CN117391423B - Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line - Google Patents
Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line Download PDFInfo
- Publication number
- CN117391423B CN117391423B CN202311684093.XA CN202311684093A CN117391423B CN 117391423 B CN117391423 B CN 117391423B CN 202311684093 A CN202311684093 A CN 202311684093A CN 117391423 B CN117391423 B CN 117391423B
- Authority
- CN
- China
- Prior art keywords
- production
- modeling
- time
- constraint
- task
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 159
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 152
- 239000000758 substrate Substances 0.000 title claims abstract description 38
- 239000000919 ceramic Substances 0.000 title claims abstract description 32
- 230000008569 process Effects 0.000 claims abstract description 69
- 238000010276 construction Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims description 45
- 238000005457 optimization Methods 0.000 claims description 14
- 238000003754 machining Methods 0.000 claims description 8
- 238000011144 upstream manufacturing Methods 0.000 claims description 5
- 230000008901 benefit Effects 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000013461 design Methods 0.000 abstract description 5
- 238000013468 resource allocation Methods 0.000 abstract description 2
- 239000000047 product Substances 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 239000000306 component Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000010845 search algorithm Methods 0.000 description 2
- 238000002922 simulated annealing Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 238000004080 punching 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0633—Workflow analysis
-
- 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
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Manufacturing & Machinery (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- General Factory Administration (AREA)
Abstract
The invention designs a multi-constraint automatic scheduling method for a production line of a chip high-multilayer ceramic package substrate, belonging to the technical field of constraint planning; firstly, collecting production data of a chip high multilayer ceramic package substrate production line and constructing structural data based on the production data; then modeling the model by adopting variable modeling, constraint modeling and target modeling methods; then, after the search process is completed by searching based on depth priority and heuristic method based on the modeling result, the minimum construction period of the scheduling plan after the modeling of the actual production plan and the scheduling scheme for realizing the minimum construction period can be obtained, and meanwhile, the maximum utilization efficiency of task arrangement is met; the invention improves the scheduling solving efficiency, can find the optimal solution of specific tasks and resource allocation when a plurality of production schedules are arranged in parallel, and improves the production quality, the production efficiency and the delivery capacity of the substrate production line.
Description
Technical Field
The invention belongs to the technical field of constraint planning, and particularly relates to a multi-constraint automatic scheduling method for a production line of a chip high-multilayer ceramic package substrate.
Background
The high multilayer ceramic package substrate is one of core components of a chip, and its production quality and production efficiency are of strategic importance for the development of the chip industry. However, in the actual production process, since the ceramic package substrate includes several complex part layers processed and stacked, the same layer is manufactured by a plurality of different processes, such as punching, cleaning, quality inspection, etc., and strict processing and assembling sequences must be ensured between the processes to ensure the quality of the final product; and the same working procedure also relates to the processing task distribution of different part layers, and part of parts are even formed by combining a plurality of other parts, so that the complexity of the production flow is increased. In addition, production resources such as manpower and equipment are limited, how to reasonably allocate and utilize the resources to meet the production requirements is a very challenging problem, and the production environment and requirements of the package substrate often change, such as increase or decrease of production schedule, equipment failure, etc., and these changes require enough flexibility and adaptability of production schedule to make adjustments rapidly. Therefore, the improvement of the production efficiency and the delivery capacity of the high multilayer ceramic package substrate production line is of great significance to the development of the chip manufacturing industry.
The traditional scheduling method relies on manual operation and experience judgment, which is time-consuming and error-prone, and cannot ensure the accuracy and reliability of scheduling. With the development of modern big data, a plurality of new scheduling methods and technologies are initiated to be applied to production practice, and the methods generally adopted are as follows:
mathematical programming methods, which can generally provide accurate optimal solutions with smaller scale and clear mathematical models, but in large-scale scheduling problems such as high multilayer ceramic package substrate production lines in the chip field, due to the lack of effective mathematical decision variables, can result in very high complexity of solution and inability to find optimal solutions, and have too high requirements on the form and constraint conditions of the problem to be suitable for complex and changeable production environments.
