CN105320105B - A kind of parallel batch processing machines Optimization Scheduling - Google Patents
A kind of parallel batch processing machines Optimization Scheduling Download PDFInfo
- Publication number
- CN105320105B CN105320105B CN201410380575.0A CN201410380575A CN105320105B CN 105320105 B CN105320105 B CN 105320105B CN 201410380575 A CN201410380575 A CN 201410380575A CN 105320105 B CN105320105 B CN 105320105B
- Authority
- CN
- China
- Prior art keywords
- workpiece
- mrow
- batch
- msub
- scheduling
- 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.)
- Expired - Fee Related
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- General Factory Administration (AREA)
Abstract
The present invention is a kind of parallel batch processing machines Optimization Scheduling, the characteristics of for criticizing scheduling problem, establish the probabilistic model based on workpiece batching, devise corresponding personal sampling and probability updating method, and the generation of compact genetic algorithm population and probability updating mechanism are improved, and then propose a kind of intelligent algorithm of new solution batch scheduling problem.On the premise of enterprise's batch processing machines batch start production requirement is met, while accelerating production, the production cycle is shortened.By flexible parameter setting, different batch scheduling problems can be solved.Unit multimachine, arrival time it is identical arrival time is different, workpiece size it is identical size is different, batch processing place capacity it is limited the different situations such as unlimited can flexibly be combined, with very strong versatility.
Description
Technical field
The invention belongs to the production schedule in scheduling field, specifically for solidifying in semiconductor assembly and test production line
A kind of parallel batch processing machines Optimization Scheduling of the scheduling scheduling of process.
Background technology
The height of production management degree decides the direct operation benefits of an enterprise, and it mainly includes three phases:It is raw
Production plan, production scheduling and production control.Wherein production scheduling is the key link in production management field, and it accepts production meter
The rough arrangement drawn, while the control to workshop gives more detailed scheduling of production.With automated manufacturing information journey
The raising of degree, production scheduling also gradually switchs to Automatic dispatching by manual dispatching once.Automatic dispatching be in general all
Carried out on the basis of practical production status monitoring, its for manual dispatching, with more comprehensively, more effective information comes
Source, can equally make more reasonable effective decision scheme.The present invention is aiming at batch this universal industrial production of production
A kind of new solution that mode is proposed.
It is the optimization problem that the class risen in early 1990s has very strong application background, its basic assumption to criticize scheduling
It is that process equipment can be by multiple workpiece while being processed.Batch processing machines have a wide range of applications in industry is manufactured,
Pre-burning such as the heat treatment in metal-processing industry, multilift lock scheduling, port handling, semiconductor integrated circuit production is grasped
Make.
The main enlightening formula algorithm of Scheduling Problem method and intelligent algorithm are criticized at present.Scheduling based on heuristic rule
Scheme lacks the adaptive ability to changing environment, and the performance quality of algorithm has very big dependence to context and lessons.Base
In the weakness of traditional optimization, and the complexity of Job Shop Scheduling itself is criticized, swarm intelligence algorithm turns into one of research
Focus.Chinese University of Science and Technology's journey Aug. 1st, Shao Hao, king's bolt lion, Xu Rui et al. enter for solving the swarm intelligence algorithm of batch scheduling problem
Go systematic research, it is proposed that solve the swarm intelligence algorithm of multi-form batch scheduling problem, including ant group algorithm, population are calculated
Method etc., tentatively embodies colony intelligence and calculates the superiority for solving production scheduling problems in such applications.But these algorithms remain unchanged
Existing defects, a class can only targetedly be solved the problems, such as by being mainly reflected in every kind of algorithm, it is impossible to meet manufacturing enterprise's variation batch
The demand of form processing.
The content of the invention
It is an object of the invention to provide a kind of parallel batch processing machines Optimization Scheduling, by flexible parameter setting with
And embedded intelligent algorithm can solve different types of batch of scheduling problem, determine the batching result of workpiece to be processed, each batch
Process sequence and the key issue such as process equipment distribution, meeting the requirement of enterprise large scale equipment batch start production
Under the premise of, while improving production efficiency, shorten the production cycle.
