conjunction_iHaulage time be designated as t_i；The volume of the volume of haulage vehicle and processing workpiece device is all designated as C；

Transport based on particle group optimizing and production scheduling method are to carry out as follows:

Step 1, the workpiece in workpiece set J is ranked up by the order of non-increasing process time, if process time is identical Then it is ranked up by the order of the non-increasing of workpiece size, thus obtains the workpiece set J ' after sequence；

Step 2, will sequence after workpiece set J ' in the 1st unappropriated workpiece put into can accommodate the 1st unallocated The size of workpiece and remaining space minimum batch in, batch remaining space be volume C and put into all workpiece sizes in corresponding batch The difference of sum；If can not accommodate the size of the 1st unallocated workpiece in the remaining space criticized, then generating volume is that the new of C is criticized, And added in new batch by the 1st unappropriated workpiece, until all workpiece in workpiece set J ' are all assigned in corresponding batch； This in batches strategy in same batch, make the residue workpiece of equipment to the greatest extent may be used the workpiece close process time as far as possible distribution simultaneously Can little, can the most less batch sum and batch process time.

Step 3, obtain in step 2 all batches are arranged by batch processing time non-increasing, obtain batch processing set B= {b₁,b₂,…,b_q,…,b_l, b_qRepresenting that q-th is criticized, process time q-th criticized is designated as p^(q), q-th criticizes b_qAdd man-hour Between p^(q)It is to be criticized b by q-th_qThe workpiece decision that middle process time is the longest；Lot count is designated as l, Table Show the smallest positive integral not less than x；Arrange by batch processing time non-increasing by criticizing, make to be assigned on every equipment processing Batch processing during be also processed by batch processing time non-increasing, in the case of there are haulage time, So can reduce the waiting time of machine.

Step 4, the parameters of initialization particle cluster algorithm, including: total number of particles N, iterations L, greatest iteration time Number L_max, crossover probability P_c1And P_c2, 1≤L≤L_max；And initialize L=1；

Step 5, generation initial population, it is thus achieved that the initial position of the kth particle in L generation And initial velocity WithRepresent that in L generation, kth particle is searched in d dimension respectively Rope position spatially and speed, wherein, 1≤d≤l；1≤k≤N；During producing initial population, one of them is individual By being assigned to make to obtain on batch machine completed the earliest by criticizing, producing individual by random manner of other, both ensure that The quality of initial population, maintains the multiformity of population, also improves convergence of algorithm speed, specifically,

Step 5.1, orderRepresent i-th equipment M_iUpper q-th criticizes b_qCompletion date, n_iRepresent i-th equipment M_iOn The quantity criticized of processing, A_iIt is i-th equipment M_iFree time；Initializen_i=1, q=1, A_i=0；

Step 5.2, judge whether q ＞ l sets up, if setting up, then it represents that generate a particle；Otherwise, step 5.3 is performed；

Step 5.3, utilize formula (1) obtain q-th criticize b_qCompletion date on i-th equipmentThus obtain q Individual crowd of b_qCompletion date on m platform equipment:

$C_{b_q}\&lsqb;i\&rsqb;=max\{t_i\&times;n_i,A_i\}+\frac{p^{\left(q\right)}}{v_i}---\left(1\right)$

In formula (1), max{x, y} represent the greater taking in x and y；

Step 5.4, criticize b from q-th_qCompletion date on m platform equipment selects setting corresponding to minimum makespan Standby, it is designated as equipment min, by q-th batch b_qTransport is processed on equipment min；Again by n_min+ 1 is assigned to n_min,It is assigned to A_min, after q+1 is assigned to q, perform step 5.2.

Based on device rate, criticize b by calculating q-th_qCompletion date on m platform equipment, reselection minimum complete man-hour Equipment corresponding between criticizes b as q-th_qProcess equipment, so criticize b than by q-th_qIt is assigned on equipment idle at first It is processed solving result more excellent.