Meta-heuristic methods, such as simulated annealing algorithms, genetic algorithms, etc., are flexible, can find high quality solutions in a short time, can be used for real-time or near-time decision making, but most of the methods only support optimization of a single target, and generally provide near-optimal solutions, which are not well suited for certain solution problems with high precision requirements.
Disclosure of Invention
Aiming at the defects of the prior art, the invention designs a multi-constraint automatic scheduling method for the production line of the chip high-multilayer ceramic package substrate, which is applied to the production scheduling of the production line of the ceramic package substrate in the chip production.
The multi-constraint automatic scheduling method for the production line of the chip high multilayer ceramic package substrate specifically comprises the following steps:
step 1: collecting production data of a chip high multilayer ceramic package substrate production line and constructing structural data based on the production data;
the production line production data of the chip high multilayer ceramic package substrate are collected: collecting production plan, procedure, layer work piece and workshop resource entity information of a production line from an upstream manufacturing execution system;
constructing structured data of different entities through information extraction and relationship abstraction, simultaneously encoding the structured data by using a JSON data format, and giving different hierarchical relationships and nested information;
the structured data obtained after encoding comprises: production plan information at the same moment, wherein each production plan comprises a plurality of working procedures, each working procedure has corresponding workshop resources, and the workshops have corresponding equipment and personnel to finish the working procedure; in each process step, the hierarchical parts to be processed are explicitly mentioned;
step 2: modeling the structured data obtained in the step 1;
after the structured data is constructed, modeling the data in the ceramic substrate production line by adopting variable modeling, constraint modeling and target modeling methods, so that the data is abstracted;
step 2.1: modeling variables; by usingiRepresenting the number of the production plan,jrepresenting a certain process in a production plan,mthe following variables are defined to represent the process task numbers in the process:
start time:the method comprises the steps of carrying out a first treatment on the surface of the End time:；
duration of time:；
and (3) processing a workshop:；
daily workshop operationInter:；
wherein the start timeAnd end timeAre all at any point in time within the executable time range, durationIs a triplet including a start time, an end time of an executable time range, and a duration constant of workpiece processing pre-calculated according to workshop productivityd，Representing the process plant to which the workpiece is assigned,representing the daily working time of the corresponding workshop;
step 2.2: constraint modeling; in order to perform data characterization on hard limiting conditions in the actual production process, respectively designing No-overlapping constraint, processing-order constraint and Maximum-time constraint;
wherein the No-overlapping constraint represents that the same shop does not allow other products to be scheduled for processing when processing a certain product, i.e. the same shop can only allocate one task at the same time; if the jth process task of the ith production planmAnd at the kth production planlThe same workshop is allocated to process each working procedure task n, so that the time periods of the working procedure tasks n cannot be overlapped; expressed in terms of the following constraints:
processing-order constraint represents constraint of sequence of different processes of the same layer, and the processes of a certain layer of workpiece to be processed must be according to fingersPerforming the steps in a fixed sequence; for the mth layer of the work piece, if process j is in processlBefore, then the ending time of j cannot be later thanlExpressed by the following constraint:
for Maximum-time, the design is made for some machining processes of the combined kit, which must not be started until the machining of other layers is completed, assuming a processjThe process has to wait for the other process set L and the layer number setMThe process can be started after completion, the constraint is defined as:
step 2.3: modeling a target; defining a shortest target of total execution period:
defining the highest target of the unit-day full load rate of the workshop:
step 3: searching the modeling result in the step 2 based on depth priority and a heuristic method;
searching the modeled variables based on depth priority and a heuristic method, wherein the searching is to narrow the range of the upper and lower bounds of the variables and the optimization targets until the optimization targets are converged in a feasible domain on the basis of meeting the constraint modeling feasibility of the step 2;
