CN108646684B - Multi-product production line production period prediction method based on variability measurement - Google Patents

Multi-product production line production period prediction method based on variability measurement Download PDF

Info

Publication number
CN108646684B
CN108646684B CN201810538128.1A CN201810538128A CN108646684B CN 108646684 B CN108646684 B CN 108646684B CN 201810538128 A CN201810538128 A CN 201810538128A CN 108646684 B CN108646684 B CN 108646684B
Authority
CN
China
Prior art keywords
station
product
logic
logical
link
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810538128.1A
Other languages
Chinese (zh)
Other versions
CN108646684A (en
Inventor
李波
张金彬
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201810538128.1A priority Critical patent/CN108646684B/en
Publication of CN108646684A publication Critical patent/CN108646684A/en
Application granted granted Critical
Publication of CN108646684B publication Critical patent/CN108646684B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
    • G05B19/41885—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by modeling, simulation of the manufacturing system
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/32—Operator till task planning
    • G05B2219/32339—Object oriented modeling, design, analysis, implementation, simulation language
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Landscapes

  • Engineering & Computer Science (AREA)
  • Manufacturing & Machinery (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • General Factory Administration (AREA)

Abstract

The invention belongs to the technical field of production line management, and provides a variability measurement of a production line with the characteristic of series-parallel series under the condition of multi-product mixed production and a corresponding product-oriented production period prediction method. The method comprises the following steps: extracting and processing original data, historical operating data and future cycle planning data of each work station of the production line; constructing a logical work station of the product according to the condition that the product actually passes through link, and connecting the logical work stations in series to obtain a logical production line of the product; calculating the variability of each logic station on the product logic production line and the variability of transmission among the logic stations according to the basic data and the logic production line composition; and calculating the queuing time of each logic station in the future period based on a VUT equation, and cumulatively adding the queuing time and the effective processing time of each logic station to obtain the overall production period of the product logic production line.