Step 6, calculate the fitness of kth particle in L generationThus obtain the local of kth particle in L generation Optimal solutionWherein,Represent that in L generation, kth particle ties up search volume at d On optimal location；

Step 7, repetition step 6, it is thus achieved that the locally optimal solution of N number of particle in L generation, and therefrom select maximum adaptation angle value Corresponding optimal solution, as the globally optimal solution in L generation, is designated asWhereinRepresent Whole particle colony optimal location on d dimension search volume in L generation；

Step 8, according to the position of kth particle in L generationAnd speedCalculate the kth in L+1 generation respectively The position of particleAnd speedThus obtain L+1 for the position of particle N number of in population and speed；

Step 8.1, by the initial position of the kth particle in L generationIt is assigned toWherein,For withThere is the variable of identical meanings, compareWithThe value of middle corresponding element is the most identical, If identical, then willThe value of middle corresponding element is set to 0, if it is different, then willMiddle corresponding element value is set toIn right Answer the value of element；Thus obtain renewalUpdateIn the value of each element represent the sequence number of equipment；And it is each Element and each batch of one_to_one corresponding in batch processing set B；

Step 8.2, by updateCriticizing corresponding to the element of middle non-zero value is dispatched to setting representated by non-zero value successively It is processed on Bei, and calculates the completion date of relevant device；Thus obtain equipment completion date set C '=C ' [1] ..., C ' [i] ..., C ' [m] } and batch size set n '={ n '₁,…,n′_i,…,n′_m}；C ' [i] represents i-th equipment M_iComplete Between man-hour, n '_iRepresent i-th equipment M_iThe quantity criticized of upper processing；

Step 8.3, by n '_i+ 1 is assigned to n '_i, C ' [i] is assigned to A '_i, wherein, variables A '_iWith A_iThere is identical meanings；

Step 8.4, by updateMiddle all values is the batch corresponding to the element of 0, by batch processing time non-increasing Arrangement obtains crowd set B '={ b '₁,…,b′_q′,…,b′_l′, wherein, l ' represents renewalIntermediate value is the total of the element of 0 Number, the most unappropriated batch total；

Step 8.5, make q '=1；

If step 8.6 q ' is ＞ l ', represent renewalAgain update, and perform step 8.9；No Then, step 8.7 is performed；

Step 8.7, formula (2) is utilized to obtain q ' individual batch b '_q′Completion date on i-th equipmentThus obtain Obtain q ' individual batch b '_q′Completion date on m platform equipment:

$C_{b_{q^{\&prime;}}^{\&prime;}}\&lsqb;i\&rsqb;=max\{t_i\&times;n_i^{\&prime;},A_i^{\&prime;}\}+\frac{p^{\left(q^{\&prime;}\right)}}{v_i}---\left(2\right)$

Step 8.8, from q ' individual batch b '_q′Completion date on m platform equipment selects corresponding to minimum makespan Equipment, be designated as equipment min ', by q ' individual batch b '_q′Transport is processed on equipment min '；Again by n '_min′+ 1 is assigned to n′_min′,It is assigned to A '_min′, q '+1 is assigned to q ', performs step 8.6；

Step 8.9, by bernoulli distribution obtain l dimension group R being made up of 0 and 1₁, array R₁In each element With again updateIn each element one_to_one corresponding；If R₁In element be 1, the most again updateMiddle corresponding element The value of element keeps constant, if R₁In element be 0, the most again updateThe value of middle corresponding element is set to 0；

Step 8.10, by again updateAs updateAnd substitute into execution in step 8.2-step 8.8, from And after obtaining third time renewal

Step 8.11, generalAsWillAsAnd substitute into execution in step 8.1-step 8.10, from And after obtaining third time renewal

Step 8.12, will third time update afterAfter updating with third timeWith probability P_c1Carry out bernoulli friendship Fork, the position of the kth particle in the L generation after being intersected

Step 8.13, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the kth particle in L+1 generation Speed

Step 8.14, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the kth particle in L+1 generation Position

Step 8.15, generalWith globally optimal solutionWith probability P_c2Carry out bernoulli intersection, it is thus achieved that after two intersect Particle, relatively prechiasmal particleWith intersect after the fitness of two particles, using particle big for fitness as The kth particle in L+1 generationPosition；Compare by the particle before and after intersecting is carried out fitness, it is ensured that will produce The high particle of fitness value remain, thus improve the quality of population.

Step 8.16, calculate the L+1 fitness for particle each in population, fitness is minimumThe individual particle particle randomly generated replaces,Represent the maximum integer less than x；Thus update L+ The position of 1 generation population.Pass throughCalculate and the population of iteration each time needs the particle with randomly generating replace The number of particles in generation, this formula increases along with iterations L, and the number of particles that the particle being randomly generated replaces is along with iteration time Number increases and gradually increases, and thus can increase the ability of Local Search in iteration early stage, make population have relatively in the iteration later stage Big multiformity, it is to avoid be absorbed in local optimum.

