CN106600058A - Prediction method for combinations of cloud manufacturing service quality of service (QoS) - Google Patents
Prediction method for combinations of cloud manufacturing service quality of service (QoS) Download PDFInfo
- Publication number
- CN106600058A CN106600058A CN201611147088.5A CN201611147088A CN106600058A CN 106600058 A CN106600058 A CN 106600058A CN 201611147088 A CN201611147088 A CN 201611147088A CN 106600058 A CN106600058 A CN 106600058A
- Authority
- CN
- China
- Prior art keywords
- service
- task
- manufacture
- lambda
- cloud
- 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 157
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012545 processing Methods 0.000 claims abstract description 27
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 238000013277 forecasting method Methods 0.000 claims description 28
- 230000008859 change Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 230000004044 response Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000003068 static effect Effects 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 claims description 4
- 230000007613 environmental effect Effects 0.000 claims description 4
- 238000011084 recovery Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 claims description 2
- 230000001537 neural effect Effects 0.000 claims 1
- 238000009826 distribution Methods 0.000 description 9
- 238000009499 grossing Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 5
- 238000007476 Maximum Likelihood Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 238000005315 distribution function Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000000205 computational method Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000003754 machining Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Health & Medical Sciences (AREA)
- Manufacturing & Machinery (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a prediction method for combinations of cloud manufacturing service quality of service (QoS), which belongs to the networked manufacturing field and comprises the steps of cloud manufacturing task modeling, cloud manufacturing service execution time predicting, cloud manufacturing service reliability predicting and service availability predicting. In the invention, the cloud manufacturing task is divided into a computer type task and a manufacturing and processing task; and based on this, predictions are made on the cloud manufacturing service QoS; a BP neural network is used to predict the cloud manufacturing service execution time; and a discrete Markov model is used to predict the service reliability; in combination with the continuous Markov model and a queuing model, predictions are made on the service availability. As different QoS indicators have different influential factors, different prediction methods are employed for the different indicators so that the prediction model is better than one individual model in terms of both efficiency and quality, the prediction accuracy is increased and that important data support can be provided for the service combination, resource optimized allocation and management in a cloud manufacturing environment.
Description
Technical field
The present invention relates to a kind of combination forecasting method of manufacture cloud service QoS, more particularly to a kind of manufacture cloud service QoS
Neutral net and Markov model combination forecasting method, belong to net instrument field.
Background technology
Manufacturing industry in national economy in occupation of critical role, traditional manufacture in information age faces enormous challenge and
New opportunity.Net instrument pattern becomes the new selection of manufacturing industry, but manufactures the net instrument patterns such as grid, Agile manufactruing
There are problems at aspects such as service mode, manufacture resource sharing distribution, physical terminal device access, information securities.Here
Under background, cloud manufacture is arisen at the historic moment.Cloud manufacture is linked into manufacturing recourses in network by technologies such as embedded, Internet of Things, is pressed
User's request tissue network based manufacturing resource (manufacturing cloud), provides the user all kinds of manufacturing services on demand, is capable of achieving manufacture enterprise
The shared resources of efficient collaboration and various manufacturing recourses between industry with it is integrated.
Proposing for user under cloud manufacturing environment for task, generally there are the substantial amounts of Services Composition for meeting functional requirement
Scheme, the functional attributes for manufacturing cloud service determine whether Services Composition scheme is feasible, manufacture the nonfunctional space of cloud service such as
Service price, execution time, reliability etc. determine that candidate service assembled scheme can to what extent meet the individual of user
Property demand.Judge whether the manufacture cloud service assembled scheme generated based on functional requirement meets QoS demand, and from function phase
As select in Services Composition scheme QoS attributes to be preferably supplied to user very necessary.And these are required in manufacture cloud clothes
Business assembled scheme performs the front QoS attributes to the Services Composition scheme and is predicted, the mark for predicting the outcome as judgement and preferentially
It is accurate.
At present, less for the research that QoS Forecasting Methodologies are serviced in cloud manufacturing environment, some scholars are pre- to Web service QoS
Survey method is studied." a kind of service quality Forecasting Methodology of Web Service " one that Shao Lingshuan etc. was delivered in 2009
Wen Zhong, by calculating user's similarity and service similarity, and based on this to the qos value of the original service of consumer
The method being predicted, but wherein to the summary a bit deficient in of temporal mode.Li Yiqing has delivered a kind of " Web clothes in 2013
The paper of business QoS dynamic prediction methods ", Liu Zhizhong was delivered " Web service penetration quality dynamic Study on Forecasting Method " in 2014
The scholars such as paper, Chen Kehan, Tang Yi are also studied Web service QoS Forecasting Methodologies.
The studies above work is concentrated mainly on Web service environment, there is the access of physical equipment resource, resource on cloud manufacturing platform
The characteristics of isomery, tissue dynamic change etc. itself is exclusive, the prediction to manufacturing cloud service QoS is considered as and solves following problem：
1. manufacture cloud and cover more extensive Service Source, except traditional calculating Service Source, also physical equipment, system
The other kinds of Service Source such as ability, manufacturing knowledge is made, the corresponding cloud clothes that these manufacturing service resources are formed Jing after virtualization
Business has polytropy in QoS features.Therefore, the premise to manufacturing cloud service QoS predictions is the impact of Correct Analysis different QoS
Factor.
2. cloud manufacturing platform has higher moving when responding cloud service request and servicing according to user's request organizational resources
State property, the online access total time of some cloud services is shorter, and its corresponding service execution historical record is relatively fewer.For this purpose, system
Make cloud service QoS Forecasting Methodologies and be considered as the situation that history reference data are not enriched.