step 3.1: the pre-solving process comprises the following steps: after variable modeling, constraint modeling and target modeling are obtained, firstly determining a numerical range after modeling according to the capacity and the production quantity of an actual production line, wherein the numerical range is divided by assuming that all production schedules are sequentially arranged without considering any constraint condition;
step 3.2: boundary search process: the boundary search needs to search nodes with nearest feasible distances to the optimization target, which are calculated according to a heuristic search method, by continuously calculating the starting time and the ending time of the new nodes and updating the optimal value of the target; comprises the following steps:
step S1: representing a node as a production planiIs one of the steps ofjIs a processing task of (a)mI.e. tripletsThe set of all possible triples constitutes a state space, which is expressed as:i.e. all processing tasks of all procedures in the production process; the task queues to be solved next are expressed as: priority solving queueStoring in the queue all task nodes for which start and end times are to be determined next;
step S2, initializing the priority solving queue in step S1QState space is setSIs placed in a queue;
step S3: to determine the order of the search process in the state set, a heuristic function is definedTo evaluate the potential advantage of each node, the interval between the end time of a completed process and the start time of the next process is taken as part of a heuristic function:
wherein the method comprises the steps ofAndrepresenting the latest end time of the last working procedure of the same layer of processing workpieces and the latest end time of the last working piece in the same working procedure respectively;
in addition, in order to better select the next candidate node, the work efficiency of the workshop related to the production line needs to be considered, and the ratio of the daily work time of the workshop to the task duration of the node is taken as another part of the heuristic function:
the value of the heuristic function represents the reciprocal of the utilization rate of the working time of the current working procedure, and the lower the proportion is, the more the workshop time is fully utilized by the current processing task;
thus, the final heuristic function is:
wherein α represents a balance parameter for balancing the influence factors of the two sub-items;
calculating the priority according to the value of the heuristic function, wherein the lower the value of the heuristic function is, the highest priority is proved, and the node with the highest priority is taken out from the queue;
step S4: expanding the node according to the Processing-order constraints mentioned in step 2.2, i.e. from the state spaceSelecting new possible nodes, calculating the values of heuristic functions of the new possible nodes, and updating the values of other heuristic functions which do not perform the task of arranging in a state space;
step S5: will be from the state spaceAdding newly generated possible nodes, namely selected possible nodes, into a priority solving queue Q, and arranging the nodes according to the value of a heuristic function from small to large;
step S6: computing the slaveThe start and end times of the node fetched in step 2.3 are updated with the target、And narrow the boundary;
repeating the process from step S3 to step S6 until the queue is empty or the target、Is equal to the upper bound and the lower bound of (2); after the search is completed, the minimum construction period of the scheduling plan after the modeling of the actual production plan and the scheduling scheme for realizing the minimum construction period can be obtained, and the maximum utilization efficiency of the task arrangement is simultaneously satisfied.
The invention has the beneficial technical effects that:
the invention provides a multi-constraint automatic scheduling method for a production line of a high-multilayer ceramic package substrate. According to the method, variable modeling is carried out on process tasks in the actual production process of the packaging substrate production line, constraint modeling is carried out on complex process relationships and other realistic constraint conditions, a plurality of optimization targets are established at the same time, constraint programming and optimization means are used on the basis, heuristic functions are designed to search solution space, backtracking, pruning and restarting means are designed to greatly reduce the search range of the solution space, therefore, the scheduling solving efficiency is improved, the optimal solution of specific tasks and resource allocation when a plurality of production plans are arranged in parallel can be found, and the production quality, the production efficiency and the delivery capacity of the substrate production line are improved.
Drawings
FIG. 1 is a diagram showing a structure of a multi-constraint automatic scheduling method for a production line of a chip high multilayer ceramic package substrate according to an embodiment of the present invention;
FIG. 2 is a flowchart of a heuristic search algorithm in accordance with an embodiment of the present invention.