Step 9, adjust the L+1 position for particle N number of in population, thus obtain N number of grain in the L+1 generation after adjustment The position of son；

Step 9.1, defined variable f and h, the greatest length of definition Local Search is f_max；Make f=1, i=1, h=m；

Step 9.2, judge f ＜ f_maxWhether setting up, if setting up, then m platform equipment being arranged by the non-increasing of completion date Row, perform step 9.3；Otherwise, L+1 is completed for particle position N number of in populationAdjustment；

Step 9.3, i-th equipment M of selection_i；Select h platform equipment M_h；

Step 9.4, judging whether h ＞ 1 sets up, if setting up, then performing step 9.5；Otherwise, L+1 is completed for population In N number of particle positionAdjustment；

Step 9.5, find any α crowd of b_α, b_α∈M_i；And any β crowd b_β, b_β∈M_h；If α crowd b_α With β crowd b_βMeet formula (3), then exchange α crowd b_αWith β crowd b_β, and perform step 9.6；Otherwise, step is performed 9.7:

$\left\{\begin{array}{c}0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{v_h}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_h}<C\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\\ 0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{v_i}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_i}<\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\end{array}\right.---\left(3\right)$

In formula (3), p^(α)Represent α crowd b_αProcess time, C [i] represents i-th equipment M_iCompletion date, p^(β)Table Show β crowd b_βProcess time, C [h] represents h platform equipment M_hCompletion date；Criticizing of formula (3) will be met swap Can shorten the completion date of the longest equipment of completion date, and the completion date of equipment that completion date is less will not be made to surpass Crossing the completion date of the longest equipment of the front completion date of exchange, therefore this exchange can shorten manufacture span.

Step 9.6, by be assigned on same equipment batch by non-increasing arrangement batch process time, and calculate individual device Completion date, is assigned to f by f+1, performs step 9.2；

Step 9.7, h-1 is assigned to h after, perform step 9.4.

Step 10, calculate L+1 for kth particle in populationFitnessAnd with L generation in kth The fitness of particleCompare, using particle position corresponding for bigger fitness value as kth particle in L+1 generation Optimal solution

Step 11, the step 10 that repeats, thus obtain the optimal solution of N particle in L+1 generation and therefrom select maximum adaptation degree The optimal solution of value correspondence is as the globally optimal solution in L+1 generation

Step 12, L+1 is assigned to L, it is judged that L ＜ L_maxWhether setting up, if setting up, then performing step 8；Otherwise, represented Become L_maxSecondary iteration, and obtain globally optimal solutionWith globally optimal solutionCorresponding scheduling scheme is adjusted as optimum Degree scheme, workpiece batching and transport and manufacture process are as shown in Figure 2.

Claims (4)

1. transport based on Modified particle swarm optimization and a procreative collaboration dispatching method, is characterized in that, will be in workpiece set After n the workpiece at place carries out batch processing, it is delivered at m platform equipment carry out production and processing by haulage vehicle；Described n workpiece The workpiece set constituted is designated as J={J₁,J₂,…,J_j,…,J_n, J_jRepresent jth workpiece, 1≤j≤n；By jth workpiece J_j Size be designated as s_j, process time be designated as p_j；Described m platform equipment is designated as M={M₁,M₂,…,M_i,…,M_m, M_iRepresent i-th to set Standby, 1≤i≤m；By i-th equipment M_iThe speed of processing workpiece is designated as v_i, i-th equipment M will be arrived at described workpiece set_i Haulage time be designated as t_i；The volume of the volume of haulage vehicle and processing workpiece device is all designated as C；

Described transport and procreative collaboration dispatching method are to carry out as follows:

Step 1, the workpiece in described workpiece set J is ranked up by the order of non-increasing process time, if process time is identical Then it is ranked up by the order of the non-increasing of workpiece size, thus obtains the workpiece set J ' after sequence；

Step 2, the 1st unappropriated workpiece in the workpiece set J ' after described sequence is put into can accommodate described 1st not The distribution size of workpiece and remaining space minimum batch in, batch remaining space be volume C and put into all workpiece in corresponding batch The difference of size sum；If can not accommodate the size of the 1st unallocated workpiece in the remaining space criticized, then generating volume is C's New batch, and the 1st unappropriated workpiece is added in new batch, until all workpiece in described workpiece set J ' are all assigned to phase Answer batch in；

Step 3, obtain in step 2 all batches are arranged by batch processing time non-increasing, obtain batch processing set B={b₁, b₂,…,b_q,…,b_l, b_qRepresenting that q-th is criticized, process time q-th criticized is designated as p^(q), q-th criticizes b_qP process time^(q) It is to be criticized b by q-th_qThe workpiece decision that middle process time is the longest；Lot count is designated as l, Represent not Smallest positive integral less than x；