The technical scheme that is used to achieve the above object of the present invention is:A kind of parallel batch processing machines Optimized Operation side
Method, comprises the following steps:
Step 1:Task-set to be processed is initialized as to the set of all processing tasks by way of this procedure, and will be to be added
Processing tasks in work task-set are ranked up according to the sequencing of preceding working procedure end time;By at the beginning of optional free device collection
Beginning turns to the set of all process equipments of this procedure;
Step 2:The sampling probability represented with matrix a model A is set up respectively for each workpiece group, and by such as lower section
Formula is initialized:
Wherein, A (i, k) is a sampling probability model A element, represents workpiece jiIt is assigned to PkIn probability;Maximum batch is represented,It is to round symbol, n is the number of the workpiece group workpiece, L is the minimum of process equipment
Batch volume;
Step 3:N kind batch scheduling schemes are generated according to sampling probability model A, the flat of N kind batch scheduling schemes is calculated respectively
The equal process-cycle, and N kind batch scheduling schemes sort from small to large by the average process-cycle, (x × N) (0 < before taking out in sequence
X < 1) batch scheduling scheme as dominant group, x is the ratio being manually set;
Step 4:All workpiece groups are carried out with the related probabilistic model Z of workpiece according to dominant group to initialize, the probability
Model Z is n × n lower triangular matrix, and row and column represents workpiece, and n is the number of the workpiece group workpiece, its element Zi,j
From the probabilistic relation for numerically having reacted workpiece batching, be worth more big then i-th of workpiece and j-th of workpiece in same batch can
Energy property is bigger;
Step 5:The sequencing and probabilistic model Z reached in each workpiece group according to workpiece is sampled respectively, makes institute
There is workpiece to complete batch operation and select corresponding process equipment by rule;Generated respectively in iterative process in batches each time
P batch scheduling schemes, select the minimum batch scheduling scheme of the average period of production and retain to next iteration;
Step 6:Crowd scheduling scheme x finally won in each iterative process is selected as advantage individual, using increment
The method of habit is updated to probabilistic model Z:
Step 7:Continuous repeat step 5 and step 6, set maximum iteration as end condition, obtain final
Criticize scheduling scheme.
The sampling probability model A is n × Max matrix, wherein row represents processing tasks, row represent batch.
It is described that N kind batch scheduling schemes are generated according to sampling probability model A, comprise the following steps:
Each workpiece group is sampled according to the sampling probability model A sequencings reached according to workpiece:For a certain
The workpiece of determination, selects the batch where the workpiece, until in batches all workpiece are completed;
After the result in batches of all workpiece groups is obtained, the arrival time the latest of each batch is calculated;
According to the priority of arrival time the latest, each batch is set to select process equipment in order.
It is described that each workpiece group is sampled according to the sampling probability model A sequencings reached according to workpiece, specifically
For:
For the workpiece j of first arrival in workpiece group1, in batch (Pk,1,Pk,2,...Pk,Max) in optional one batch
It is secondary;
For workpiece ji, i ≠ 1, according to roulette method choice and workpiece (j1,j2...ji-1) in some workpiece it is same
One batch, or in batch (Pk,1,Pk,2,...Pk,Max) select one and workpiece (j1,j2...ji-1) selected by batch it is different and still
The batch of non-granted full, if there is no such batch, then the batch for randomly choosing a not yet granted full is processed;
For a workpiece group k batch Pk,i, 1≤i≤Max, its capacity has been maxed out capacity S, then sampling is general
Kth row all elements in rate model A are set to 0, and remaining each row is normalized, so that remaining each row of probabilistic model A
Every a line probability and for 1.
Carried out for multiple not compatible workpiece groups in batches in scheduling process, independently set up sampling probability model
And probabilistic model, make the in batches separate of each workpiece group.After all workpiece groups are completed in batches, it will integrate in batches
To be processed the selection and process sequence determination of equipment by the arrival time sequence the latest of the workpiece of each batch.
The priority of basis arrival time the latest, makes each batch select process equipment in order, enters according to following rule
OK:
According to present lot Pk,pEarliest permission process time on each boardWith the processing on each board
TimePass throughSelection makesThe minimum equipment m of value is used as batch Pk,pPlus
Construction equipment, more new lot Pk,pMachine time and equipment m release time;
The earliest permission process timeThe deadline t of last consignment of workpiece is processed for each boardmWith this batch of workpiece
In arrival time max (R the latestk,i) in maximum;It is batch Pk,iProcess time in m platform equipment.
