CN109523136A - A kind of scheduling knowledge management system towards intelligence manufacture - Google Patents
A kind of scheduling knowledge management system towards intelligence manufacture Download PDFInfo
- Publication number
- CN109523136A CN109523136A CN201811269262.2A CN201811269262A CN109523136A CN 109523136 A CN109523136 A CN 109523136A CN 201811269262 A CN201811269262 A CN 201811269262A CN 109523136 A CN109523136 A CN 109523136A
- Authority
- CN
- China
- Prior art keywords
- scheduling
- knowledge
- sample
- control
- scheduling knowledge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 81
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000011156 evaluation Methods 0.000 claims abstract description 14
- 238000009412 basement excavation Methods 0.000 claims abstract description 6
- 230000004044 response Effects 0.000 claims abstract description 4
- 238000007726 management method Methods 0.000 claims description 22
- 230000008569 process Effects 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 16
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000009826 distribution Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 241001106462 Ulmus Species 0.000 description 4
- 239000004065 semiconductor Substances 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000007418 data mining Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000004377 microelectronic Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 238000010845 search algorithm Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 206010021703 Indifference Diseases 0.000 description 1
- 241000131894 Lampyris noctiluca Species 0.000 description 1
- 241001114959 Ulmus procera Species 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 235000021050 feed intake Nutrition 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- 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
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Manufacturing & Machinery (AREA)
- General Factory Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The scheduling knowledge management system towards intelligence manufacture that the present invention relates to a kind of, comprising: scheduling knowledge study module is carried out knowledge excavation for sample for dispatching data, obtained scheduling knowledge by data-driven;Scheduling knowledge application module, the response when receiving scheduling request, the scheduling strategy to match for calling the scheduling knowledge to generate with intelligent workshop production status, and obtain scheduling result;Scheduling knowledge online evaluation module responds after obtaining scheduling result, for judging scheduling knowledge validity according to the scheduling result;Scheduling knowledge update module is responded when determining scheduling knowledge failure, for applying on-line study method, is adjusted on the basis of increasing more new samples to scheduling knowledge.Compared with prior art, the present invention can preferably improve the adaptability of scheduling knowledge, it is ensured that scheduling validity, and the advantage with sustainable Intelligent Optimal workshop overall performance.
Description
Technical field
The present invention relates to intelligence manufacture fields, more particularly, to a kind of scheduling knowledge management system towards intelligence manufacture.
Background technique
In intelligent workshop, due to the extensive use of technology of Internet of things and various information systems, such as Enterprise Resources Plan system
Unite (Enterprise Resource Planning, ERP), manufacturing execution system (Manufacturing Execution
System, MES), quality control system (Quality Control System, QCS) etc., store more and more data.
Information relevant to the production and operation is also contained in these data, is reflected the operating condition in intelligent workshop, is richly stored with
Production scheduling knowledge can extract scheduling knowledge relevant to workshop running using data mining technology from data, and can
It is applied in intelligent workshop Operation Decision, instructs workshop operation, improves the adaptivity in intelligent workshop.
A kind of Chinese patent " Dynamic Job-shop Scheduling rule intelligent selecting method of creation data driving " (patent disclosure
Number: CN 107767022A) invent a kind of Dynamic Job-shop Scheduling rule intelligent selection side based on creation data driving
Method, this method are based on Multi-Pass algorithm and establish job shop production scheduling emulation platform, generate the sample number of scheduled production scheduling
According to;The sample data of acquisition is screened, scheduling parameter collection is generated;It designs under different regulation goals for scheduling knowledge study
BP neural network model;And using a kind of training of improved glowworm swarm algorithm Optimized BP Neural Network model, NFA-BP is obtained
Model;By the NFA-BP model set under each regulation goal at an intelligent scheduling module, and carried out with job shop MES system
It is integrated, instruct on-line scheduling.The Shahzad A of French Nantes university et al. (" Discovering Dispatching
Rules For Job Shop Scheduling Problem Through Data Mining ", is shown in 8th
International Conference of Modeling and Simulation, 2010) it is directed to Job Shop dynamic dispatching
Problem proposes that a kind of scheduling rule Selection Framework based on data mining, the frame obtain optimization using tabu search algorithm
Scheduling data learn to obtain the scheduling rule of near-optimization by decision Tree algorithms on this basis.Chinese patent is " based on opinion
Domain dynamic divides and the fuzzy scheduling rule digging method of study " (patent publication No.: 104698838 A of CN), for group
The characteristics of criticizing the microelectronics production line scheduling problem of feature is proposed the new fuzzy scheduling rule format of one kind and is calculated based on Aprior
The fuzzy scheduling rule intelligent excavating algorithm of method, while devising a kind of harmonic search algorithm and is optimized to key parameter
It practises, can produce preferable scheduling in microelectronics production line scheduling problem of the mean transit time as regulation goal to minimize
Effect.