3. in existing service QoS Forecasting Methodologies, grey forecasting model requires less data, and principle is directly perceived, calculates
Complexity is suitable, and prediction effect is good, and manufacture cloud service QoS Forecasting Methodologies are contrasted to illustrate that this is pre- with grey forecasting model
The superiority of survey method.
The content of the invention
The purpose of the present invention is that the manufacture cloud service QoS in cloud manufacturing environment is predicted, and predicting the outcome can comment QoS
The manufacture cloud service resource managements such as valency, preferred, the scheduling of resource distribution of cloud service combination and decision-making produce positive role, it is proposed that one
Plant the combination forecasting method of manufacture cloud service QoS.
The present invention core technology thought be：Consider that there is physical equipment access, resource dynamic change in cloud manufacturing environment
The features such as, the task requests of user are modeled initially with ontology description language, by the similarity between calculating task,
Obtain the similar historical data of task；Secondly, the execution time for manufacturing cloud service is predicted using BP neural network；Using
Discrete Markov Model is predicted to reliability of service；The availability for servicing is carried out using continuous Markov model
Prediction.On this basis, obtain manufacturing predicting the outcome for cloud service QoS.
A kind of combination forecasting method of manufacture cloud service QoS, mainly includes following four step：
Step one, user task request modeling and the acquisition of similar historical data, specially：
Step 1.1 utilizes ontology description language (OWL-S), the manufacturing operation request submitted to user Xiang Yun manufacturing platforms
It is described；
Step 1.2 chooses full by calculating user's current task and calling the similarity between the historic task for servicing
The historic task of sufficient similarity threshold constraint, and referred to as similar historical task；
Similar historical number of the qos value of all similar historical tasks that step 1.3 obtains step 1.2 as user task
According to obtaining the similar historical data acquisition system of user's current task；
Step 2, manufacture cloud service running time prediction, specially：
Step 2.1 task requests current to user are analyzed, and be classified as calculating generic task and manufacture processing class is appointed
Business；
Different types of cloud manufacturing operation performs the influence factor of time in step 2.2 analytical procedure 2.1, and task is held
The row time is divided into quiet hour and dynamic time, and the quiet hour is carried out the static part of time, not outside influences,
The quiet hour for calculating generic task is ideally task execution time on a virtual machine, the static state of manufacture processing generic task
Time, the Conventional Time of machining task can be regarded as；Dynamic time is carried out the dynamic change part of time, can be with outer
The change of portion's factor and change, calculate the quiet hour and prediction dynamic time, for manufacturing operation predict when statistics
For in meaning, mean value can be seen that the integral level of manufacturing time, ignore the impact of dynamic factor.
If representing tasks carrying total time, τ with τsRepresent the quiet hour of tasks carrying, τdRepresent the dynamic of tasks carrying
The state time, then the expression of task execution time such as formula (1)：
τ=τs+τd (1)
Wherein, quiet hour τsBy being calculated；Dynamic time τdObtained by BP neural network prediction；
Specifically, to the quiet hour for calculating generic task and manufacture processing generic task of step 2.1 output, it was calculated
Journey is respectively：
For generic task is calculated, the computing resource service called finally is performed on a virtual machine, calculates the static state of generic task
Time is ideally task execution time on a virtual machine, i.e., virtual machine is idle, network it is unimpeded in the case of, task
The execution time on a virtual machine.Its static execution time τsComputing formula such as formula (2)：
τs=D/V (2)
Wherein, D is the amount of calculation of virtual task, and V is given virtual machine processing speed；
Manufacture processing generic task is considerably complicated, and the quiet hour is calculated without direct formula, manufacture processing generic task
Quiet hour, the Conventional Time of machining task can be regarded as, be said from the statistical significance, the mean value calculated with historical data
Geostationary Conventional Time can be regarded as.So, for the quiet hour of manufacture processing generic task, by calculating historical data
Be averagely worth to quiet hour τs, calculated with equation below (3)：
Wherein, τiFor the historical data that manufacture processing generic task performs the time, n is the number of historical data；
Calculating generic task and manufacture processing generic task for step 2.1 output, the Dynamic Execution Time Calculation mistake of task
Journey is respectively：
For calculate generic task for, impact task perform on a virtual machine time change principal element be cpu load,
Memory size and the network bandwidth.By using cpu load, memory size, the network bandwidth, perform the time historical data as sample
And BP neural network is trained, it is possible to achieve dynamic time τdPrediction；
For manufacture processing generic task, the principal element that task performs time change in manufacturing equipment is affected to be behaviour
Author's human factor, apparatus factor and environmental factor.By by these factors and perform the time historical dataAs sample simultaneously
BP neural network is trained, it is possible to achieve to dynamic time τdPrediction.Wherein, human factor can be divided into person works Jing
Test and personnel specialty level of skill；Apparatus factor can be divided into plant maintenance level and equipment failure rate；Environmental factor can be divided into ring
Border temperature and humidity；
Step 3, manufacture cloud service reliability prediction, specially：
The historical data of step 3.1 pair manufacture cloud service reliability is analyzed, using Discrete Markov Model to clothes
Business reliability is described；
The state-transition matrix of Discrete Markov Model in step 3.2 calculation procedure 3.1, the reliability to manufacturing cloud service
Property is predicted；
Wherein, the state-transition matrix of Discrete Markov Model represents and calculates equation below (4) and formula (5) respectively
It is shown：
In formula (4), PiRepresent service RSiState-transition matrix,To service the probability for keeping failure,For service by
Failure is changed into available probability,It is to service from the available probability for being changed into and failing,Available probability is kept for service；Formula (5)
In, E0Represent service response failure, E1Represent service response success, (E0→E0) represent service holding failure, (E0→E1) represent
Service is changed into available from failure, (E1→E0) represent that service is changed into failure, (E from available1→E1) represent that service holding is available；ni、
NiService RS is represented respectivelyiKeep the number of times and failure total degree, m of failurei、MiService RS is represented respectivelyiKeep available number of times
With available total degree；
Step 4, manufacture cloud service availability prediction, specially：
The Failure Factors for manufacturing cloud service are divided into two by step 4.1：One is service failure itself, and another is service
The network failure at place；
The continuous Markov model of step 4.2 is described respectively to service failure and network failure, service availability
Including two aspects of service availability itself and network availability, cloud service is only in the case of service and network are simultaneously available
It is called, to respond the task requests of user.