Described pair of all workpiece groups carry out the related probabilistic model Z of workpiece initialization, comprise the following steps:
For the workpiece j of a certain workpiece groupi, the workpiece (j in dominant group is counted respectively1,j2...ji-1) and workpiece ji
With a batch of number (c1,c2...ci-1), and by (c1/(x×N),c2/(x×N)...ci-1/ (x × N)) (Z is assigned to respectively
(i, 1), Z (i, 2) ... Z (i, i-1)), represent workpiece jiWith (j1,j2...ji-1) respectively with a batch of probability;Statistics is excellent
Workpiece j in gesture colonyiWith (j1,j2...ji-1) not with a batch of number ci, and by ci/ (x × N) is assigned to Z (i, i), table
Show workpiece jiWith (j1,j2...ji-1) not with a batch of probability.To probabilistic model Z normalizing after the assignment to Z is completed
Change is handled, and is specially:
Wherein, SumiFor the sum of the i-th row in probabilistic model Z.
The method of the use incremental learning is updated to probabilistic model Z, is specially:
Wherein, l is iterative algebra, and n is workpiece (j1,j2...ji-1) and jiWith a batch of number of parts, if (j1,
j2...ji-1) and (j1,j2...ji-1) different crowdes then n=1.β ∈ (0,1) learning rate;Li,j(l) it is the letter that is defined as below
Number:
The present invention has advantages below and beneficial effect:
(1) on the premise of enterprise's batch processing machines batch start production requirement is met, while accelerating production, shorten
Production cycle.
(2) method can solve different batch scheduling problems by flexible parameter setting.Unit multimachine, arrival time
It is identical arrival time is different, workpiece size it is identical size is different, batch processing machines finite capacity the different situations such as unlimited all may be used
To be flexibly combined, with very strong versatility.
(3) improved compact genetic algorithm is employed, relative to rule-based batch dispatching method parameter is less, optimization effect
Fruit is more preferably.
(4) defect existed for cGA (compact Genetic Algorithm) algorithms when solving challenge,
Following improvement has been done to cGA algorithms on the basis of holding algorithm is simple clearly:
A increases population scale, still produces two individuals, and relatively more individual fitness value by probability matrix every time, protects
Stay advantage individual, give up inferior position individual then produced again by probabilistic model one newly it is individual compared with original advantage individual,
Reservation advantage individual, giving up inferior position individual, the rest may be inferred, until the individual of generation meets the population scale np of setting, takes out excellent
Gesture individual probabilistic model is updated so processing ensure that it is each it is alternative come update probabilistic model individual be all one
Compare outstanding individual (rather than excellent individual in two), be conducive to probabilistic model along being correctly oriented convergence.
B retains the defect individual in every generation, participates in follow-on comparison.Standard cGA algorithms can produce two every time
New individual, so processing make it that excellent individual and average individual are identicals to the convergent contribution degree of probabilistic model, can cause
The concussion of algorithm, while can not ensure that the individual that algorithm is finally provided is the best individual encountered in search procedure so knot
The advantage of traditional genetic algorithm is closed, retains the triumph individual per a generation, participates in follow-on competition.
Brief description of the drawings
Fig. 1 is improvement cGA algorithm flow charts used in the present invention;
Fig. 2 is initialization sampling probability illustraton of model;
Fig. 3 is that the workpiece obtained using the present invention is schemed in batches;
Fig. 4 is that rule-based obtained workpiece is schemed in batches.
Embodiment
Below in conjunction with the accompanying drawings and embodiment the present invention is described in further detail.
Improved compact genetic algorithm flow chart used in the present invention is as shown in Figure 1.Manufactured with semiconductor rear section
It is illustrated in journey exemplified by curing process, Fig. 2 is initialization sampling probability illustraton of model, Fig. 3 is to be obtained using the present invention
Workpiece scheme in batches, Fig. 4 is first processed the obtained workpiece of rule and schemed in batches based on arriving first.
The implementation process of the present invention is explained below in conjunction with technical scheme example and accompanying drawing.
The preceding working procedure that table 1 depicts curing process in semiconductor rear section manufacturing process terminates process time and solidification
The data such as process process time, Lot types, process equipment minimum lot size and maximum batch.
Table 1
According to the data message in table 1, the implementation process to whole dispatching method is described.Preceding working procedure and this road work
The intermediate transportation time between sequence is without considering.
Algorithm performs process carries out as follows:
1) it is all processing operations after this procedure, a shared Liang Ge working groups, work to initialize task-set to be processed
Part sum N=14;The equipment of all procedures of initialization is concentrated to optional free device, and sum is 2, and batch processing machines are minimum
Manufacturing batch is 2, and maximum manufacturing batch is 4.
2) task type 1 and task type 2 have 7 workpiece respectively, and maximum batch is:
Each workpiece group at most has 4 batches, and initialization sampling probability model is shown in accompanying drawing 2.