It is not difficult to find out that according to scheduling knowledge good dispatching effect can be obtained with Instructing manufacture.In manual dispatching, adjust
Spending decision can be with information such as goods in process inventory in the type of product, the operating load of workshop appliance, order delivery date, workshop
And dynamic change equally for the scheduling knowledge generated by machine learning, also there is same requirement.And in intelligent workshop, it is raw
Production is the demand dynamic change with market, and therefore, scheduling knowledge is also required to be dynamic, this also means that, if intelligent vehicle
Between state vary widely, the training sample for the scheduling knowledge being previously generated fails to cover current production status (including product
The factors such as type and equipment), then scheduling knowledge can not match current production environment, then, it is obtained by scheduling knowledge application module
The scheduling strategy obtained can not meet regulation goal, not can guarantee the Continuous optimization of workshop operation performance finally.For this reason, it may be necessary to right
Scheduling knowledge carries out the management of Life cycle, including, the generation of scheduling knowledge, the application of scheduling knowledge, scheduling knowledge are commented
Estimate four parts of update with scheduling knowledge.But there has been no scheduling knowledges to manage relevant document and patent at present.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind towards intelligence manufacture
Scheduling knowledge management system.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of scheduling knowledge management system towards intelligence manufacture, comprising:
Scheduling knowledge study module carries out knowledge excavation for sample for dispatching data by data-driven, obtains scheduling and knows
Know;
Scheduling knowledge application module, the response when receiving scheduling request, for calling the scheduling knowledge to generate and intelligence
The scheduling strategy that energy Workshop Production state matches, and obtain scheduling result;
Scheduling knowledge online evaluation module responds after obtaining scheduling result, for judging to adjust according to the scheduling result
Spend knowledge effectiveness;
Scheduling knowledge update module responds when determining scheduling knowledge failure, for applying on-line study method, is increasing
Scheduling knowledge is adjusted on the basis of more new samples.
Further, the scheduling knowledge characterization: under set regulation goal, the matching of production status and scheduling strategy
Relationship.
Further, in the scheduling knowledge study module, knowledge excavation is carried out using extreme learning machine algorithm, study obtains
Obtain scheduling knowledge.
Further, the scheduling knowledge online evaluation module includes:
Control figure generation unit generates scheduling satisfaction control for obtaining scheduling satisfaction according to the scheduling result
Figure;
Knowledge effectiveness judging unit, for judging scheduling knowledge according to the distribution situation of the scheduling satisfaction control figure
Whether effectively.
Further, the scheduling satisfaction refers to the ratio of actual schedule result Yu desired scheduling result, expression formula are as follows:
Wherein, λ is scheduling satisfaction, and P is actual schedule as a result, P ' is expectation scheduling result.
Further, the generating process of the scheduling satisfaction control figure specifically:
Using multiple scheduling satisfaction values as a sample group, the average value mean value and mean pole of multiple sample groups are calculated
The average value of difference, multiple sample groups forms scheduling satisfaction control figure, and upper and lower control limit, calculation formula is arranged are as follows:
Wherein,Respectively upper and lower control limit,For the average value mean value of sample group,For sample
The mean range of group, σ are sample group standard error of the mean, A2For control figure coefficient.