Step 4, the parameters of initialization particle cluster algorithm, including: total number of particles N, iterations L, maximum iteration time L_max, crossover probability P_c1And P_c2, 1≤L≤L_max；And initialize L=1；

Step 5, generation initial population, it is thus achieved that the initial position of the kth particle in L generation And initial velocity WithRepresent that in L generation, kth particle is searched in d dimension respectively Rope position spatially and speed, wherein, 1≤d≤l；1≤k≤N；

Step 6, calculate the fitness of kth particle in L generationThus obtain the local optimum of kth particle in L generation SolveWherein,Represent that in L generation, kth particle is on d dimension search volume Optimal location；

Step 7, repetition step 6, it is thus achieved that the locally optimal solution of N number of particle in L generation, and it is corresponding therefrom to select maximum adaptation angle value Optimal solution as the globally optimal solution in L generation, be designated asWhereinRepresent L Whole particle colony optimal location on d dimension search volume in Dai；

Step 8, according to the position of kth particle in L generationAnd speedCalculate the kth particle in L+1 generation respectively PositionAnd speedThus obtain L+1 for the position of particle N number of in population and speed；

Step 9, adjust the described L+1 position for particle N number of in population, thus obtain N number of grain in the L+1 generation after adjustment The position of son；

Step 10, calculate L+1 for kth particle in populationFitnessAnd with L generation in kth particle FitnessCompare, using particle position corresponding for bigger fitness value as the optimum of kth particle in L+1 generation Solve

Step 11, the step 10 that repeats, thus obtain the optimal solution of N particle in L+1 generation and therefrom select maximum adaptation angle value pair The optimal solution answered is as the globally optimal solution in L+1 generation

Step 12, L+1 is assigned to L, it is judged that L ＜ L_maxWhether setting up, if setting up, then performing step 8；Otherwise, expression completes L_maxSecondary iteration, and obtain globally optimal solutionWith globally optimal solutionCorresponding scheduling scheme is as optimal scheduling Scheme.

Production scheduling method based on particle group optimizing the most according to claim 1, it is characterised in that described step 5 is created During building initial population, N-1 particle is to be produced by random fashion, and a remaining particle is to produce as follows Raw:

Step 5.1, orderRepresent i-th equipment M_iUpper q-th criticizes b_qCompletion date, n_iRepresent i-th equipment M_iUpper processing Batch quantity, A_iIt is i-th equipment M_iFree time；Initializen_i=1, q=1, A_i=0；

Step 5.2, judge whether q ＞ l sets up, if setting up, then it represents that generate a particle；Otherwise, step 5.3 is performed；

Step 5.3, utilize formula (1) obtain q-th criticize b_qCompletion date on i-th equipmentThus obtain q-th and criticize b_qCompletion date on m platform equipment:

$C_{b_q}\&lsqb;i\&rsqb;=max\{t_i\&times;n_i,A_i\}+\frac{p^{\left(q\right)}}{v_i}---\left(1\right)$

In formula (1), max{x, y} represent the greater taking in x and y；

Step 5.4, criticize b from q-th_qCompletion date on m platform equipment selects the equipment corresponding to minimum makespan, note For equipment min, by q-th batch b_qTransport is processed on equipment min；Again by n_min+ 1 is assigned to n_min,Assignment To A_min, after q+1 is assigned to q, perform step 5.2.

Production scheduling method based on particle group optimizing the most according to claim 1, it is characterised in that described step 8 is Carry out as follows:

Step 8.1, by the initial position of the kth particle in L generationIt is assigned to Wherein,For withThere is the variable of identical meanings, compareWithThe value of middle corresponding element is the most identical, if phase Same, then willThe value of middle corresponding element is set to 0, if it is different, then willMiddle corresponding element value is set toMiddle corresponding element The value of element；Thus obtain renewalUpdateIn the value of each element represent the sequence number of equipment；And each element With each batch of one_to_one corresponding in batch processing set B；

Step 8.2, by updateCriticizing on the equipment being dispatched to successively representated by non-zero value corresponding to the element of middle non-zero value It is processed, and calculates the completion date of relevant device；Thus obtain equipment completion date set C '=C ' [1] ..., C ' [i] ..., C ' [m] } and batch size set n '={ n '₁,…,n′_i,…,n′_m}；C ' [i] represents i-th equipment M_iCompletion Time, n '_iRepresent i-th equipment M_iThe quantity criticized of upper processing；

Step 8.3, by n '_i+ 1 is assigned to n '_i, C ' [i] is assigned to A '_i, wherein, variables A '_iWith A_iThere is identical meanings；