3) 20 kinds of solutions are generated according to sampling probability model, the average production week of 20 kinds of solutions is calculated respectively
Phase.A solution is produced by sampling probability model with next example introduction and the average production of the solution is calculated
Cycle:
For workpiece group 1, produce the random number between 70 to 1 respectively, 7 workpiece of correspondence, as a result for (0.1,0.3,0.9,
0.56,0.7,0.6,0.73, then will produce 4 batches for workpiece group 1 is respectively:Batch 1.1 (workpiece 1.1), batch 1.2
(workpiece 1.2), batch 1.3 (workpiece 1.4,1.5,1.6,1.7), batch 1.4 (workpiece 1.3).
For workpiece group 2, produce the random number between 70 to 1 respectively, 7 workpiece of correspondence, as a result for (0.3,0.43,
0.72,0.8,0.75,0.22,0.73, then will produce 4 batches for workpiece group 1 is respectively:Batch 2.1 (workpiece 2.6),
Batch 2.2 (workpiece 2.1, workpiece 2.2), batch 2.3 (workpiece 2.3,2.5,2.7), batch 2.4 (workpiece 2.3).
By the way that 8 batches can be obtained in batches, the workpiece time reached the latest in each batch is calculated:Batch 1.1 is 0,
Batch 1.2 is 1, and batch 1.3 is 4, and batch 1.4 is 1, and batch 2.1 is 3, and batch 2.2 is 1, and batch 2.3 is 4, and batch 2.4 is 2.
By workpiece, arrival time is ranked up to this 8 batches and obtains sequence the latest:Batch 1.1, batch 1.2, batch 2.2, batch
1.4, batch 2.4, batch 2.1, batch 1.3, batch 2.3.Select equipment to process respectively by this sequence, 8 batches can be obtained
Corresponding completion date is:
Batch 1.1 is 3, and batch 1.2 is 4, and batch 2.2 is 7, and batch 1.4 is 7, and batch 2.4 is 11, and batch 2.1 is 11,
Batch 1.3 is 14, and batch 2.3 is 15.It is 10.9 that the average period of production, which can be obtained,.
4) individual (each individual represents a kind of feasible batch scheduling scheme) is respectively obtained by above-mentioned steps, obtained every
The individual average process-cycle, and individual sorts from small to large by the average process-cycle, before taking out in sequence (x × N)
(0 < x < 1) individual is used as advantage individual BetterGroup.In the present embodiment, x takes 25%, that is, selects 20 kinds of solutions
The minimum 5 excellent alternatives generation dominant group BetterGroup of the middle average period of production.
By dominant group BetterGroup, we need to complete the initialization of the related probabilistic model Z of workpiece in algorithm.
For the probabilistic model Z of workpiece group k algorithmskIt is the lower triangular matrix of a n × n (n is the number of the workpiece group workpiece), wherein
Row and column represents workpiece, and workpiece is obtained into workpiece sequence for q by the sequencing of arrival time:(j1,j2...jn),Zi,j(0
≤ i≤n, 0 < j < i) represent that i-th of workpiece and j-th of workpiece are in the probability of same batch, Z in sampling processi,j(0≤
I≤n, j=i) represent to come i in workpiece and q before workpiece (0,1,2 ... i-1) not with a batch of probability.Matrix
Z is from the probabilistic relation for numerically having reacted workpiece batching, Zi,jBigger i-th of workpiece and j-th of workpiece in same batch can
Energy property is bigger.ZkInitial method be:
For workpiece j in workpiece group 1,2i, the workpiece (j in dominant group BetterGroup is counted respectively1,j2...ji-1)
With workpiece jiWith a batch of number (c1,c2...ci-1), and by (c1/(0.25×N),c2/(0.25×N)...ci-1/
(0.25 × N)) it is assigned to respectively (Z (i, 1), Z (i, 2) ... Z (I, i-1)), represent workpiece j1With (j1,j2...ji-1) respectively same
Probability in one batch;Statistics workpiece (j in dominant group BetterGroup1,j2...ji-1) and (j1,j2...ji-1) do not exist
With a batch of number cj, and by cj/ (0.25 × N) is assigned to Z (j, j), represents workpiece jiWith (j1,j2...ji-1) not same
The probability of batch.Complete after initialization assignment, each row of probabilistic model is normalized, the probability mould initialized
Type.