Description

Multi-product production line production period prediction method based on variability measurement
Technical Field
The invention belongs to the technical field of production line management, and particularly relates to production cycle prediction under a multi-product mixed production condition.
Background
The Cycle Time (CT) is one of the key performance indicators of a product (e.g., semiconductor chip package test) production line, and refers to the average of the Time spent in the production line by a batch of products as Work In Process (WIP). The production cycle is a key parameter for making a production plan, if the CT of the production line can be accurately predicted, the made production plan is more reasonable, and the production efficiency is correspondingly improved. The semiconductor chip packaging test production line comprises a plurality of working procedures, each working procedure corresponds to a corresponding work station (Workstation), and products as products in process flow through the work stations in sequence according to a process route for processing. The dynamic uncertain factors including faults, mold change and the like exist in a machining Process of a work station, the machining Time of the work station can fluctuate due to the dynamic uncertain factors, the machining Time considering the dynamic uncertain factors of the work station is called Effective machining Time (EPT), the fluctuation phenomenon of the Effective machining Time is called variability of the work station, even if the average machining Time of each work station on a production line is equal and the feeding rate is equal to the average machining Time, a product can still experience waiting Time before the work station due to the variability, and the prediction of a production period is more complicated due to the addition of the waiting Time. The quantitative relation among the waiting time, the variability (V), the utilization rate (U) and the effective processing time (T) is well described by a VUT equation provided by factory physics, the waiting time before each product station can be accurately calculated as long as the variability of the stations and the variability of the transmission among the stations can be accurately measured, and finally, the production period of the whole production line can be obtained by accumulating the queuing time and the processing time of each station.
Referring to fig. 1, the inside of a station of a semiconductor chip package test production line is generally composed of a plurality of links connected in parallel, the links are composed of a plurality of devices with variable parameters connected in series, and different types of products can be processed by changing the parameters of the same link. In actual operation, the same station in the same production cycle may mix and pass through multiple products. The equipment composition characteristics of multi-product mixed production and series-parallel-series enable effective processing time composition of a work station to become very complex, and further measurement complexity of work station variability is caused.
Although, in the patent application with application number 200810201503.X, entitled "production cycle target measurement and system", a method for predicting (or measuring) a production cycle by directly fitting the relationship between utilization rate and queuing time is disclosed, which can also achieve the goal of production cycle prediction, it is modeled only in consideration of utilization rate parameters, so that the method will fail once variability changes (product and link changes) between production cycles; in addition, the scheme does not distinguish the period of each product under the condition of multi-product mixed flow, so that the production period of the whole of various products can be calculated. Namely, the work station is completely regarded as a black box, so that the specific influence of each parallel device (device corresponding link) in the work station and different products passing through the same work station on the production period cannot be distinguished, and direction guidance with finer granularity cannot be provided for further performance optimization.
Disclosure of Invention
The invention aims to: aiming at the difficulties and the defects of the existing production period prediction method, the station variability measurement under the conditions of multi-product mixing and multi-equipment parallel connection is realized through mechanism analysis, the production period prediction of the production line with the characteristics is further realized based on a VUT equation, and meanwhile, an analysis basis is provided for the optimization of the manufacturing system.
The technical scheme of the invention is as follows: a multi-product production line production cycle prediction method based on variability measurement comprises the following steps:
step 1: extracting and processing data of each physical station on the production line, wherein the processing data comprises original performance data, historical shutdown data, historical production scheduling data and production scheduling data of a future period;
step 2: actually passing the product through links on each physical work station to form a logical work station of the product on the current physical work station, and connecting the logical work stations of the same product in series to obtain a logical production line of the product;