Wherein, the calculating of service availability includes the following two kinds situation：
If 4.2A services RSiOriginal state is available, then service availability, is designated as Pi, calculated by formula (6)：
If 4.2B services RSiOriginal state is failure, then service availability Pi, calculated by formula (7)：
In formula (6),To service RSiAvailability itself,For RSiPlace network availability, both with regard to
The function of time t；μ1To service RSiCrash rate, μ2To service RSiFault recovery rate, λ1To service RSiThe mistake of place network
Efficiency, λ2To service RSiThe fault recovery rate of place network；
For manufacture processing class service, the characteristics of considering manufacturing equipment resource is also needed to when its availability is calculated,
I.e. equipment has exclusivity, and during a task is performed other task requests can not be responded, and this point is provided with class is calculated
Source service is different；
For manufacture processing class service, it is assumed that b is the service request number that its most multipotency is received, and μ is service rate, i.e.,
The service number of unit interval interior energy response, λ is the user's request number that the unit time reaches, and (is arrived first and first taken using FCFS
Business) service response rule, service requester needs line for service successively, then this process can with a queuing model come
Description；
Make piRepresent there is the probability of i service request in queue, then can be obtained by queuing theory：
Wherein,Service intensity is represented, in general ρ ＜ 1.p0Represent in current queue without the probability of request.
For manufacture processing class service, service availability PiComputing formula be：
So far, from step one to step 4, a kind of combination forecasting method of manufacture cloud service QoS is completed.
Beneficial effect
A kind of combination forecasting method of manufacture cloud service QoS, compared with other Forecasting Methodologies, has the advantages that：
1. the research of existing service QoS predictions concentrates on Web service, does not consider the different resource service class such as physical equipment
Type, the present invention for cloud manufacturing environment the characteristics of, research have targetedly manufacture cloud service QoS Forecasting Methodologies, can be certain
The application that the field is filled up in degree is blank；
2. pair cloud manufacturing operation is classified, and the execution time of task is divided into into quiet hour and dynamic time, and analysis is not
The influence factor of same type task execution time, using influence factor as input the dynamic change portion of task execution time is predicted
Point, compared with conventional method, task execution time has more preferable estimated accuracy；
3. the present invention is predicted using Discrete Markov Model to the reliability for manufacturing cloud service, is had an advantage in that not
Need to consider impact of the various factors to service availability, by the analysis to historical data, state-transition matrix is calculated, to clothes
Business availability is predicted；
4. the availability and the availability of manufacturing equipment resource service that the present invention services computing resource separately considers, to meter
Calculate the availability of resource service, it is considered to two kinds of situations of network failure and service failure, carried out using continuous Markov model pre-
Survey；For manufacturing equipment resource service availability, in addition it is also necessary to consider the exclusivity of cloud service, be predicted using queuing model.
Description of the drawings
Fig. 1 is a kind of flow chart of the combination forecasting method and embodiment 1 of manufacture cloud service QoS of the present invention；
Fig. 2 is the manufacture cloud service body that a kind of combination forecasting method embodiment 2 of manufacture cloud service QoS of the present invention determines
Description graph of a relation；
Fig. 3 is the auto parts and components production work in a kind of combination forecasting method embodiment 1 of manufacture cloud service QoS of the present invention
Make flow chart；
Fig. 4 is to manufacture cloud service in a kind of combination forecasting method embodiment of manufacture cloud service QoS of the present invention to perform the time
Influence factor graph of a relation；
Fig. 5 is to calculate generic task dynamic time in a kind of combination forecasting method embodiment 2 of manufacture cloud service QoS of the present invention
BP neural network predicts structure chart；
Fig. 6 is manufacture processing generic task dynamic in a kind of combination forecasting method embodiment 2 of manufacture cloud service QoS of the present invention
Time BP neural network predicts structure chart；
Fig. 7 is the execution time of candidate service in a kind of combination forecasting method embodiment 2 of manufacture cloud service QoS of the present invention
The comparison diagram of predicted value, Smoothing Prediction value, gray prediction value and actual value；
Fig. 8 is candidate service RS in a kind of combination forecasting method embodiment 2 of manufacture cloud service QoS of the present invention3 1Execution
The comparison diagram of temporal predictive value, Smoothing Prediction value, gray prediction value and actual value；
Fig. 9 is candidate service RS1 1The comparison of reliability prediction value, gray prediction value and actual value；
Figure 10 is candidate service RS3 1The comparison of reliability prediction value, gray prediction value and actual value；
Figure 11 is candidate service RS1 1The comparison of availability predicted value, gray prediction value and actual value；
Figure 12 is candidate service RS3 1The comparison of availability predicted value, gray prediction value and actual value.