Further, in the Effective judgement unit, when the distribution situation of scheduling satisfaction control figure meets following appoint
It is determined as that scheduling knowledge fails when one condition:
1) the average value X of sample groupiBeyond the upper lower control limit in control figure;
2) continuous three XiIn there are two XiRegion A is fallen within, region A includesWith
3) continuous five XiIn there are four XiRegion B is fallen within, region B includesWith
Further, the scheduling knowledge online evaluation module further include:
Control figure Effective judgement unit is used for recording dispatching knowledge Failure count, scheduling knowledge is occurring twice in succession
When failure, determines the failure of scheduling satisfaction control figure, the control figure generation unit is called to regenerate scheduling satisfaction control
Figure.
Further, the scheduling knowledge update module uses online extreme learning machine algorithm, has update sample based on newly-increased
This sample set, training adjust existing scheduling knowledge.
Further, the selection mode of the more new samples are as follows: new samples and original sample are described by condition with approximately linear
The variation of this collection, when new samples and original sample Line independent, using new samples as more new samples.
Compared with prior art, the present invention is for big data technology under industrial 4.0 environment in intelligent workshop operation management
Widely apply this trend, propose the lifecycle management of scheduling knowledge, have with following the utility model has the advantages that
1) present invention is by the way that the management of scheduling knowledge to be applied in intelligent workshop operation, to the entire life of scheduling knowledge
Period is managed, study, scheduling knowledge including scheduling knowledge application, scheduling knowledge online evaluation and scheduling knowledge
Update, can preferably improve the adaptability of scheduling knowledge, it is ensured that scheduling validity, sustainable Intelligent Optimal workshop globality
Energy.
2) present invention analyzes scheduling satisfaction of the intelligent workshop after application schedules knowledge by control figure, judges to dispatch
Whether process is controlled, realizes the online evaluation of scheduling knowledge, intuitively, effectively.
3) the new samples data that the present invention is updated by (ALD) selection for scheduling knowledge using approximately linear, are adopted simultaneously
With on-line learning algorithm, existing scheduling knowledge is updated, improves the adaptability of scheduling knowledge.
4) present invention has important application value to the intelligent workshop running for solving data-driven, to the intelligent workshop of raising
Production management level have important directive significance.
Detailed description of the invention
Fig. 1 is the structural diagram of the present invention;
Fig. 2 is the influence that scheduling knowledge updates to bottleneck device congestion lengths in the embodiment of the present invention;
Fig. 3 is the influence that scheduling knowledge updates to productivity in the embodiment of the present invention;
Fig. 4 is the influence that scheduling knowledge updates to average daily mobile step number in the embodiment of the present invention;
Fig. 5 is scheduling satisfaction control figure in the embodiment of the present invention.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with technical solution of the present invention
Premised on implemented, the detailed implementation method and specific operation process are given, but protection scope of the present invention is not limited to
Following embodiments.
As shown in Figure 1, the present invention provides a kind of scheduling knowledge management system towards intelligence manufacture, including scheduling knowledge
Module, scheduling knowledge application module, scheduling knowledge online evaluation module and scheduling knowledge update module are practised, it can be to scheduling knowledge reality
Existing lifecycle management.Scheduling knowledge study module carries out knowledge excavation by data-driven, for sample for dispatching data, obtains
Scheduling knowledge is obtained, scheduling knowledge is characterized under set regulation goal, the matching relationship of production status and scheduling strategy;Scheduling is known
Application module response when receiving scheduling request is known, for calling the scheduling knowledge to generate and intelligent workshop production status phase
Matched scheduling strategy, and obtain scheduling result;Scheduling knowledge online evaluation module responds after obtaining scheduling result, is used for root
According to the scheduling result, scheduling knowledge validity is judged;Scheduling knowledge update module is responded when determining scheduling knowledge failure, is used
In applying on-line study method, scheduling knowledge is adjusted on the basis of increasing more new samples.
The present embodiment is illustrated by taking Benchmark Models Sets as an example.Benchmark Models Sets are by Arizona, USA
One group of semiconductor production line model that Department of Industrial Engineering, state university semiconductors manufacture laboratory provides.The sets of data is all derived from
True semiconductor foundry combines publication by several field of semiconductor manufacture esbablished corporations and scholar.It is therein
BenchMark6 model is the biggish semiconductor manufacturing plant model of a production scale.The model contains 104 and sets
Standby group, 228 equipment have every kind of product of 9 kinds of products to have procedures up to a hundred in model, and the process time of individual processes is extremely
It is long, it is longest to can reach 18 hours, meanwhile, the time of feeding and blanking is also contemplated in the model, is more tallied with the actual situation.It should
Production line model is in large scale, process flow is complicated, has contained data abundant, is a typical intelligent workshop.