and step 3: based on the processing data of the physical work station corresponding to each logic work station, measuring the effective processing time and the variability of each logic work station on the logic production line of each product: firstly, measuring the effective processing time and the variability corresponding to each link of each logic station; then constructing a regression model to fuse the effective processing time and the variability of all links of each logic station to obtain the effective processing time and the variability of each logic station;
and 4, step 4: and calculating the queuing time of each logic station about different products in the future period based on a VUT equation, and iteratively accumulating the queuing time and the processing time on the logic production line to obtain the overall production period of each product.
The invention provides a multi-product production line production period prediction method based on variability measurement, and provides a method for converging processing variability of multiple products on a single link based on a proportional weighting mode, and the method can calculate the link processing variability under any product combination based on the respective basic variability of each product after the products produced in a mixed mode change, thereby solving the variability measurement problem under the condition of multi-product mixed flow; the regression model is introduced to depict the rule that the processing variability of a plurality of links connected in parallel in the station converges to the overall processing variability of the station, so that the variability measurement of the parallel link structure is realized, the variability measurement can adapt to the variation of the link structure, and the problem of the variability measurement of the parallel links is solved; and then parameter extraction can be carried out by combining logic station division, and the prediction of the production period of a specific product under the condition of multi-product mixed flow can be realized through a VUT equation.
Further, in step3, the measuring of the effective processing time and the variability of each product corresponding to each link of each logic station specifically includes:
i represents a logical station identifier, j represents a link identifier, t represents a product type identifier, and the jth link of the logical station i and the tth product p are calculated based on the shutdown data type of the linktEffective time t ofeijtAnd effective machining time standard deviation sigmaeijt：
If the shutdown data of the current link only comprises preemptive shutdown data, then the formula is followed
Figure GDA0002636051520000031
Calculating the effective processing time teijtThen according to
Figure GDA0002636051520000032
Calculating the corresponding effective machining time standard deviation sigmaeijt；
If the shutdown data of the current link only includes non-preemptive shutdown data, then the formula is followed
Figure GDA0002636051520000033
Calculating the effective processing time teijtThen according to
Figure GDA0002636051520000034
Calculating the corresponding effective machining time standard deviation sigmaeijt；
If the shutdown data for the current link includes preemptive shutdown data and non-preemptive shutdown data, then the formula is followed
Figure GDA0002636051520000035
Calculating the effective processing time teijtThen according to
Figure GDA0002636051520000036
Calculating the corresponding effective machining time standard deviation sigmaeijt；
Wherein, t0ijtJ link representing logical station i to t product ptTheoretical working time of (1), wherein ptBelong to the product set linkijCorresponding set of classes, linkijA set of products representing the jth link that passes through logical station i; c0ijtJ link representing logical station i to t product ptCoefficient of variation of the theoretical processing time of (1); sigma0ijt＝C0ijt×t0ijt(ii) a Parameter Aij＝mfij/(mfij+mrij) Wherein m isfijMean value of down time intervals, m, caused by failure of jth link representing logical station irijRepresenting the mean time interval of fault repair of the jth link of the logic work station i; sigmarijThe time interval standard deviation of fault repair of the j link of the logic work station i is represented; wherein t issijMean value of the moulding-change time intervals, σ, of the j-th link representing the logical station isijVariance, N, of the modulo time interval representing the j link of logical station isijRepresenting the average number of processed products between two die changes of the jth link of the logic work station i;
fusing multiple products to obtain the effective processing time of each product on the j link of the logic station i
Figure GDA0002636051520000041
And mobility
Figure GDA0002636051520000042
Wherein
Figure GDA0002636051520000043
Figure GDA0002636051520000044
k is a product identifier, M represents the number of product categories passing through the current link, r0tDenotes the t product ptFeeding ratio of (1), nitDenotes the t product ptNumber of parallel links, m, of passing logical station iiktDenotes the t product ptAnd the kth product pkNumber of links passing simultaneously at logical station i, where pkE P, P represents the product set produced by mixing.
Further, the regression model specifically includes: y isγ＝fγ(x) W' x + e, where x ═ x (x)1x2x3x4x5x6x7) ' representing a feature vector, x1Representing the effective processing time of all links of the current logical station
Figure GDA0002636051520000045
Mean value of (1), x2Representing the effective processing time of all links of the current logical station
Figure GDA0002636051520000046
Variance of (a), x3Representing the volatility of all links of the current logical station
Figure GDA0002636051520000047
Root mean square, x4Representing the effective processing time of all links of the current logical station
Figure GDA0002636051520000048
Maximum value of (a), x5Representing the effective processing time of all links of the current logical station