Detailed Description
The invention will be further described with reference to the accompanying drawings and examples;
in the invention, three high-multilayer ceramic package substrate production plans with simultaneous production scheduling requirements in a certain day of a certain factory are taken as an example, each production plan contains information such as required processing start time, required processing quantity, processing procedures, part layers required to be processed in each procedure, workshop resources required by processing, daily productivity and the like, the latest required delivery date, priority and the like of the production plan, the part layers required to be processed need to be processed in sequence of a plurality of procedures, and parts of different levels can be involved in one procedure and can only be distributed in a specific workshop. Because three production plans have production scheduling requirements at the same time, the processing factory needs to comprehensively consider various factors influencing the production plan scheduling, and makes an arrangement plan for the production time and the production shift of each part layer to be processed, and delivers each production plan in time on the premise of meeting the basic premise of meeting various constraint jump-in non-conflict. Therefore, the high multilayer ceramic package substrate production line has high requirements on scientific, efficient and intelligent scheduling methods.
A multi-constraint automatic scheduling method for a chip high multilayer ceramic package substrate production line comprises the processes of data collection, structured data generation, constraint planning modeling (comprising variable modeling, constraint modeling and target modeling), heuristic search, solving and mapping a solution to an actual part processing production line. The whole structure of the technical scheme is shown in figure 1:
step 1: collecting production data of a chip high multilayer ceramic package substrate production line and constructing structural data based on the production data;
the production line production data of the chip high multilayer ceramic package substrate are collected: collecting production plan, procedure, layer work piece and workshop resource entity information of a production line from an upstream manufacturing execution system;
constructing structured data of different entities through information extraction and relationship abstraction, simultaneously encoding the structured data by using a JSON data format, and giving different hierarchical relationships and nested information;
the structured data obtained after encoding comprises: production plan information at the same moment, wherein each production plan comprises a plurality of working procedures, each working procedure has corresponding workshop resources, and the workshops have corresponding equipment and personnel to finish the working procedure; in each process step, the hierarchical parts to be processed are explicitly mentioned;
the production data of the high multilayer ceramic package substrate production line is usually from an upstream ERP system, and for the actual production line, the main characteristics are as follows:
(1) The number of scheduled production plans is large at the same time. In a practical production line, a factory side receives a plurality of different production plan arrangements, the information structures of the production plans are similar, but the required processing quantity and start-stop time of each production plan are inconsistent, and the production plans can be distinguished from each other in terms of urgency. In order to ensure that each production schedule can be delivered on time, it is necessary to quantify the priority of the production schedule information.
(2) The ceramic package substrate has a large number of layers and the process for each layer is complicated. The processing steps of ceramic package substrates are very tedious, a single substrate may be composed of many layers of parts, each layer of parts must be processed strictly according to the sequence of the process within a specified time, otherwise the quality may be disqualified, and the downstream production task is affected. In addition, some of the components require other components to be assembled together. Therefore, the whole substrate processing flow has complexity, and a proper data structure needs to be designed to ensure that each part meets the sequence or combination requirements.
(3) Shop resources of the factory are limited and difficult to schedule. In a practical production line, factories often distribute part processing tasks of the same procedure to fixed workshops, i.e. one workshop can only be responsible for one type of processing task, and no other processing task is allowed to be discharged to the workshop, meanwhile, the working conditions of the workshops of the factories are different, and some workshops are three shift values all day and some workshops are one shift value all day, which increases the difficulty of a scheduling algorithm. In actual production, there are situations in which some workshops are left idle despite limited resources, because a scheduling means based on human experience may result in the upstream processing task of the task assigned to the workshop not being completed. Therefore, in order to fully utilize the plant resources and fully load the plant utilization as much as possible, it is necessary to design a related scheduling algorithm to ensure the full utilization of the resources.
In order to solve the above problems, it is necessary to model the information of the production plan, the process, the layer workpieces, and the shop resources of the production line to satisfy the demand of intelligent scheduling. The method uses a JSON data format to encode and generate input information, and different hierarchical relations and nested information are given.
The input information includes 3 production schedule information at the same time, each production schedule includes 10 processes, and each process corresponds to one shop resource, so the number of the workshops is also 10. In each process, the hierarchical parts to be processed are explicitly mentioned, and the number of processing tasks in each process is 10, and the total number of tasks is 300. The duration of each task was 4 hours.