It is objective for implementation, the work of the above-mentioned scheduling knowledge management system towards intelligence manufacture using BenchMark6 model
Process is as follows:
Step 1, sample information A is definedP={ Si,Di|Si∈Rn,Di∈Rm, i=1,2,3 ..., N }, wherein Si=(sI, 1,
sI, 2..., sI, n) the production feature of intelligent workshop production status is described;Di=(dI, 1, dx, 2..., dI, m) describe in current state
SiUnder, meet regulation goal P it is optimal when used optimal scheduling strategy.Simulation model is run, initial sample is acquired, goes forward side by side
The optimization of row scheduling strategy, obtains above-mentioned optimal sample set AP.Choose the average process-cycle MCT of workpiece, Workshop Production rate PROD,
Production status of per day mobile these three performance indicators of step number MDayMOV of workpiece as optimizing scheduling target, in sample information
Information S definition is shown in Table 1.
Production status S is defined in 1 BenchMark6 sample information of table
Step 2, scheduling knowledge study module carries out off-line learning to sample, obtains according to the regulation goal in intelligent workshop
Corresponding scheduling knowledge.According to the optimal sample set under different regulation goals, using ELM algorithm, to characterize the life of production status
Producing feature S is input, characterizes the D of scheduling strategy as output, training study, the matching for obtaining production status and scheduling strategy is closed
System, i.e. ELM model, the model are the scheduling knowledge for learning to obtain.Application schedules knowledge can be obtained under the regulation goal
Optimal scheduling strategy.Scheduling knowledge under different regulation goals constitutes scheduling knowledge library.
Learn the detailed process of acquisition scheduling knowledge using the extreme learning machine algorithm are as follows:
1a) setting parameter k is the training sample batch that training is added, and is trained at this time to initial sample, i.e. k=0;
1b) select N0A initial sample (Xi,Yi), wherein inputting Xi=[xi1,xi2,...xin]T∈RnFor production status Si,
Export Yi=[yi1,yi2,...yim]T∈RmFor scheduling strategy Di, give hidden node number L, initialization network parameter (Wi,
bi), i=1,2 ... L, wherein Wi=[wi1, wi2…win]TFor the weight vectors of i-th of hidden node and input node, biFor
The biasing of i-th of hidden node;
1c) calculate initial hidden layer output matrix H0, wherein G (x) is excitation function:
1d) calculate initial output weightWherein
1e) learn obtained scheduling knowledge KP: f (X)=H0·β0。
The scheduling knowledge in scheduling data, detailed process are excavated based on ELM algorithm are as follows:
2a) generate training set and test set.Sample set is divided into training sample set and test sample collection, wherein training sample
Collection is used to training study scheduling knowledge, and test sample collection evaluates the scheduling knowledge learnt;
2b) the study of scheduling knowledge.Application training sample learning scheduling knowledge (establishes ELM model), production status collection S
As input, scheduling strategy D is output;
2c) scheduling knowledge learnt is tested with test sample, analyzes test result, evaluates established model
Effect.
Step 3, scheduling knowledge application module application schedules knowledge carries out on-line scheduling to workshop.It was sampling week with 4 hours
Phase acquires the production status in workshop, application schedules knowledge acquisition scheduling strategy, traffic control strategy, with the workshop of a cycle
Production performance is as scheduling result, to determine the validity of scheduling knowledge.Production performance after recording intelligent workshop a cycle;
Meanwhile expectation production performance is obtained by simulation model, scheduling knowledge online evaluation module is according to the actual production performance of record
The scheduling satisfaction in each period is calculated with desired production performance.
The scheduling satisfaction refers to the ratio of actual schedule result Yu desired scheduling result, it is expected that scheduling result can be by pre-
It surveys or emulation obtains, expression formula are as follows:
Wherein, λ is scheduling satisfaction, reflects scheduling knowledge to the applicability of current intelligent workshop production status, P is real
Border scheduling result, P ' are desired scheduling result.