Figure GDA0002636051520000049
Minimum value of (1), x6Representing the volatility of all links of the current logical station
Figure GDA00026360515200000410
Maximum value of (a), x7Representing the volatility of all links of the current logical station
Figure GDA00026360515200000411
Minimum value of (d); symbol (·)' denotes transposition, w denotes coefficient vector, e denotes error vector, y denotesγA subscript γ of 1,2, wherein y1Representing the effective processing time after the integration of the logical station as a whole, i.e. y1Effective processing time for arbitrary logical work station of product
Figure GDA00026360515200000412
y2Representing the variability of the logical station after the fusion of the whole, i.e. y2Being any logical work station of a product
Figure GDA00026360515200000413
Further, step4 specifically includes:
step 401: calculating expected utilization of future production cycles
Figure GDA00026360515200000414
Wherein u isiRepresents the expected utilization of the logical station i, rtIs the rate of feed, rbEffective processing rate, r, for a logical production line bottleneck stationeiEffective processing Rate for logical station i, miThe number of the parallel links of the logic station i is shown;
step 402: and calculating the queuing time of each product before each logic station based on a VUT equation, adding the effective processing time to obtain the production period of a single logic station, and circularly executing the step on each logic station and accumulating the production period to obtain the overall predicted production period of the logic production line.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
(1) the invention explicitly models the variability factors such as multi-product and parallel link structures, so that the variability after strain can be measured after the production product and the link structure are changed, and the production period calculation under the corresponding condition is realized.
(2) The method provided by the invention can be used for calculating the production period of a specific product under the condition of multi-product mixed flow, and better meets the actual requirements of a semiconductor chip packaging test production line.
Drawings
FIG. 1 is an internal block diagram of a single physical station of a semiconductor chip package test line;
FIG. 2 is a block diagram of the method of the present invention;
FIG. 3 is a flow chart of an embodiment of the method of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
The invention adopts Coefficient of Variation (CV) of effective processing time as a variability measurement index, predicts the production cycle of a multi-product production line based on the index, and can be realized by constructing different processing modules when the production cycle prediction method is realized. Referring to fig. 2, in order to implement the production cycle prediction method of the present invention, 4 processing modules are constructed, including: an input data processing module, a product logic production line construction module, a variability measurement module and a production cycle prediction module, referring to fig. 3, the specific execution process of each module is as follows:
(1) an input data processing module: and extracting and processing the processing data of each physical work station of the production line, and transmitting the extraction and processing results to a product logic production line construction module.
The processing data includes raw performance data (raw data for short), historical shutdown data, historical production schedule data, and production schedule data for future cycles.
The original data are theoretical parameters for processing a certain product by each link of a physical station, and the theoretical parameters comprise: theoretical processing time, coefficient of variation of theoretical processing time; the historical shutdown data is the shutdown record data of each link and is divided into preemptive shutdown data and non-preemptive shutdown data. Preemptive shutdown data such as: a sequence of downtime and repair events; non-preemptive shutdown data such as: a die change stopping interval sequence and a die change processing product quantity sequence data; the production schedule data refers to the condition that each product passes through links of each physical work station, namely the number of links occupied by each product, and is divided into historical production schedule data and production schedule data of a future period.
(2) A product logic production line construction module: and according to the collected actual link condition of the product, constructing a logic station of the product, connecting the logic stations in series to obtain a logic production line of the product, and transmitting the logic production line to the variability measurement module and the production period prediction module.
For a particular product pkThe corresponding logical station on the ith physical station is represented as a link set of the product passing through the physical station, and can be specifically represented as formula (1),
wik＝{lis|s∈Sik} (1)
wherein lisThe S-th link, S representing the ith physical stationikRepresents pkSet of link numbers passing through physical station i.
Based on this, product pkThe logical production line of (a) can be represented by a vector as equation (2),
plk＝(w1k,w2k,...,wnk) n∈W (2)
for the j-th link of the ith logical station (for a specific product, there is only one logical station on the physical station, so the station identifier can be used in common), the collection of products processed by the j-th link can be represented by formula (3),
linkij＝{pk} pk∈P,i∈W,j∈Li(3)