Step 8.4, by updateMiddle all values is the batch corresponding to the element of 0, arranges by batch processing time non-increasing Obtain crowd set B '={ b '₁,…,b′_q′,…,b′_l′, wherein, l ' represents renewalIntermediate value is total number of the element of 0, The most unappropriated batch total；

Step 8.5, make q '=1；

If step 8.6 q ' is ＞ l ', represent renewalAgain update, and perform step 8.9；Otherwise, hold Row step 8.7；

Step 8.7, formula (2) is utilized to obtain q ' individual batch b '_q′Completion date on i-th equipmentThus obtain Q ' individual batch b '_q′Completion date on m platform equipment:

$C_{b_{q^{\&prime;}}^{\&prime;}}\&lsqb;i\&rsqb;=max\{t_i\&times;n_i^{\&prime;},A_i^{\&prime;}\}+\frac{p^{\left(q^{\&prime;}\right)}}{v_i}---\left(2\right)$

Step 8.8, from q ' individual batch b '_q′Completion date on m platform equipment selects setting corresponding to minimum makespan Standby, it is designated as equipment min ', by q ' individual batch b '_q′Transport is processed on equipment min '；Again by n '_min′+ 1 is assigned to n′_min′,It is assigned to A '_min′, q '+1 is assigned to q ', performs step 8.6；

Step 8.9, by bernoulli distribution obtain l dimension group R being made up of 0 and 1₁, array R₁In each element with again Secondary renewalIn each element one_to_one corresponding；If R₁In element be 1, the most again updateMiddle corresponding element Value keeps constant, if R₁In element be 0, the most again updateThe value of middle corresponding element is set to 0；

Step 8.10, by again updateAs updateAnd substitute into execution in step 8.2-step 8.8, thus obtain After obtaining third time renewal

Step 8.11, generalAsWillAsAnd substitute into execution in step 8.1-step 8.10, thus obtain After obtaining third time renewal

Step 8.12, will third time update afterAfter updating with third timeWith probability P_c1Carry out bernoulli intersection, The position of the kth particle in the L generation after being intersected

Step 8.13, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the speed of the kth particle in L+1 generation

Step 8.14, generalWithWith probability P_c1Carry out bernoulli intersection, obtain the position of the kth particle in L+1 generation

Step 8.15, generalWith globally optimal solutionWith probability P_c2Carry out bernoulli intersection, it is thus achieved that the grain after two intersections Son, relatively prechiasmal particleWith intersect after the fitness of two particles, using particle big for fitness as L+1 The kth particle in generationPosition；

Step 8.16, calculate the L+1 fitness for particle each in population, fitness value is minimum The individual particle particle randomly generated replaces,Represent the maximum integer less than x；Thus update the L+1 position for population Put.

Production scheduling method based on particle group optimizing the most according to claim 1, it is characterised in that described step 9 is It is adjusted as follows:

Step 9.1, defined variable f and h, the greatest length of definition Local Search is f_max；Make f=1, i=1, h=m；

Step 9.2, judge f ＜ f_maxWhether set up, if setting up, then m platform equipment is arranged by the non-increasing of completion date, hold Row step 9.3；Otherwise, L+1 is completed for particle position N number of in populationAdjustment；

Step 9.3, i-th equipment M of selection_i；Select h platform equipment M_h；

Step 9.4, judging whether h ＞ 1 sets up, if setting up, then performing step 9.5；Otherwise, L+1 is completed for N number of in population Particle positionAdjustment；

Step 9.5, find any α crowd of b_α, b_α∈M_i；And any β crowd b_β, b_β∈M_h；If α crowd b_αWith β Individual crowd of b_βMeet formula (3), then exchange α crowd b_αWith β crowd b_β, and perform step 9.6；Otherwise, step 9.7 is performed:

$\left\{\begin{array}{c}0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{v_h}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_h}<C\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\\ 0<\frac{p^{\left(\mathrm{\&alpha;}\right)}}{v_i}-\frac{p^{\left(\mathrm{\&beta;}\right)}}{v_i}<C\&lsqb;i\&rsqb;-C\&lsqb;h\&rsqb;\end{array}\right.---\left(3\right)$

In formula (3), p^(α)Represent α crowd b_αProcess time, C [i] represents i-th equipment M_iCompletion date, p^(β)Represent the β crowd b_βProcess time, C [h] represents h platform equipment M_hCompletion date,

Step 9.6, by be assigned on same equipment batch by non-increasing arrangement batch process time, and calculate the completion of individual device Time, f+1 is assigned to f, performs step 9.2；

Step 9.7, h-1 is assigned to h after, perform step 9.4.