Step 5, using control map analysis scheduling satisfaction, whether controllable (scheduling knowledge is scheduling knowledge online evaluation module
No failure), it analyzes recorded.
The drawing process of scheduling satisfaction control figure is: n sample group of acquisition, the capacity of every group of sample is 4, i.e., by four
Scheduling satisfaction value is combined into a sample group, calculate the average value of each sample group with it is very poor:
RiMinimum value in maximum value-sample group in=sample group
Calculate the mean value of all sample cell meansAnd sample group mean range
Control figure is drawn, center line is the mean value of sample cell mean, and upper lower control limit is sample at a distance from center line
3 times of this standard error of the mean σ, calculation formula are as follows:
Center line:
Upper control limit:
Lower control limit:
Wherein A2It is control figure coefficient, is derived from control figure coefficient table (such as table 2), value is related with sample size.Table 2 controls
Figure coefficient table (Wayne C.Turner, Joe H.Mize writes, and Zhang Xuzhu is translated, " Industrial Engineering outline ",
Publishing house, Tsinghua University, 2007)
As shown in table 3, it is judged to dispatching when the distribution situation of scheduling satisfaction control figure meets following either condition and knows
Know failure:
1) the average value X of sample groupiBeyond the upper lower control limit in control figure;
2) continuous three XiIn there are two XiRegion A is fallen within, region A includesWith
3) continuous five XiIn there are four XiRegion B is fallen within, region B includesWith
3 scheduling knowledge replacement criteria of table
Scheduling knowledge online evaluation module determines whether control figure fails also according to the frequency that scheduling knowledge updates, specifically
Are as follows: when scheduling knowledge update occurs twice in succession, it can determine that control figure fails, need to repaint control figure, such as 4 institute of table
Show.
4 control figure replacement criteria of table
Practical productivity, expectation productivity and the scheduling satisfaction of 5 sample of table
Table 5 is 40 datas record, including practical productivity, expectation productivity and scheduling satisfaction, sample size are set to
4,40 samples are determined as 10 sample groups.It searches control figure coefficient table (table 2), at this time A2=0.729, according to the number of record
According to, the center line and upper lower control limit of control figure are calculated separately, are as follows:
Sample group scheduling satisfaction mean of mean are as follows:
Sample group scheduling satisfaction mean value it is very poor are as follows:
Center line:
Upper control limit:
Lower control limit:
Fig. 5 is that drawn control figure then needs scheduling knowledge when sample meets the replacement criteria of 3 scheduling knowledge of table
It updates;When meeting the replacement criteria of 4 control figure of table, control figure is also updated.As can be seen from Figure 5 preceding 9 groups of samples
Mean value in upper lower control limit, be unsatisfactory for all control figure replacement criterias described in table 4, show that current scheduling knowledge is suitable
For Workshop Production state, it is not required to be scheduled the update of knowledge;And the mean value of the 10th group of sample as shown in Figure 5 has exceeded control
The Lower Limits of figure meet the replacement criteria of scheduling knowledge, illustrate that current scheduling knowledge is mismatched with Workshop Production state at this time,
It needs to update scheduling knowledge.
Step 6, when judging scheduling knowledge failure, (Approximate linear is relied on using approximately linear
Dependence, ALD) method acquisition more new samples, using the incremental learning algorithm of ELM, i.e., online extreme learning machine algorithm
(Online Sequential Extreme Learning Machine, OS-ELM) method learns new samples, updates
Scheduling knowledge continues Instructing manufacture scheduling using updated scheduling knowledge.
The selection mode of more new samples are as follows: the variation for describing new samples Yu original sample collection by condition with approximately linear, when
When new samples and original sample Line independent, using new samples as more new samples.ALD value is defined as:
Wherein, xlFor training sample, i.e. training sample described in step 2, q is sample number, xq+1For new samples, ν is specified
Threshold value, αq+1=[α1 α2… αq]TFor weight coefficient.When approximate error is less than threshold value ν, representing new samples and original sample is
It linearly relies on, it can by original sample linear expression, then cannot be used as more new samples.Otherwise illustrate new samples and original sample
Originally it is Line independent, can be used as more new samples.