Specific embodiment
Below according to accompanying drawing, the present invention will be further described with embodiment 1.
Embodiment 1
This example illustrates the specific embodiment flow process of the method for the invention.Fig. 1 is the present invention " one kind manufacture cloud clothes
The flow chart of the combination forecasting method of business QoS ".
As seen from Figure 1, a kind of flow process of the combination forecasting method of manufacture cloud service QoS of the present invention is as follows：
Step A incoming task is asked；
Step B task requests are modeled；
The historical data of step C similar tasks is obtained；
The similar historical data of step D incoming task；
D.1 the historical data of execution time is input into, BP neural network is built, when predicts execution of the task in candidate service
Between；
D.2 the historical data of availability is input into, judges that whether task is to calculate generic task, and proceed as follows：
If D.2.1 calculating generic task, the Y in correspondence Fig. 1 then builds the Markov model of service and network, prediction
The availability of service；
D.2.2 if not calculating generic task, the N in correspondence Fig. 1 then builds the Markov model of service and network, structure
The queuing model of service is built, the availability of service is predicted.
D.3 the historical data of reliability is input into, Markov model is built, reliability of service is predicted.
Embodiment 2
So that certain automobile manufacturing enterprise produces certain part as an example, to each subtasks of task requests in corresponding candidate service
The execution time be predicted.
User task request modeling, specially：
Production task T is submitted to manufacture cloud service platform by the enterprise, and cloud manufacturing management system carries out Business Stream to task T
Journey is decomposed, and obtains subtask set T={ T1,T2,T3,T4,T5, here subtask modeling is carried out using ontology description language,
As shown in Figure 2.
As can be seen from Figure 2：The task of user's request is connected with relation, attribute and concept, and the expression of task can be by closing
System, attribute and concept are described.Relation can be divided into inheritance, component relationship and functional relationship；Attribute can be divided into shape category
Property, material properties and precision attribute, shape attribute can be divided into length, height and width again, and material properties can be divided into specification, type again
Number and performance, precision attribute can be divided into dimensional accuracy and surface accuracy again；Concept can be divided into mission number, task names and task
Type.Assume that each subtask can be by single component manufacture cloud service complete independently, the logical relation between each subtask is by Fig. 3
Shown digraph is given, wherein T1、T2To calculate generic task, T3、T4、T5Generic task is processed for manufacture, table 1 gives each height
The candidate service set of task.Here to subtask T1Candidate service RS1 1With subtask T3Candidate service RS3 1QoS enter
Row prediction.Due to T1To calculate generic task, then corresponding RS1 1To calculate class service；T3Process generic task for manufacture, then it is corresponding
RS3 1For manufacture processing class service.
The auto parts and components of table 1 manufacture subtask and its candidate service collection
Calculate subtask and called its candidate service historic task between similarity, computational methods are as follows：
Assume subtask TxThere is m attribute, then use Tx=[a1x,a2x,a3x...,amx] represent subtask TxM attribute
Value；sim(aix,aiy) represent task TxWith task TyBetween ith attribute similarity, according to the characteristics of manufacturing operation attribute,
Attributes similarity computational methods can be divided into 3 classes：
Numeric Attributes：a,b∈[A,B]
Boolean property：
Enumeration type attribute：Sim (a, b)=f (a, b), f (a, b) are an enumeration function, according to the feature of specific object
It is fixed.
According to attribute type, the similarity of each attribute between two tasks is calculated, then the Similarity Measure of two tasks
Method is as follows：
In formula (1), ωiFor subtask Tx、TyThe weight of ith attribute, can be set when actually used according to expertise
It is fixed.
Manufacture cloud service running time prediction, specially：
Assume subtask T1In its correspondence candidate service RS1 1On the historical data such as table for performing time and its influence factor
Shown in 2；Subtask T3In its correspondence candidate service RS3 1On the historical data for performing time and its influence factor it is as shown in table 3：
Table 2 services RS1 1Execution time and its historical data (unit of influence factor：Minute)
Table 3 services RS3 1Execution time and its historical data (unit of influence factor：Hour)
So subtask T1In service RS1 1On quiet hour τs：
τs=D/V=35 (2)
In formula (2), D is calculating generic task T1Amount of calculation, V be candidate service RS1 1The given process speed of place virtual machine
Degree, both can be obtained by cloud data center inquiry.
Subtask T3In service RS3 1On quiet hour τs：
In formula (3), τiFor subtask T3I-th data in k historical data of execution time.
Manufacture cloud service performs the influence factor of time as shown in figure 4, because different task performs the influence factor of time
Difference, its corresponding BP neural network forecast model is also different, by using perform the time influence factor as network inputs, can
With the Dynamic Execution time τ to taskdIt is predicted.Subtask T1、T3Dynamic time τdThe pre- geodesic structure of BP neural network point
Not as shown in Figure 5, Figure 6.Total predicted value of service execution time can be by formula τ=τs+τdIt is calculated.