Assuming that through counting in 5 excellent alternatives, workpiece 1.5 and workpiece 1.1 are in being 2 with batch of number of times, with workpiece 1.2
In being 1 with batch of number of times, it is 3 to be in workpiece 1.3 with batch of number of times, is in workpiece 1.4 and with batch of number of times is
0, it is 3 to be in workpiece 1.6, workpiece 1.7 with batch of number of times.Then corresponding element is respectively in matrix ZIt is 2+1 to be wherein equal to each element denominator
+ 3+0+3 is 9.
5) sequencing reached in each compatible workpiece group according to workpiece is sampled.For first arrival in workpiece group
Workpiece j1Can be in (Pk,1,Pk,2,...Pk,4) in an optional batch.For workpiece ji(i ≠ 1) can be according to wheel disc bet method
Selection and (j1,j2...ji-1) in some workpiece processed or in (P in same batchk,1,Pk,2,...Pk,4) selection one
With (j1,j2...ji-1) difference is criticized and the batch of not yet granted full is processed, if there is no such batch, then random choosing
The batch for selecting a not yet granted full is processed.For the batch of granted full, due to new workpiece can not be added, so will still
Unsampled workpiece is set to 0 with workpiece probability in same batch in the batch, and normalizes again.By that analogy until
It is all to complete batch operation.Generate np individual respectively in iterative process each time, select optimum individual and retain under
A generation.
Assuming that workpiece is workpiece 1.1 in batch 1.1, workpiece 1.3, the workpiece in batch 1.2 is the work in 1.4, batch 1.3
Part is 1.2, and the 5th row element is respectively in probabilistic model: For workpiece 1.5, the random number r between one 0 to 1 is produced, ifThen workpiece is dispensed on batch
1.1 in;IfThen workpiece is dispensed in batch 1.3;IfThen workpiece is dispensed on batch 1.1
In;IfThen workpiece is dispensed in batch 1.4.
6) the selection individual x that per generation finally wins is as advantage individual, using the method update probability of following incremental learning
Model:
Wherein, l is iterative algebra, and n is workpiece (j1,j2...ji-1) and jiWith a batch of number pieces, if (j1,
j2...ji-1) and (j1,j2...ji-1) different crowdes then n=1.β ∈ (0,1) learning rate;Li,j(l) it is the letter that is defined as below
Number:
Wherein, l is iterative algebra, and n is workpiece (j1,j2...ji-1) and jiWith a batch of number of parts, if (j1,
j2...ji-1) and (j1,j2...ji-1) different crowdes then n=1.β ∈ (0,1) learning rate;Li,j(l) it is the letter that is defined as below
Number:
Can prove by formula (5) and (6) it is updated probability matrix per a line and still be 1.
Illustrate model renewal process by taking workpiece 1.5 in workpiece group 1 as an example below:
Assuming that workpiece 1.5 and workpiece 1.1 in the outstanding solution remained, 1.2 points of workpiece is in same batch, β
ForThe 5th row element is respectively in current probability model: Then the probabilistic model after renewal is:
7) 5 and operation 6 are constantly repeated, until number of repetition is more than or equal to the maximum iteration that algorithm is set, are obtained
To result in batches as shown in Figure 3, the process time of each task and process equipment difference are as shown in table 2, are obtained using of the invention
The average period of production arrived is 6.71, and the rule-based obtained average period of production is 7, by contrast it can be found that using this
Invention can effectively reduce the average period of production of workpiece.
Table 2
Claims (7)
1. a kind of parallel batch processing machines Optimization Scheduling, it is characterised in that comprise the following steps:
Step 1:Task-set to be processed is initialized as to the set of all processing tasks by way of this procedure, and by be processed
The processing tasks that business is concentrated are ranked up according to the sequencing of preceding working procedure end time;By the initialization of optional free device collection
For the set of all process equipments of this procedure;
Step 2:The sampling probability represented with matrix a model A is set up respectively for each workpiece group, and as follows just
Beginningization:
<mrow>
<mi>A</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>M</mi>
<mi>a</mi>
<mi>x</mi>
</mrow>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, A (i, k) is a sampling probability model A element, represents workpiece jiIt is assigned to PkIn probability;
Maximum batch is represented,It is to round symbol, n is the number of the workpiece group workpiece, L is the minimum batch volume of process equipment;It is described
Sampling probability model A is n × Max matrix, wherein row represents processing tasks, row represent batch;
Step 3:N kind batch scheduling schemes are generated according to sampling probability model A, average the adding of N kind batch scheduling schemes is calculated respectively
The work cycle, and N kind batch scheduling schemes sort from small to large by the average process-cycle, (x × N) (0 < x < before taking out in sequence
1) batch scheduling scheme is as dominant group, and x is the ratio being manually set;
Step 4:All workpiece groups are carried out with the related probabilistic model Z of workpiece according to dominant group to initialize, the probabilistic model Z
It is n × n lower triangular matrix, row and column represents workpiece, and n is the number of the workpiece group workpiece, its element Zi,jFrom numerical value
On reacted the probabilistic relation of workpiece batching, be worth the more big possibility of then i-th of workpiece and j-th of workpiece in same batch just
It is bigger;
Step 5:The sequencing and probabilistic model Z reached in each workpiece group according to workpiece is sampled respectively, makes all works
Part completes batch operation and selects corresponding process equipment by rule;Generating p in iterative process respectively in batches each time
Scheduling scheme is criticized, the minimum batch scheduling scheme of the average period of production is selected and retains to next iteration;
Step 6:Crowd scheduling scheme x finally won in each iterative process is selected as advantage individual, using incremental learning
Method is updated to probabilistic model Z:
Step 7:Continuous repeat step 5 and step 6, set maximum iteration as end condition, obtain final batch tune
Degree scheme.