Step 2: modeling the structured data obtained in the step 1;
after the structured data is constructed, modeling the data in the ceramic substrate production line by adopting variable modeling, constraint modeling and target modeling methods, so that the data is abstracted;
step 2.1: modeling variables; by usingiRepresenting the number of the production plan,jrepresenting a certain process in a production plan,mthe following variables are defined to represent the process task numbers in the process:
start time:the method comprises the steps of carrying out a first treatment on the surface of the End time:；
duration of time:；
and (3) processing a workshop:；
daily shop hours:；
wherein the start timeAnd end timeAre all at any point in time within the executable time range, durationIs a triplet, comprising respectively a start time, an end time of the executable time range, a duration constant d of the workpiece processing pre-calculated according to the shop capacity,representing the process plant to which the workpiece is assigned,representing the daily working time of the corresponding workshop;
the ranges of the variables in this embodiment are:，，；
step 2.2: constraint modeling; in order to perform data characterization on hard limiting conditions in the actual production process, respectively designing No-overlapping constraint, processing-order constraint and Maximum-time constraint;
the No-overlapping constraint represents that 10 workshops cannot allow other products to be arranged for processing when processing a certain product, namely, the same workshop can only allocate one task at the same time, and once the task starts, the task cannot stop; if the jth process task of the ith production planmAnd at the kth production planlThe same workshop is allocated to process each working procedure task n, so that the time periods of the working procedure tasks n cannot be overlapped; expressed in terms of the following constraints:
processing-order constraint represents constraint of sequence of different working procedures of the same layer, and working procedures of a certain layer of workpieces to be processed must be executed according to a specified sequence; as in the case of the 3 rd and 4 th tasks of the 2 nd process in the production plan numbered 1 in this example, the start time of the 4 th task must be later than the end time of the 3 rd task. For the mth layer of the work piece, if process j is in processlBefore, then the ending time of j cannot be later thanlExpressed by the following constraint:
for Maximum-time, the design is directed to machining processes for some composite sets that must not begin until the machining of other layers is complete, e.g., lamination operations must not be performed until at least four layers of work pieces are complete. The 1 st task of the 10 th process in the production plan, numbered 2 in this example, has a combination kit requirement, must be waited for until the 7 th-9 process tasks in the plan are all completed, and then the start time of the task must be later than the latest end time of the previous 7-9 processes. Assuming that there is a process j that must wait for the other process set L and layer number set M to finish before starting, the constraint is defined as:
step 2.3: modeling a target; the most important of the whole ceramic package substrate is the delivery date, so that each production schedule can be delivered as expected, and the total production time can be shortened as much as possible, therefore, the optimization target is to minimize the total execution period, and the shop full load rate per day is the highest;
in order to minimize the total execution period, the maximum end time and the maximum start time of each production plan are made to be different, and if the difference is minimum, the total execution period is shortest; thus defined as the following targets:
in order to maximize the plant unit day full load rate, it is necessary to maximize the working time and the utilization efficiency of the plant under the condition of meeting No-overlapping, the utilization efficiency being the ratio of the task time allocated to the plant to the workable time of the plant on the same day, and therefore defined as the following targets:
in this embodiment, the difference between the latest end time and the earliest start time of the tasks in the three production plans is found among a plurality of different scheduling schemes, and the optimization objective positioning is minimized to the difference, that is, the total execution period of the production plans is minimized. Maximizing the efficiency of the utilization plant on this basis, assuming that the working time of each plant is 8 hours and the total time per day allocated to the current plant is 6 hours, the current plant's today utilization is 66.7%, so another goal is to determine the maximum daily average utilization of the plant.