Online extreme learning machine algorithm is to existing scheduling knowledge KP: f (X) is adjusted, specifically:
6a)+1 lot sample notebook data (X of kth for being newly addedk+1,Yk+1), calculate hidden layer output matrix Hk+1
6b) update output weight betak+1:
K=k+1 6c) is enabled, continues incremental learning until training sample training terminates to obtain output weight beta.Updated
Scheduling knowledge K afterwardsP: f (X)=H β.
Scheduling knowledge is subjected to online updating respectively and without the dispatching method of online updating, is applied to BenchMark6
Scheduling result in model, under more two different modes.
The online updating of scheduling knowledge refers in scheduling process, obtains new sample by (ALD) method using approximately linear
This, and learn new sample data using online extreme learning machine (OS-ELM) method, to adjust existing scheduling knowledge, and adopt
With updated scheduling knowledge Instructing manufacture, i.e. scheduling knowledge is that dynamic updates.
The not online updating of scheduling knowledge refers to learns scheduling knowledge from off-line data, and is applied to intelligent Job-Shop
In the process, and in entire scheduling process, scheduling knowledge is not adjusted, i.e., scheduling knowledge is static fixed.
Intelligent workshop service condition is as follows: the intelligent workshop service condition and equipment situation that two ways is acted on are homogeneous
Together, entire intelligent workshop operation is divided into two time zones, each time zone includes 50 sampling periods, between each sampling period
It is divided into 4 hours., there are 9 kinds of products in the characteristics of according to BenchMark6 on production line, and the quantity and mixing ratio of product can all influence
The operation in intelligence manufacture workshop, in time zone 1, it is 1 that the daily every kind of product of 9 kinds of products, which is arranged, to feed intake several;In time zone 2, if
Set the number random distribution between [0,3] that feeds intake of every kind of product.What the scheduling knowledge and static state that online updating is respectively adopted were fixed
Scheduling knowledge carries out production scheduling to BenchMark6, and records corresponding production performance, as a result as shown in figs 2-4.
Scheduling result is analyzed, carries out production scheduling using the scheduling knowledge of online updating, generated bottleneck device waits
Team leader's average value is 12.366lot, and productivity is 6.11Lot/ days, and average daily mobile step number is 1743.3 steps.It is fixed using static state
Scheduling knowledge carry out production scheduling, generated bottleneck device congestion lengths average value is 12.996Lot, and productivity is
6.01Lot/ days, average daily mobile step number was 1720.4 steps.
Therefore, it can be seen from Fig.2-Fig.4 that, in process of production, applied scheduling knowledge is updated, i.e., using dynamic
The scheduling knowledge of update carries out production scheduling, can obtain more preferably production performance.It is carried out using the scheduling knowledge of online updating
Scheduling, compared with using static fixed scheduling knowledge, the optimization of bottleneck device congestion lengths is obvious, optimizes 5%, production
1.67% and 1.34% has been separately optimized in rate and average daily mobile step number.Trend is adopted in production process early period in analysis chart 2- Fig. 4
With identical scheduling knowledge, production performance indifference;In the production process later period, scheduling knowledge is updated, obtained scheduling
It is tactful different, different production performances is resulted in, also, updated scheduling knowledge can adapt to the variation of production status, make
Production process is obtained well to run.
The preferred embodiment of the present invention has been described in detail above.It should be appreciated that those skilled in the art without
It needs creative work according to the present invention can conceive and makes many modifications and variations.Therefore, all technologies in the art
Personnel are available by logical analysis, reasoning, or a limited experiment on the basis of existing technology under this invention's idea
Technical solution, all should be within the scope of protection determined by the claims.
Claims (10)
1. a kind of scheduling knowledge management system towards intelligence manufacture characterized by comprising
Scheduling knowledge study module is carried out knowledge excavation for sample for dispatching data, is obtained scheduling knowledge by data-driven;
Scheduling knowledge application module, the response when receiving scheduling request, for calling the scheduling knowledge to generate and intelligent vehicle
Between the scheduling strategy that matches of production status, and obtain scheduling result;
Scheduling knowledge online evaluation module responds after obtaining scheduling result, for judging that scheduling is known according to the scheduling result
Know validity;
Scheduling knowledge update module is responded when determining scheduling knowledge failure, for applying on-line study method, is updated increasing
Scheduling knowledge is adjusted on the basis of sample.