2. a kind of parallel batch processing machines Optimization Scheduling according to claim 1, it is characterised in that the basis is taken out
Sample probabilistic model A generation N kind batch scheduling schemes, comprise the following steps:
Each workpiece group is sampled according to the sampling probability model A sequencings reached according to workpiece:Determined for a certain
Workpiece, the batch where the workpiece is selected, until in batches all workpiece are completed;
After the result in batches of all workpiece groups is obtained, the arrival time the latest of each batch is calculated;
According to the priority of arrival time the latest, each batch is set to select process equipment in order.
3. a kind of parallel batch processing machines Optimization Scheduling according to claim 2, it is characterised in that described to each
Workpiece group is sampled according to the sampling probability model A sequencings reached according to workpiece, is specially:
For the workpiece j of first arrival in workpiece group1, in batch (Pk,1,Pk,2,...Pk,Max) in an optional batch;
For workpiece ji, i ≠ 1, according to roulette method choice and workpiece (j1,j2...ji-1) in some workpiece it is same a collection of
It is secondary, or in batch (Pk,1,Pk,2,...Pk,Max) select one and workpiece (j1,j2...ji-1) selected by batch it is different and not yet full
The batch criticized, if there is no such batch, then the batch for randomly choosing a not yet granted full is processed;
For a workpiece group k batch Pk,i, 1≤i≤Max, its capacity has been maxed out capacity S, then by sampling probability mould
Kth row all elements in type A are set to 0, and remaining each row is normalized, so that remaining each row of probabilistic model A is every
A line probability and for 1.
4. a kind of parallel batch processing machines Optimization Scheduling according to claim 2, it is characterised in that for it is multiple not
Compatible workpiece group is being carried out in batches in scheduling process, independently set up sampling probability model and probabilistic model, make each
Workpiece group it is in batches separate;After all workpiece groups are completed in batches, the workpiece by each batch will be integrated in batches
Arrival time sequence the latest is processed the selection of equipment and process sequence is determined.
5. a kind of parallel batch processing machines Optimization Scheduling according to claim 2, it is characterised in that the basis is most
The priority of late arrival time, makes each batch select process equipment in order, is carried out according to following rule:
According to present lot Pk,pEarliest permission process time on each boardWith the process time on each boardPass throughSelection makesThe minimum equipment m of value is used as batch Pk,pProcessing set
It is standby, more new lot Pk,pMachine time and equipment m release time;
The earliest permission process timeThe deadline t of last consignment of workpiece is processed for each boardmWith in this batch of workpiece most
Late arrival time max (Rk,i) in maximum;It is batch Pk,iProcess time in m platform equipment.