wherein p iskDenotes a product, subscript is a product identifier, P denotes a product set of a mixed production, W denotes a physical station number set of a physical production line, LiAnd a link number set representing the ith physical station.
(3) A variability metric module: and measuring the variability and the effective processing time of each logic station on the product logic production line and the variability transmitted among the logic stations, and transmitting the measurement result to the production cycle prediction module.
For a specific link and a specific product of a certain logic station, the theoretical variability is the coefficient of variation of the theoretical processing time, and the measurement formula is (4):
Figure GDA0002636051520000061
in the formula (4), t0To theoretical working time, σ0Standard deviation of theoretical processing time, and C0、t0The method can be directly obtained based on theoretical parameters of certain products processed by each link in the original data of the physical work station.
Step 301: coefficient of variation C based on theoretical processing time0Theoretical machining time t0And standard deviation of theoretical machining time sigma0And measuring the corresponding effective processing time and the corresponding variability of each link by combining the shutdown data type of each link.
Because the shutdown data of the link is divided into preemptive shutdown data and non-preemptive shutdown data, the invention adopts a corresponding measurement mode to obtain the effective processing time and the variability of a specific link of a certain logic station about a specific type of products according to the type of the shutdown data included in the link.
(1) For preemptive volatility, taking the shutdown volatility caused by a typical fault as an example, the measurement flow of the volatility is as follows: step 1: carrying out statistics according to the fault shutdown and repair event sequence to obtain basic parameters shown in the table 1;
TABLE 1 Fault shutdown variability metrics basic parameters
Figure GDA0002636051520000071
Step 2: calculating an availability factor parameter A according to a formula (5);
Figure GDA0002636051520000072
step 3: calculating the effective processing time considering the fault shutdown factor according to a formula (6);
Figure GDA0002636051520000073
step 4: according to the formula (7), the corresponding preemptive mobility C is calculated by combining the theoretical mobility measured by longitudep；
Figure GDA0002636051520000074
(2) For non-preemptive volatility, taking the typical shutdown caused by mode change as an example, the volatility metric flow is: step 1: obtaining the basic parameters shown in the table 2 according to the statistics of the die change event sequence;
TABLE 2 Change of mold shut-down variability measurement basic parameters
Figure GDA0002636051520000075
Step 2: calculating the effective processing time considering the die change shutdown factor according to a formula (8);
Figure GDA0002636051520000076
step 3: according to the formula (9), the corresponding non-preemptive variability C is calculated by combining the theoretical variability measured by longitude and the processing timenp：
Figure GDA0002636051520000081
(3) For the case that both the preemptive and non-preemptive changeability exist, the effective processing time t of the preemptive is calculated firsteAnd variability Cp(i.e., calculated according to equations 6 and 7, respectively), and then apply teAnd CpSubstituting the theoretical machining time and the theoretical variability into the formulas (8) and (9) can obtain the effective machining time and the variability fusing the preemptive shutdown and the non-preemptive shutdown, namely obtaining the measure of the variability fusing the preemptive shutdown and the non-preemptive shutdown
Figure GDA0002636051520000082
Wherein
Figure GDA0002636051520000083
And is composed of
Figure GDA0002636051520000084
The effective processing time for fusing preemptive and non-preemptive shutdowns is obtained.
Step 302: applying the process of the previous step to all the products on all the links of any logic station i, the effective processing time t of each link about each product can be obtainedeijtAnd variability Ceijt(corresponding to the above-mentioned three modes of preemptive, non-preemptive and preemptive-non-preemptive fusion, the volatility Cp、Cnp、Cfuse). This step will pool and blend the variability and effective processing time of multiple products passing through a single link in a weighted average manner based on the proportion of each product on the link to obtain the effective processing time and variability of a particular link with respect to a particular product logic station.
For a link described by equation (3), the aggregate equation for its multi-product variability and effective processing time is shown by equation (10):
Figure GDA0002636051520000085
wherein the content of the first and second substances,
Figure GDA0002636051520000086
respectively representing jth link and tth product p representing logic work station itWherein the superscript i is a physical station (logical station) identifier, k is a product identifier, the subscript j is a link identifier, and linkijRepresents the set of all products, p, that pass through the jth link of the ith logical stationtFor the t product produced for link mix, M represents the number of product categories passing through the current link, teijtEffective processing time, σ, of jth link representing ith logical station with respect to tth producteijtEffective processing time scale of jth link for ith logic station relative to tth productTolerance, and σeijt＝teijt×Ceijt，CeijtRepresents the variability, r, of the jth link of the ith logical station with respect to the tth product0tIs ptOriginal feed ratio of the product, nitThe number of parallel links of t products passing through the logic station i is called the parallelism of the products relative to the logic station, and the formula (11) is shown based on the formula (2).