Fig. 7 gives candidate cloud service RS1 1Execution time the inventive method predicted value, Smoothing Prediction value, grey are pre-
The comparison of measured value and actual value.In Fig. 7, abscissa represents the execution number of times of service, and ordinate represents the execution time of service, single
Position is minute.To candidate cloud service RS3 1The prediction of execution time, Fig. 8 gives the inventive method predicted value, Smoothing Prediction
The comparison of value, gray prediction value and actual value.In Fig. 8, abscissa represents the execution number of times of service, and ordinate represents holding for service
Row time, unit is hour.As can be seen that the inventive method is compared with exponential smoothing, grey method from Fig. 7, Fig. 8,
Its more closing to reality value that predicts the outcome, while the mean square error for predicting the outcome also illustrate that the superiority of the inventive method.It is right
In candidate service RS1 1, mean square error MSE=1.027 of the inventive method predicted value, the mean square error of exponential smoothing predicted value
MSE=7.35, mean square error MSE=5.42 of gray prediction；For candidate service RS3 1, the inventive method predicted value it is square
Error MSE=0.716, mean square error MSE=9.17 of exponential smoothing predicted value, mean square error MSE=of gray prediction
6.31。
Manufacture cloud service reliability prediction, specially：
Subtask T1In its correspondence candidate service RS1 1On execution historical data as shown in table 3, subtask T3In its correspondence
Candidate service RS3 1On execution historical data it is as shown in table 4.In table 3 and table 4, " 0 " represents service response failure, and " 1 " represents
Service normal response, performs record by the sequencing arrangement for being from left to right, from top to bottom.
Table 3 services RS1 1Reliability history data
Table 4 services RS3 1Reliability history data
According to the historical data in table 3, service RS can be tried to achieve1 1State-transition matrix P1, solution procedure is as follows：
In the same manner, the historical data in table 4 tries to achieve service RS3 1State-transition matrix P3For：
The reliability that subsequent time manufactures cloud service can be predicted by state-transition matrix, Fig. 9 gives time
Select cloud service RS1 1The comparison of reliability the inventive method predicted value, gray prediction value and actual value, wherein abscissa represent service
Execution number of times, ordinate represents reliability of service；Figure 10 gives candidate cloud service RS3 1Reliability the inventive method is predicted
The comparison of value, gray prediction value and actual value, wherein abscissa represents the execution number of times of service, and ordinate represents the reliability of service
Property.The mean square error of two methods prediction is calculated, for candidate service RS1 1, mean square error MSE=of the inventive method prediction
0.975, mean square error MSE=1.451 of gray prediction, for candidate service RS3 1, the mean square error of the inventive method prediction
MSE=1.565, mean square error MSE=2.414 of gray prediction method.
Manufacture cloud service availability prediction, specially：
The Failure Factors of manufacture cloud service are divided into two：One is service failure itself, and one is to service the network being located
Failure.Assume candidate service RS1 1Normal pot life and fault time as shown in table 4, take in nearest five days in nearest five days
The time good for use and fault time of business place network is as shown in table 5, candidate service RS3 1Normally can use in the middle of nearest five days
Time and fault time as shown in table 6, service the time good for use and fault time such as institute of table 7 of place network in nearest five days
Show.
Candidate service RS of table 41 1Normal pot life and fault time in the middle of nearest five days
Table 5 services RS1 1Normal pot life and fault time during place network is nearest five days
Candidate service RS of table 63 1Normal pot life and fault time in the middle of nearest five days
Table 7 services RS3 1Normal pot life and fault time during place network is nearest five days
The pot life of manufacture cloud service and fault time can be described with exponential distribution, according to table 4 and the system of table 5
Evaluation, it is possible to it is determined that service RS1 1The parameter of exponential distribution, so as to draw their distribution function.
It is assumed herein that time X good for use1Obedience parameter is μ1Exponential distribution, i.e.,
t≥0,μ1＞ 0, its distribution density is：
It is assumed herein that fault time Y1Obedience parameter is μ2Exponential distribution, i.e.,
t≥0,μ2＞ 0, its distribution density is：In order to determine ginseng
Number μ1And μ2Value, here parameter is estimated with Maximum Likelihood Estimation Method.
According to RS in table1 1Pot life and the statistics of fault time, take likelihood function：
Both members are taken the logarithm, and show that log-likelihood function is：
Order
Then λ is obtained1Maximum-likelihood estimation be：
Therefore, for RS1 1Pot life distribution, μ1Maximum-likelihood estimation be
Then RS1 1Annual distribution function good for use be：
Likewise, λ can be obtained2Maximum-likelihood estimation be：
Then RS1 1Time to failure distribution function be：
Therefore, RS1 1" available-failure " transition rates of service are：
Can obtain in the same manner, " available-failure " transition rates of network are：
Here, the original state of service and network is all available, according to the prediction public affairs derived in availability prediction
Knowable to formula, RS1 1At any time the availability of t is：
Can obtain in the same manner, candidate service RS3 1The availability of t at any time be：
Due to RS3 1It is manufacturing equipment resource service, so when availability is predicted, in addition it is also necessary to consider that manufacturing equipment resource takes
The exclusivity of business, i.e. service can only be taken by a task, be described with queuing model.
In order to predict the whether occupied probability of manufacturing equipment resource service, the clothes for knowing manufacturing equipment resource service are needed
Business rate and the arrival rate of user task request.It is assumed here that manufacturing equipment resource service RS3 1Service rate and user task request
Arrival rate it is as shown in table 8.
Table 8 services RS3 1Service rate and user arrival rate
Make piRepresent there is the probability of i request in service queue, then can be obtained by queuing theory：
In formula (12), b is to service the service request number (manufacture processing class service b=1) that most multipotency is received, when μ is unit
Interior treatable service number, i.e. service rate, λ is the number of users that the unit time reaches；ρ represents service intensity, can be byMeter
Obtain, p0Represent the probability that current queue is not asked.