6. a kind of parallel batch processing machines Optimization Scheduling according to claim 1, it is characterised in that described pair is owned
Workpiece group carries out the related probabilistic model Z of workpiece initialization, comprises the following steps:
For the workpiece j of a certain workpiece groupi, the workpiece (j in dominant group is counted respectively1,j2...ji-1) and workpiece jiSame
Number (the c of batch1,c2...ci-1), and by (c1/(x×N),c2/(x×N)...ci-1/ (x × N)) be assigned to respectively (Z (i, 1),
Z (i, 2) ... Z (i, i-1)), represent workpiece jiWith (j1,j2...ji-1) respectively with a batch of probability;Statistics is in dominant group
Middle workpiece jiWith (j1,j2...ji-1) not with a batch of number ci, and by ci/ (x × N) is assigned to Z (i, i), represents workpiece
jiWith (j1,j2...ji-1) not with a batch of probability;At normalization after the assignment to Z is completed to probabilistic model Z
Reason, be specially:
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Sum</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mi>n</mi>
</mrow>
</munderover>
<msub>
<mi>Z</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo><</mo>
<mi>i</mi>
<mo><</mo>
<mi>n</mi>
<mo>,</mo>
<mn>1</mn>
<mo><</mo>
<mi>j</mi>
<mo><</mo>
<mi>n</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Z</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>Z</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msub>
<mrow>
<msub>
<mi>Sum</mi>
<mi>i</mi>
</msub>
</mrow>
</mfrac>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mn>0</mn>
<mo><</mo>
<mi>i</mi>
<mo><</mo>
<mi>n</mi>
<mo>,</mo>
<mn>1</mn>
<mo><</mo>
<mi>j</mi>
<mo><</mo>
<mi>n</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, SumiFor the sum of the i-th row in probabilistic model Z.
7. a kind of parallel batch processing machines Optimization Scheduling according to claim 1, it is characterised in that described using increasing
The method of amount study is updated to probabilistic model Z, is specially:
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>Z</mi>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>-</mo>
<mi>&beta;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>Z</mi>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mfrac>
<mi>&beta;</mi>
<mi>n</mi>
</mfrac>
<msub>
<mi>L</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>&le;</mo>
<mi>j</mi>
<mo>&le;</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, l is iterative algebra, and n is workpiece (j1,j2...ji-1) and jiWith a batch of number of parts, if (j1,
j2...ji-1) and (j1,j2...ji-1) different crowdes then n=1, β ∈ (0,1) learning rate;Li,j(l) it is the letter that is defined as below
Number:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410380575.0A CN105320105B (en) | 2014-08-04 | 2014-08-04 | A kind of parallel batch processing machines Optimization Scheduling |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410380575.0A CN105320105B (en) | 2014-08-04 | 2014-08-04 | A kind of parallel batch processing machines Optimization Scheduling |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105320105A CN105320105A (en) | 2016-02-10 |
CN105320105B true CN105320105B (en) | 2017-09-29 |
Family
ID=55247690
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410380575.0A Expired - Fee Related CN105320105B (en) | 2014-08-04 | 2014-08-04 | A kind of parallel batch processing machines Optimization Scheduling |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105320105B (en) |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106611236A (en) * | 2016-04-27 | 2017-05-03 | 四川用联信息技术有限公司 | Production scheduling algorithm for solving job shop scheduling problem |
US10783469B2 (en) * | 2016-11-01 | 2020-09-22 | Hitachi, Ltd. | Production support system using multiple plan candidates and different types of learning devices |
CN107024867B (en) * | 2017-06-01 | 2019-10-25 | 合肥工业大学 | A kind of optimal control method for the compatible workpiece race considering forward sight distance |
CN107102552B (en) * | 2017-06-12 | 2018-03-09 | 合肥工业大学 | Gather the parallel machine dispatching method and system for leapfroging and becoming neighborhood processing based on mixing |
CN108171372B (en) * | 2017-12-25 | 2022-06-14 | 上海大学 | Multi-variety production batch scheduling method with preparation time |
CN108776862B (en) * | 2018-05-02 | 2021-08-10 | 华南理工大学 | Intelligent scheduling method supporting splitting of process task quantity |
CN109709916B (en) * | 2018-12-20 | 2019-09-20 | 宁波大学 | A kind of dispatching method based on Gibbs sampling method |
CN111160711B (en) * | 2019-12-06 | 2022-09-16 | 安徽大学 | Parallel machine batch scheduling method based on ant colony algorithm |
CN111007821B (en) * | 2019-12-20 | 2020-12-29 | 韩山师范学院 | Workshop scheduling method with batch processing time being limited by total weight and maximum size |
CN111507641B (en) * | 2020-04-27 | 2024-04-16 | 上海华力集成电路制造有限公司 | Batch processing equipment scheduling method and device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB0422900D0 (en) * | 2003-10-24 | 2004-11-17 | Agilent Technologies Inc | A method and machine-readable medium for using matrices to automatically analze network events |