nit＝Count{wit} (11)
Wherein, the Count operator represents the number of the elements of the calculation set. At the same time, miktRepresents ptProduct pass pkThe number of links in the logic work station corresponding to the product is similarly expressed by the formula (12):
mikt＝Count{wik∩wit} (12)
step 303: aiming at a specific logic work station, fitting the rule that the variability of different links in the logic work station converges to the overall variability of the logic work station by establishing a linear regression model; and measuring the whole variability of the logic station based on the trained model. Table 3 shows the input and output of the regression model, and the corresponding signs and calculation methods.
TABLE 3 regression model input and output parameters
Figure GDA0002636051520000091
Figure GDA0002636051520000101
The corresponding regression model is shown in equation (13).
yγ＝fγ(x)＝w'x+e,γ＝1,2 (13)
Wherein, yiThe subscript i ═ 1,2, and x ═ x (x)1x2x3x4x5x6x7) ' denotes a feature vector, sign (·) denotes transposition, w is a coefficient vector, and e denotes an error vector. The above model can be trained using least squares (training objective: minimizing error vector e), and mostThe small-two-times algorithm is shown in equation (14).
w*＝argmin(y-Xw)'(y-Xw) (14)
Where X represents the feature vector of the sample, y ∈ { y }1,y2}。
Step 304: the variability can generate influence in the work station and can flow among the work stations, the variability of the effective processing time of the previous logic work station can be transferred to the downstream logic work station in the departure time interval variability and directly becomes the arrival time interval variability of the downstream logic work station, so the invention uses the formula (15) to measure the variability transfer rule:
Figure GDA0002636051520000102
wherein, Ca(. cndot.) denotes the arrival variability of the station in brackets, i.e. Ce(i) Expressing the arrival variability of the logic station i, namely the variability of the logic station after the whole integration according to the formula (13); ca(i +1) represents the arrival variability of the downstream station of the logical station i, uiRepresents the expected utilization rate, m, of the station iiRepresenting the number of parallel devices (or parallel links) of station i.
(4) A production cycle prediction module: and calculating the queuing time of each logic work station in the future period based on a VUT equation, and iteratively summing the queuing time and the effective processing time on the logic production line to obtain the overall production period of the product.
Step 401: the expected utilization estimate for the future production cycle, the utilization of the production line, can be calculated using equation (16).
Figure GDA0002636051520000103
Wherein u isiRepresents the expected utilization of the logical station i, rtIs the rate of feed, rbEffective processing rate, r, for a logical production line bottleneck stationeiEffective processing Rate for logical station i, miAnd (5) connecting the number of links in parallel for the logic work station i.
Step 402: calculating the queuing time of each product before each logic station based on a VUT equation, adding effective processing time to obtain the production period of a single logic station, circularly executing the step on each logic station and accumulating the production period to obtain the overall predicted production period of the logic production line, wherein the specific execution flow is as follows:
step 1: initializing an identifier i of 1 for the logical station, initializing an inter-arrival time variability CaI.e. initialise Ca(1) Value of (C), typically empirically initializedaIn general, C may be seta(1)＝0；
Step 2: calculating the arrival time interval variability C based on the stepsa(for the first station, the preset value is directly obtained, and for the non-first station, the calculation is carried out according to the formula (15)), and the mobility CeEffective processing time teAnd the expected utilization rate u.
Step 3: and calculating the queuing time before the logic work station by using a VUT equation, wherein the VUT equation is shown as an equation (17).
Figure GDA0002636051520000111
For any logical station i, any product pkk, the product pkQueuing time for logical station i
Figure GDA0002636051520000112
Comprises the following steps:
Figure GDA0002636051520000113
wherein
Figure GDA0002636051520000114
Representing logical stations i with respect to products pkThe effective processing time of the process is reduced,
Figure GDA0002636051520000115
representing logical stations i with respect to products pkVariability of (u)iRepresents the expected utilization of the logical station i, miRepresenting the number of parallel links passing through the logical station i.
Step 4: calculating the production cycle of a single logical station according to equation (18):
Figure GDA0002636051520000116
step 5: and (4) circularly executing Step 2-4 in the Step 402 to the last logical station for the logical station i +1, so as to obtain the production period of each logical station about the current product.
Step 6: summing production cycles of each logical station with respect to the product
Figure GDA0002636051520000117
And obtaining the production period of the whole logic production line relative to the product.
While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.