2. the scheduling knowledge management system according to claim 1 towards intelligence manufacture, which is characterized in that the scheduling is known
Know characterization: under set regulation goal, the matching relationship of production status and scheduling strategy.
3. the scheduling knowledge management system according to claim 1 towards intelligence manufacture, which is characterized in that the scheduling is known
Know in study module, knowledge excavation is carried out using extreme learning machine algorithm, study obtains scheduling knowledge.
4. the scheduling knowledge management system according to claim 1 towards intelligence manufacture, which is characterized in that the scheduling is known
Knowing online evaluation module includes:
Control figure generation unit generates scheduling satisfaction control figure for obtaining scheduling satisfaction according to the scheduling result;
Knowledge effectiveness judging unit, for whether judging scheduling knowledge according to the distribution situation of the scheduling satisfaction control figure
Effectively.
5. the scheduling knowledge management system according to claim 4 towards intelligence manufacture, which is characterized in that the scheduling is full
Meaning degree refers to the ratio of actual schedule result Yu desired scheduling result, expression formula are as follows:
Wherein, λ is scheduling satisfaction, and P is actual schedule as a result, P ' is expectation scheduling result.
6. the scheduling knowledge management system according to claim 4 towards intelligence manufacture, which is characterized in that the scheduling is full
The generating process of meaning degree control figure specifically:
Using multiple scheduling satisfaction values as a sample group, the average value mean value and mean range of multiple sample groups are calculated, it is more
The average value of a sample group forms scheduling satisfaction control figure, and upper and lower control limit, calculation formula is arranged are as follows:
Wherein,Respectively upper and lower control limit,For the average value mean value of sample group,For the flat of sample group
Very poor, σ is sample group standard error of the mean, A2For control figure coefficient.
7. the scheduling knowledge management system according to claim 6 towards intelligence manufacture, which is characterized in that the validity
In judging unit, it is determined as that scheduling knowledge fails when the distribution situation of scheduling satisfaction control figure meets following either condition:
1) the average value X of sample groupiBeyond the upper lower control limit in control figure;
2) continuous three XiIn there are two XiRegion A is fallen within, region A includesWith
3) continuous five XiIn there are four XiRegion B is fallen within, region B includesWith
8. the scheduling knowledge management system according to claim 4 towards intelligence manufacture, which is characterized in that the scheduling is known
Know online evaluation module further include:
Control figure Effective judgement unit is used for recording dispatching knowledge Failure count, and scheduling knowledge failure is occurring twice in succession
When, determine the failure of scheduling satisfaction control figure, the control figure generation unit is called to regenerate scheduling satisfaction control figure.
9. the scheduling knowledge management system according to claim 1 towards intelligence manufacture, which is characterized in that the scheduling is known
Know update module and use online extreme learning machine algorithm, based on the newly-increased sample set for there are more new samples, training adjusts existing tune
Spend knowledge.