CN101231720A (en) * | 2008-02-01 | 2008-07-30 | 北京航空航天大学 | Enterprise process model multi-target parameter optimizing method based on genetic algorithm |
CN101334663A (en) * | 2008-06-11 | 2008-12-31 | 清华大学 | Machine group scheduling method based on process capacity characteristic model |
CN102566431A (en) * | 2012-01-16 | 2012-07-11 | 浙江工业大学 | Method for dispatching and controlling fuzzy parameter operation workshop under dynamic disturbance based on improved genetic algorithm |
CN102929263A (en) * | 2012-11-16 | 2013-02-13 | 北京理工大学 | Hybrid flow shop scheduling method |
CN103246240A (en) * | 2012-10-30 | 2013-08-14 | 中国科学院沈阳自动化研究所 | Batch preparation continuous single-processing prediction control scheduling method |
-
2014
- 2014-08-04 CN CN201410380575.0A patent/CN105320105B/en not_active Expired - Fee Related
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB0422900D0 (en) * | 2003-10-24 | 2004-11-17 | Agilent Technologies Inc | A method and machine-readable medium for using matrices to automatically analze network events |
CN101231720A (en) * | 2008-02-01 | 2008-07-30 | 北京航空航天大学 | Enterprise process model multi-target parameter optimizing method based on genetic algorithm |
CN101334663A (en) * | 2008-06-11 | 2008-12-31 | 清华大学 | Machine group scheduling method based on process capacity characteristic model |
CN102566431A (en) * | 2012-01-16 | 2012-07-11 | 浙江工业大学 | Method for dispatching and controlling fuzzy parameter operation workshop under dynamic disturbance based on improved genetic algorithm |
CN103246240A (en) * | 2012-10-30 | 2013-08-14 | 中国科学院沈阳自动化研究所 | Batch preparation continuous single-processing prediction control scheduling method |
CN102929263A (en) * | 2012-11-16 | 2013-02-13 | 北京理工大学 | Hybrid flow shop scheduling method |
Also Published As
Publication number | Publication date |
---|---|
CN105320105A (en) | 2016-02-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105320105B (en) | A kind of parallel batch processing machines Optimization Scheduling | |
Yuan et al. | Research on intelligent workshop resource scheduling method based on improved NSGA-II algorithm | |
Magalhaes-Mendes | A comparative study of crossover operators for genetic algorithms to solve the job shop scheduling problem | |
CN113792924A (en) | Single-piece job shop scheduling method based on Deep reinforcement learning of Deep Q-network | |
Liu et al. | Solving distributed and flexible job-shop scheduling problems for a real-world fastener manufacturer | |
CN108563200B (en) | Multi-target workpiece scheduling method and device based on ant colony algorithm | |
CN110501978B (en) | Scheduling method for scheduling robot product production workshop | |
CN104408528A (en) | Optimization scheduling method in raw material leaching process for chemical industry production | |
CN105843197A (en) | Teaching-and-learning-algorithm-based static scheduling optimization system for discrete manufacturing shop | |
CN105302078B (en) | The control system and method for boiler tube board operation in a kind of semiconductor fabrication process | |
CN108647859A (en) | The double population collaborative learning strategies of displacement assembly line and optimization method of Knowledge driving | |
CN113987936B (en) | Equipment test resource overall allocation method based on chaotic genetic algorithm | |
CN106611275A (en) | Production scheduling algorithm for solving job shop production problem | |
Huang et al. | An improved genetic algorithm for job-shop scheduling problem with process sequence flexibility | |
CN107807518A (en) | A kind of Multiobjective Optimal Operation method of Chemical Manufacture raw material still-process | |
CN109543921B (en) | Production scheduling method of special petroleum pipe flow shop based on improved genetic algorithm | |
CN110928261A (en) | Distributed estimation scheduling method and system for distributed heterogeneous flow shop | |
Qin et al. | Multi-objective discrete migratory bird optimizer for stochastic disassembly line balancing problem | |
CN104698838A (en) | Discourse domain based dynamic division and learning fuzzy scheduling rule mining method | |
CN116011723A (en) | Intelligent dispatching method and application of coking and coking mixed flow shop based on Harris eagle algorithm | |
CN114462770A (en) | Semiconductor production scheduling method, system and computer readable storage medium | |
CN103996080A (en) | Manufacturing system configuration optimization method for achieving the highest connectedness | |
Vimala Rani et al. | Performance evaluation of ATC based greedy heuristic algorithms in scheduling diffusion furnace in wafer fabrication | |
CN113723695A (en) | Scene-based remanufacturing scheduling optimization method | |
Moreno et al. | A constructive heuristic algorithm for concurrently selecting and sequencing jobs in an FMS environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170929 Termination date: 20180804 |