Service intensity can be obtained by table 8Then manufacturing equipment resource service is currently without the probability asked：
Therefore, RS is serviced3 1Availability predictor formula be：
Figure 11 gives manufacture cloud service RS1 1The ratio of availability the inventive method predicted value, gray prediction value and actual value
Compared with Tu11Zhong, abscissa represents the execution number of times of service, and ordinate represents the availability of service；Figure 12 gives manufacture cloud clothes
Business RS3 1The comparison of the predicted value, gray prediction value and actual value of availability, Tu12Zhong, abscissa represents the execution number of times of service,
Ordinate represents the availability of service.Solid line represents availability actual value in Figure 11, Figure 12, and dotted line represents predicted value, actual situation line
Represent gray prediction value.Can be seen that from Figure 11, Figure 12, the inventive method compared with gray prediction value, more closing to reality value.
The mean square error of two methods prediction is calculated, for candidate service RS1 1, it is square that availability Forecasting Methodology of the present invention predicts the outcome
Error MSE=8.562, mean square error MSE=12.134 of gray prediction result；For candidate service RS3 1, availability of the present invention
Mean square error MSE=5.92 that Forecasting Methodology predicts the outcome, mean square error MSE=12.082 of gray prediction result.
From predicting the outcome and comparative analysis figure, the present invention performs time, reliability, availability to manufacturing cloud service
Predict the outcome and meet real data tendency, and predicated error relative to conventional method have be obviously improved, it was demonstrated that the present invention carried
Method has higher reasonability and science for the prediction that QoS is serviced in cloud manufacturing environment.
The above is presently preferred embodiments of the present invention, and the present invention should not be limited to the embodiment and accompanying drawing institute is public
The content opened.It is every without departing from complete equivalent or modification under spirit disclosed in this invention, both fall within the model of present invention protection
Enclose.
Claims (9)
1. a kind of combination forecasting method of manufacture cloud service QoS, it is characterised in that：Core technology thought is：Consider cloud manufacture ring
The features such as there is physical equipment access, resource dynamic change in border, initially with task requests of the ontology description language to user
It is modeled, by the similarity between calculating task, obtains the similar historical data of task；Secondly, using BP neural network
The execution time to manufacturing cloud service is predicted；Reliability of service is predicted using Discrete Markov Model；Adopt
The availability for servicing is predicted with continuous Markov model, finally gives predicting the outcome for manufacture cloud service QoS.
2. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 1, it is characterised in that：Mainly include
Following four step：
Step one, the modeling of cloud manufacturing operation and the acquisition of similar historical data；
Step 2, manufacture cloud service running time prediction；
Step 3, manufacture cloud service reliability prediction；
Step 4, manufacture cloud service availability prediction；
So far, from step one to step 4, a kind of combination forecasting method of manufacture cloud service QoS is completed.
3. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 2, it is characterised in that：Step one, tool
Body is：
Step 1.1 utilizes ontology description language (OWL-S), and the task requests that user's Xiang Yun manufacturing platforms are submitted are described；
Step 1.2 by calculating user task and calling the similarity between the historic task for servicing, is chosen similarity and is existed again
Historic task in certain threshold range, referred to as similar historical task；
Step 1.3 uses the qos value of the similar historical task that step 1.2 exports as the similar historical data of user task, obtains
To the similar historical data acquisition system of user task.
4. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 2, it is characterised in that：Step 2, tool
Body is：
Different types of user task is analyzed in the cloud manufacturing environment that step 2.1 is exported to step 1.3, is classified as calculating
Generic task and manufacture processing generic task；
The influence factor of different type task execution time, by task execution time static state is divided in step 2.2 analytical procedure 2.1
Time and dynamic time, and calculate quiet hour and prediction dynamic time.
5. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 4, it is characterised in that：In step 2.2
Quiet hour, be designated as：τs；Dynamic time, is designated as：τd, task execution time is designated as τ, and the expression of task execution time is as public
Formula (1)：
τ=τs+τd (1)
Wherein, the quiet hour is by being calculated；Dynamic time is, using the influence factor of task execution time as input, to pass through
Prediction BP neural network is obtained；
Specifically, for the quiet hour calculating process point for calculating generic task and manufacture processing generic task of step 2.1 output
It is not：
For generic task is calculated, this generic task is finally performed on a virtual machine, quiet hour τsComputing formula such as formula (2)：
τs=D/V (2)
Wherein, D is the amount of calculation of virtual task, and V is the processing speed that virtual machine gives；
For the quiet hour of manufacture processing generic task, τ is averagely worth to by calculating historical datas, with equation below (3)
Calculate：
Wherein, τiFor the historical data that manufacture processing generic task performs the time, n is the number of historical data；
It is respectively for the dynamic time calculating process for calculating generic task and manufacture processing generic task of step 2.1 output：
For generic task is calculated, it is cpu load, internal memory that impact task performs on a virtual machine the principal element of time change
Size and the network bandwidth；Cpu load, memory size, the network bandwidth, the historical data of execution time is neural to BP as sample
Network is trained, can be to dynamic time τdIt is predicted；
For manufacture processing generic task, affect task to perform the principal element of time change in manufacturing equipment for personnel because
Element, apparatus factor and environmental factor；These factors and execution time are trained as sample to BP neural network, can be right
Dynamic time τdIt is predicted；Wherein, human factor can be divided into the working experience of personnel and the professional skill level of personnel；Equipment
Factor can be divided into the maintenance levels of equipment and the crash rate of equipment；Environmental factor can be divided into temperature and humidity.
6. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 2, it is characterised in that：Step 3, tool
Body is：
The historical data of step 3.1 pair manufacture cloud service reliability is analyzed, can to service using Discrete Markov Model
It is described by property；
Step 3.2 calculates state-transition matrix, and the reliability to manufacturing cloud service is predicted.
7. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 6, it is characterised in that：In step 3,
Respectively equation below (4) and formula (5) are shown for the state-transition matrix expression of Discrete Markov Model and calculating：
In formula (5), E0Represent service response failure, E1Represent service normal response；P00For the probability that service keeps failure, P01For
Service is changed into available probability, P from failure10To service from the available probability for being changed into and failing, P11Keep available general for service
Rate；N is to service the number of times for keeping failure, N1The total degree of service failure is represented, m keeps available number of times, N for service2For service
Available total degree.
8. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 2, it is characterised in that：Step 4, tool
Body is：
The Failure Factors for manufacturing cloud service are divided into two by step 4.1：One is service failure itself, and one is that service is located
Network failure；
The continuous Markov model of step 4.2 is described respectively to service failure and network failure, and cloud service is only in clothes
It is called in the case of two kinds of business availability and network availability, to respond other service requests；
On this basis, manufacturing equipment class cloud service also needs to consider the exclusivity of service, i.e. synchronization, manufacture step 4.3
The cloud service of equipment class can only respond a service request, if current cloud service is occupied, next service request will be refused
Absolutely, this process is described with queuing model.
9. the combination forecasting method of a kind of manufacture cloud service QoS according to claim 8, it is characterised in that：In step 4,
The calculating of service availability includes the following two kinds situation：
If 4.2A service original state be available, i.e. service availability, being designated as P, by formula (6) calculating：
If 4.2B service original states are failure, service availability P is calculated by formula (7)：
In formula (6), PserFor service availability, PnetFor network availability；μ1For the crash rate of service, λ1Failure for service is extensive
Multiple rate, μ2For the crash rate of network, λ2For the fault recovery rate of network；
For manufacture processing class service, the characteristics of considering manufacturing equipment, i.e. equipment tool are also needed to when its availability is calculated
There is exclusivity, equipment can not be called during a task is performed by other task requests, this puts and calculates class service
It is different；
For manufacture processing class service, the service request number that its most multipotency is received is b, treatable service in the unit interval
Number, i.e. service rate μ, the number of users that the unit interval reaches is λ, then its service regulation can be described as：First Come First Served, after
The person that has come needs to queue up successively in the buffer.This process can be described with a queuing model；
Make piRepresent there is the probability of i request in queue, then can be obtained by queuing theory：
Wherein,Service intensity is represented, in general ρ ＜ 1.p0Represent the probability that current queue is not asked；
For manufacture processing class service, the computing formula of service availability P is：
P=Pser(t)*Pnet(t)*P0 (9)。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611147088.5A CN106600058A (en) | 2016-12-13 | 2016-12-13 | Prediction method for combinations of cloud manufacturing service quality of service (QoS) |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611147088.5A CN106600058A (en) | 2016-12-13 | 2016-12-13 | Prediction method for combinations of cloud manufacturing service quality of service (QoS) |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106600058A true CN106600058A (en) | 2017-04-26 |
Family
ID=58802027
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611147088.5A Pending CN106600058A (en) | 2016-12-13 | 2016-12-13 | Prediction method for combinations of cloud manufacturing service quality of service (QoS) |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106600058A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108009023A (en) * | 2017-11-29 | 2018-05-08 | 武汉理工大学 | Method for scheduling task based on BP neural network time prediction in mixed cloud |
| CN108038612A (en) * | 2017-12-08 | 2018-05-15 | 温州大学 | A kind of the manufacture cloud service self-organizing evaluation system and method for user behavior driving |
| CN108170404A (en) * | 2017-12-25 | 2018-06-15 | 南京航空航天大学 | A kind of Web service combination verification method based on parameterized model |
| CN108984301A (en) * | 2018-07-04 | 2018-12-11 | 中国人民解放军国防科技大学 | Self-adaptive cloud resource allocation method and device |
| CN109167673A (en) * | 2018-07-18 | 2019-01-08 | 广西大学 | A kind of Novel cloud service screening technique of abnormal fusion Qos Data Detection |
| CN109615081A (en) * | 2018-09-26 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of Model forecast system and method |
| CN109981749A (en) * | 2019-01-30 | 2019-07-05 | 北京理工大学 | A kind of cloud workflow task running time prediction method promoted based on limit gradient |
| CN111083743A (en) * | 2019-12-31 | 2020-04-28 | 上海无线通信研究中心 | Distributed QoS prediction method, system and device based on 5G access network |
| WO2021024076A1 (en) * | 2019-08-05 | 2021-02-11 | International Business Machines Corporation | Automated operational data management dictated by quality-of-service criteria |