10. the scheduling knowledge management system according to claim 1 or described in 9 towards intelligence manufacture, which is characterized in that it is described more
The selection mode of new samples are as follows: the variation for describing new samples Yu original sample collection by condition with approximately linear, when new samples and original
When sample Line independent, using new samples as more new samples.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811269262.2A CN109523136A (en) | 2018-10-29 | 2018-10-29 | A kind of scheduling knowledge management system towards intelligence manufacture |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811269262.2A CN109523136A (en) | 2018-10-29 | 2018-10-29 | A kind of scheduling knowledge management system towards intelligence manufacture |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109523136A true CN109523136A (en) | 2019-03-26 |
Family
ID=65773071
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811269262.2A Pending CN109523136A (en) | 2018-10-29 | 2018-10-29 | A kind of scheduling knowledge management system towards intelligence manufacture |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109523136A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112862326A (en) * | 2021-02-19 | 2021-05-28 | 同济大学 | System scheduling optimization method based on big data mining |
US11762376B2 (en) | 2019-12-03 | 2023-09-19 | Industrial Technology Research Institute | Quick dispatching rule screening method and apparatus |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103310285A (en) * | 2013-06-17 | 2013-09-18 | 同济大学 | Performance prediction method applicable to dynamic scheduling for semiconductor production line |
CN104536412A (en) * | 2014-12-23 | 2015-04-22 | 清华大学 | Photoetching procedure dynamic scheduling method based on index forecasting and solution similarity analysis |
CN104766100A (en) * | 2014-10-22 | 2015-07-08 | 中国人民解放军电子工程学院 | Infrared small target image background predicting method and device based on machine learning |
CN105045243A (en) * | 2015-08-05 | 2015-11-11 | 同济大学 | Semiconductor production line dynamic scheduling device |
CN108492013A (en) * | 2018-03-09 | 2018-09-04 | 同济大学 | A kind of manufacture system scheduling model validation checking method based on quality control |
-
2018
- 2018-10-29 CN CN201811269262.2A patent/CN109523136A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103310285A (en) * | 2013-06-17 | 2013-09-18 | 同济大学 | Performance prediction method applicable to dynamic scheduling for semiconductor production line |
CN104766100A (en) * | 2014-10-22 | 2015-07-08 | 中国人民解放军电子工程学院 | Infrared small target image background predicting method and device based on machine learning |
CN104536412A (en) * | 2014-12-23 | 2015-04-22 | 清华大学 | Photoetching procedure dynamic scheduling method based on index forecasting and solution similarity analysis |
CN105045243A (en) * | 2015-08-05 | 2015-11-11 | 同济大学 | Semiconductor production line dynamic scheduling device |
CN108492013A (en) * | 2018-03-09 | 2018-09-04 | 同济大学 | A kind of manufacture system scheduling model validation checking method based on quality control |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11762376B2 (en) | 2019-12-03 | 2023-09-19 | Industrial Technology Research Institute | Quick dispatching rule screening method and apparatus |
CN112862326A (en) * | 2021-02-19 | 2021-05-28 | 同济大学 | System scheduling optimization method based on big data mining |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Liu et al. | Many-objective job-shop scheduling: A multiple populations for multiple objectives-based genetic algorithm approach | |
CN110163410B (en) | Line loss electric quantity prediction method based on neural network-time sequence | |
US20180356775A1 (en) | Heuristic method of automated and learning control, and building automation systems thereof | |
CN109478045A (en) | Goal systems is controlled using prediction | |
CN102402716B (en) | Intelligent production decision support system | |
Peng et al. | Deep reinforcement learning approach for capacitated supply chain optimization under demand uncertainty | |
Papageorgiou et al. | Application of fuzzy cognitive maps to water demand prediction | |
CN107644297B (en) | Energy-saving calculation and verification method for motor system | |
CN108876001A (en) | A kind of Short-Term Load Forecasting Method based on twin support vector machines | |
CN109919421A (en) | Short-term power load prediction model establishment method based on VMD-PSO-BPNN | |
CN111898867B (en) | Airplane final assembly production line productivity prediction method based on deep neural network | |
CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
Tran et al. | Using Fuzzy Clustering Chaotic-based Differential Evolution to solve multiple resources leveling in the multiple projects scheduling problem | |
Xin et al. | An adaptive BPSO algorithm for multi-skilled workers assignment problem in aircraft assembly lines | |
Wang et al. | An interoperable adaptive scheduling strategy for knowledgeable manufacturing based on SMGWQ-learning | |
CN109523136A (en) | A kind of scheduling knowledge management system towards intelligence manufacture | |
US20220269835A1 (en) | Resource prediction system for executing machine learning models | |
CN108492013A (en) | A kind of manufacture system scheduling model validation checking method based on quality control | |
CN108829846A (en) | A kind of business recommended platform data cluster optimization system and method based on user characteristics | |
Napalkova et al. | Multi-objective stochastic simulation-based optimisation applied to supply chain planning | |
Khoroshev et al. | Adaptive clustering method in intelligent automated decision support systems | |
CN115016405A (en) | Process route multi-objective optimization method based on deep reinforcement learning | |
Neukart et al. | High order computational intelligence in data mining a generic approach to systemic intelligent data mining | |
CN108596781A (en) | A kind of electric power system data excavates and prediction integration method | |
Owda et al. | Using artificial neural network techniques for prediction of electric energy consumption |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190326 |