Claims (4)

1. A multi-product production line production cycle prediction method based on variability measurement is characterized by comprising the following steps:
step 1: extracting and processing data of each physical station on the production line, wherein the processing data comprises original performance data, historical shutdown data, historical production arrangement actual data and production arrangement data of a future period;
step 2: actually passing the product through link on each physical station to form a logical station of the product on the current physical station, and connecting the logical stations of the same product in series to obtain a logical production line of the product;
and step 3: based on the processing data of the physical work station corresponding to each logic work station, measuring the effective processing time and the variability of each logic work station on the logic production line of each product: firstly, measuring the effective processing time and the variability corresponding to each link of each logic station; then constructing a regression model to fuse the effective processing time and the variability of all links of each logic station to obtain the effective processing time and the variability of each logic station;
and 4, step 4: calculating the queuing time of each logic station about different products in a future period based on a VUT equation, and iteratively accumulating the queuing time and the processing time on a logic production line to obtain the overall production period of each product;
the effective processing time and the variability of each product corresponding to each link of each logic station are measured as follows:
using i to represent a logical station identifier, j to represent a link identifier, and t to represent a product category identifier;
shutdown data type calculation logic station i j th link to t th product p based on linktEffective time t ofeijtAnd effective machining time standard deviation sigmaeijt：
If the shutdown data of the current link only comprises preemptive shutdown data, then the formula is followed
Figure FDA0002636051510000011
Calculating the effective processing time teijtThen according to
Figure FDA0002636051510000012
Calculating the corresponding effective machining time standard deviation sigmaeijt；
If the shutdown data of the current link only includes non-preemptive shutdown data, then the formula is followed
Figure FDA0002636051510000013
Calculating the effective processing time teijtThen according to
Figure FDA0002636051510000014
Calculating the corresponding effective machining time standard deviation sigmaeijt；
If the shutdown data of the current link includes preemptive shutdown data and non-preemptive shutdownData according to the formula
Figure FDA0002636051510000015
Calculating the effective processing time teijtThen according to
Figure FDA0002636051510000021
Calculating the corresponding effective machining time standard deviation sigmaeijt；
Wherein, t0ijtJ link representing logical station i to t product ptTheoretical working time of (1), wherein ptBelong to the product set linkijCorresponding set of classes, linkijA set of products representing the jth link that passes through logical station i; c0ijtJ link representing logical station i to t product ptCoefficient of variation of the theoretical processing time of (1); sigma0ijt＝C0ijt×t0ijt(ii) a Parameter Aij＝mfij/(mfij+mrij) Wherein m isfijMean value of down time intervals, m, caused by failure of jth link representing logical station irijRepresenting the mean time interval of fault repair of the jth link of the logic work station i; sigmarijThe time interval standard deviation of fault repair of the j link of the logic work station i is represented; wherein t issijMean value of the moulding-change time intervals, σ, of the j-th link representing the logical station isijVariance, N, of the modulo time interval representing the j link of logical station isijRepresenting the average number of processed products between two die changes of the jth link of the logic work station i;
Figure FDA0002636051510000022
indicating a mobility of the preemptive shutdown;
fusing multiple products to obtain the effective processing time of each product on the j link of the logic station i
Figure FDA0002636051510000023
And mobility
Figure FDA0002636051510000024
Wherein
Figure FDA0002636051510000025
k is a product identifier, M represents the number of product categories passing through the current link, r0tDenotes the t product ptFeeding ratio of (1), nitDenotes the t product ptNumber of parallel links, m, of passing logical station iiktDenotes the t product ptAnd the kth product pkNumber of links passing simultaneously at logical station i, where pkE P, P represents the product set produced by mixing.
2. The method of claim 1, wherein the regression model is specifically: y isγ＝fγ(x)＝w'x+e；
Wherein x is (x)1,x2,x3,x4,x5,x6,x7) ' representing a feature vector, x1Representing the effective processing time of all links of the current logical station
Figure FDA0002636051510000026
Mean value of (1), x2Representing the effective processing time of all links of the current logical station
Figure FDA0002636051510000027
Variance of (a), x3Representing the volatility of all links of the current logical station
Figure FDA0002636051510000028
Root mean square, x4Representing the effective processing time of all links of the current logical station
Figure FDA0002636051510000029
Maximum value of (a), x5Representing the effective processing time of all links of the current logical station
Figure FDA00026360515100000210
Minimum value of (1), x6Representing the volatility of all links of the current logical station
Figure FDA0002636051510000031
Maximum value of (a), x7Representing the volatility of all links of the current logical station
Figure FDA0002636051510000032
Minimum value of (d); the symbol (·)' denotes transpose, w denotes coefficient vector, e denotes error vector, yγA subscript γ of 1,2, wherein y1Representing the effective processing time after the integration of the logical station as a whole, i.e. y1Effective processing time for arbitrary logical work station of product
Figure FDA0002636051510000033
y2Representing the variability of the logical station after the fusion of the whole, i.e. y2Being any logical work station of a product
Figure FDA0002636051510000034
3. The method according to claim 2, wherein step4 is specifically:
step 401: calculating expected utilization of future production cycles
Figure FDA0002636051510000035
Wherein u isiRepresents the expected utilization of the logical station i, rtIs the rate of feed, rbEffective processing rate, r, for a logical production line bottleneck stationeiEffective processing Rate for logical station i, miThe number of the parallel links of the logic station i is shown;
step 402: and calculating the queuing time of each product before each logic station based on a VUT equation, adding the effective processing time to obtain the production period of a single logic station, and circularly executing the step on each logic station and accumulating the production period to obtain the overall predicted production period of the logic production line.
4. The method of claim 3, wherein step 402 is specifically:
402-1: initializing an identifier i of the logic station to be 1, initializing arrival time interval variability, and recording an initial value of the arrival time interval variability as Ca(1)；
402-2: calculation of current product p using VUT equationkQueuing time for logical station i
Figure FDA0002636051510000036
Wherein
Figure FDA0002636051510000037
Wherein
Figure FDA0002636051510000038
Representing logical stations i with respect to products pkThe effective processing time of the process is reduced,
Figure FDA0002636051510000039
representing logical stations i with respect to products pkVariability of (u)iRepresents the expected utilization of the logical station i, miRepresenting the number of parallel links passing through the logic station i;
402-3: computing logical stations i about products pkProduction cycle CTi kWherein
Figure FDA00026360515100000310
402-4: updating the identifier i to i +1 and according to the formula
Figure FDA00026360515100000311
Calculating departure time interval variability C of current logic work stationa(i+1)；
402-5: circularly executing the steps 402-2-402-4 until the last logic work station, thereby obtaining the production period of each logic work station;
402-6: adding the production cycle CT of each logical stationi kObtaining a complete logical production line concerning product pkThe production cycle of (1).
CN201810538128.1A 2018-05-30 2018-05-30 Multi-product production line production period prediction method based on variability measurement Active CN108646684B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810538128.1A CN108646684B (en) 2018-05-30 2018-05-30 Multi-product production line production period prediction method based on variability measurement