| CN112462720A (en) * | 2020-12-05 | 2021-03-09 | 宜昌湖蓝科技开发有限公司 | Process module standardization system for large-scale and customized production of clothing tools |
| US11277317B2 (en) | 2019-08-05 | 2022-03-15 | International Business Machines Corporation | Machine learning to predict quality-of-service needs in an operational data management system |
-
2016
- 2016-12-13 CN CN201611147088.5A patent/CN106600058A/en active Pending
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108009023A (en) * | 2017-11-29 | 2018-05-08 | 武汉理工大学 | Method for scheduling task based on BP neural network time prediction in mixed cloud |
| CN108009023B (en) * | 2017-11-29 | 2022-06-03 | 武汉理工大学 | Task scheduling method based on BP neural network time prediction in hybrid cloud |
| CN108038612A (en) * | 2017-12-08 | 2018-05-15 | 温州大学 | A kind of the manufacture cloud service self-organizing evaluation system and method for user behavior driving |
| CN108038612B (en) * | 2017-12-08 | 2021-06-15 | 温州大学 | User behavior driven manufacturing cloud service self-organizing evaluation system and method |
| CN108170404B (en) * | 2017-12-25 | 2021-04-06 | 南京航空航天大学 | Web service combination verification method based on parameterized model |
| CN108170404A (en) * | 2017-12-25 | 2018-06-15 | 南京航空航天大学 | A kind of Web service combination verification method based on parameterized model |
| CN108984301A (en) * | 2018-07-04 | 2018-12-11 | 中国人民解放军国防科技大学 | Self-adaptive cloud resource allocation method and device |
| CN109167673A (en) * | 2018-07-18 | 2019-01-08 | 广西大学 | A kind of Novel cloud service screening technique of abnormal fusion Qos Data Detection |
| CN109167673B (en) * | 2018-07-18 | 2021-04-02 | 广西大学 | Novel cloud service screening method integrating abnormal Qos data detection |
| CN109615081A (en) * | 2018-09-26 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of Model forecast system and method |
| CN109981749A (en) * | 2019-01-30 | 2019-07-05 | 北京理工大学 | A kind of cloud workflow task running time prediction method promoted based on limit gradient |
| WO2021024076A1 (en) * | 2019-08-05 | 2021-02-11 | International Business Machines Corporation | Automated operational data management dictated by quality-of-service criteria |
| US11277317B2 (en) | 2019-08-05 | 2022-03-15 | International Business Machines Corporation | Machine learning to predict quality-of-service needs in an operational data management system |
| CN114208127A (en) * | 2019-08-05 | 2022-03-18 | 国际商业机器公司 | Machine learning to predict quality of service in an operational data management system |
| US11310126B2 (en) | 2019-08-05 | 2022-04-19 | International Business Machines Corporation | Automated operational data management dictated by quality of service criteria |
| GB2602213A (en) * | 2019-08-05 | 2022-06-22 | Ibm | Automated operational data management dictated by quality-of-service criteria |
| GB2602213B (en) * | 2019-08-05 | 2023-03-29 | Ibm | Automated operational data management dictated by quality-of-service criteria |
| CN111083743A (en) * | 2019-12-31 | 2020-04-28 | 上海无线通信研究中心 | Distributed QoS prediction method, system and device based on 5G access network |
| CN112462720A (en) * | 2020-12-05 | 2021-03-09 | 宜昌湖蓝科技开发有限公司 | Process module standardization system for large-scale and customized production of clothing tools |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600058A (en) | Prediction method for combinations of cloud manufacturing service quality of service (QoS) | |
| CN110737529B (en) | Short-time multi-variable-size data job cluster scheduling adaptive configuration method | |
| Wang | A fuzzy project scheduling approach to minimize schedule risk for product development | |
| Saliu et al. | Supporting software release planning decisions for evolving systems | |
| Nembhard | Heuristic approach for assigning workers to tasks based on individual learning rates | |
| CN104765642B (en) | Virtual machine deployment method and system based on dynamic prediction model under cloud environment | |
| Chakravarthi et al. | TOPSIS inspired budget and deadline aware multi-workflow scheduling for cloud computing | |
| CN109726926A (en) | Machine tool equipment resource supply and demand matching process based on Grey Relation Algorithm under a kind of constraint of multivariate quality | |
| Tuli et al. | GOSH: Task scheduling using deep surrogate models in fog computing environments | |
| Chen | An intelligent hybrid system for wafer lot output time prediction | |
| CN110209467A (en) | A kind of flexible resource extended method and system based on machine learning | |
| Liu et al. | A deep multi-agent reinforcement learning approach to solve dynamic job shop scheduling problem | |
| Mazidi et al. | An autonomic decision tree‐based and deadline‐constraint resource provisioning in cloud applications | |
| Aytug et al. | Measures of subproblem criticality in decomposition algorithms for shop scheduling | |
| Chen | Intelligent scheduling approaches for a wafer fabrication factory | |
| Liu | Scheduling via reinforcement | |
| Puliafito et al. | Systems Modelling: Methodologies and Tools | |
| Smilgyte et al. | Artificial neural networks application in software testing selection method | |
| Liu et al. | An intelligent system for estimating full product Life Cycle Cost at the early design stage | |
| Schönmann et al. | Anticipative strategic production technology planning considering cyclic interactions | |
| Stojanov et al. | Evaluating software maintenance processes in small software company based on fuzzy screening | |
| Thakkar et al. | Naarprec: A novel approach for adaptive resource prediction in cloud | |
| CN112257977B (en) | Logistics project construction period optimization method and system with resource limitation under fuzzy man-hour | |
| Bi et al. | Adaptive prediction of resources and workloads for cloud computing systems with attention-based and hybrid LSTM | |
| Liu et al. | Task allocation optimization model in mechanical product development based on Bayesian network and ant colony algorithm |
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: 20170426 |