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810538128.1A CN108646684B (en) 2018-05-30 2018-05-30 Multi-product production line production period prediction method based on variability measurement

Publications (2)

Publication Number Publication Date
CN108646684A CN108646684A (en) 2018-10-12
CN108646684B true CN108646684B (en) 2020-10-23

Family

ID=63758663

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810538128.1A Active CN108646684B (en) 2018-05-30 2018-05-30 Multi-product production line production period prediction method based on variability measurement

Country Status (1)

Country Link
CN (1) CN108646684B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109740813B (en) * 2018-12-29 2020-11-24 上海华力微电子有限公司 Analysis and prediction method for on-line product batch running state in wafer manufacturing
CN113780907B (en) * 2020-06-10 2023-09-12 宝山钢铁股份有限公司 Consumable demand and inventory control method, device, system and storage medium
CN113807513A (en) * 2020-06-12 2021-12-17 中芯国际集成电路制造(上海)有限公司 Method, system and computing device for determining impact factor weight of semiconductor production process
CN111857081B (en) * 2020-08-10 2023-05-05 电子科技大学 Chip packaging test production linear energy control method based on Q-learning reinforcement learning

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070239300A1 (en) * 2006-03-30 2007-10-11 Burda Richard G Method and system for setting rates and targets in a range management system
CN101727098B (en) * 2008-10-21 2012-01-25 中芯国际集成电路制造(上海)有限公司 Cycle time target measuring method and system
CN105045236B (en) * 2015-07-21 2018-02-13 江苏云道信息技术有限公司 A kind of production line balance dispatching method and system
CN107944695B (en) * 2017-11-21 2022-03-15 电子科技大学 Key chain buffer area setting method based on production uncertainty

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于变动性的半导体制造生产线性能预测与优化策略研究;彭歆然;《中国优秀硕士学位论文全文数据库 信息科技辑》;20140115(第1期);全文 *

Also Published As

Publication number Publication date
CN108646684A (en) 2018-10-12

Similar Documents

Publication Publication Date Title
CN108646684B (en) Multi-product production line production period prediction method based on variability measurement
CN109978403B (en) Quality control method, device and equipment for product assembly process
JP2021051769A (en) Data cleansing system and method for inferring feed composition
CN105046402B (en) A kind of state evaluating method applied to secondary equipment of intelligent converting station
TWI443776B (en) An automated state estimation system for cluster tools and a method of operating the same
Abd Rahman et al. Enhancement of overall equipment effectiveness (OEE) data by using simulation as decision making tools for line balancing
CN106447201B (en) The best maintenance decision method of power transmission and transforming equipment based on Markovian decision process
Gilanifar et al. Multi-task logistic low-ranked dirty model for fault detection in power distribution system
CN107360026B (en) Distributed message middleware performance prediction and modeling method
CN108664700A (en) Acceleration degradation information Fusion Modeling Method based on uncertain data Envelope Analysis
CN107679089A (en) A kind of cleaning method for electric power sensing data, device and system
CN113111955A (en) Line loss abnormal data expert system and detection method
CN104915730A (en) Device multi-attribute maintenance decision method based on weight
Vladov et al. Neural network method for helicopters turboshaft engines working process parameters identification at flight modes
CN105389656A (en) Performance evaluation method of secondary equipment of power system
US20080133031A1 (en) System and method for validating data for a machine
CN115169426B (en) Anomaly detection method and system based on similarity learning fusion model
Piplani et al. Simplification strategies for simulation models of semiconductor facilities
CN113408915B (en) Production line bottleneck identification method and system
CN105699043B (en) A kind of wind tunnel sensors that improve measure stability and the method for precision
Yan et al. Characteristic curves and cycle time control of re-entrant lines
JP2019148876A (en) Control device, data processing method and program
Yang et al. Modeling assembly systems with repetitive operations
Cupek et al. Estimation of the number of energy consumption profiles in the case of discreet multi-variant production
Laktionov Improving the methods for determining the index of quality of subsystem element interaction

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
GR01 Patent grant
GR01 Patent grant