Step 3: searching the modeling result in the step 2 based on depth priority and a heuristic method;
searching the modeled variables based on depth priority and a heuristic method, wherein the searching is to narrow the range of the upper and lower bounds of the variables and the optimization targets until the optimization targets are converged in a feasible domain on the basis of meeting the constraint modeling feasibility of the step 2;
step 3.1: the pre-solving process comprises the following steps: after variable modeling, constraint modeling and target modeling are obtained, firstly determining a numerical range after modeling according to the capacity and the production quantity of an actual production line, wherein the numerical range is divided by assuming that all production schedules are sequentially arranged without considering any constraint condition; so that the feasibility check is performed under such conditions, the existence of the solution of the scheduling object can be determined in advance, thereby saving unnecessary computing resources; secondly, carrying out equivalent replacement on various complex constraint conditions to obtain simple sub constraint conditions so as to improve solving efficiency;
the total processing time of the production line in this example ranges from 0,3×10×10×4 hours. Each task has a duration of 4 hours and whenever the task is assigned to begin execution, the next task must be continued after the 4 hours of execution.
Step 3.2: boundary search process: the boundary search needs to search nodes with nearest feasible distances to the optimization target, which are calculated according to a heuristic search method, by continuously calculating the starting time and the ending time of the new nodes and updating the optimal value of the target. As shown in fig. 2, the method comprises the following steps:
step S1: representing a node as a processing task m, i.e. triplet, of a process j of a production plan iThe set of all possible triples constitutes a state space, which is expressed as:i.e. all processing tasks of all procedures in the production process; the task queues to be solved next are expressed as: priority solving queueStoring in the queue all task nodes for which start and end times are to be determined next;
step S2, initializing a priority solving queue Q in the step S1, and putting a first node of a state space S into the queue;
step S3: to determine the order of the search process in the state set, a heuristic function is definedTo evaluate the potential advantage of each node, the interval between the end time of a completed process and the start time of the next process is taken as part of a heuristic function:
wherein the method comprises the steps ofAndrepresenting the latest end time of the last working procedure of the same layer of processing workpieces and the latest end time of the last working piece in the same working procedure respectively; the heuristic function greatly shortens the time gap between the front and rear processing workpieces, so that the time arrangement of the whole arrangement is tighter;
in addition, in order to better select the next candidate node, the work efficiency of the workshop related to the production line needs to be considered, and the ratio of the daily work time of the workshop to the task duration of the node is taken as another part of the heuristic function:
the value of the heuristic function represents the reciprocal of the utilization rate of the working time of the current working procedure, and the lower the proportion is, the more the workshop time is fully utilized by the current processing task;
thus, the final heuristic function is:
where α represents a balance parameter to balance the influence factors of the two sub-items.
Calculating the priority according to the value of the heuristic function, wherein the lower the value of the heuristic function is, the highest priority is proved, and the node with the highest priority is taken out from the queue;
step S4: expanding the node according to the Processing-order constraints mentioned in step 2.2, i.e. from the state spaceSelecting new possible nodes, calculating the values of heuristic functions of the new possible nodes, and updating the values of other heuristic functions which do not perform the task of arranging in a state space;
step S5: will be from the state spaceAdding newly generated possible nodes, namely selected possible nodes, into a priority solving queue Q, and arranging the nodes according to the value of a heuristic function from small to large;
step S6: computing the slaveThe start and end times of the node fetched in step 2.3 are updated with the target、And narrow the boundary;
repeating the process from step S3 to step S6 until the queue is empty or the target、Is equal to the upper bound and the lower bound of (2); finish searchingAfter the completion of the project, the minimum construction period of the project plan after the modeling of the actual production plan and the project plan for realizing the minimum construction period can be obtained, and the maximum utilization efficiency of the task arrangement is simultaneously satisfied. And searching an optimal solution in a state space through a boundary search algorithm, minimizing the total execution period and maximizing the full load rate of the unit daily working time of the workshop.
The boundary searching process in this embodiment: selecting tasks from a state space, namely a set formed by the tasks, adding the tasks into a solving queue, calculating a heuristic function value of each task through the heuristic function constructed in the step 3.2, sequentially solving the starting time and the ending time of each task according to the value, updating the value of the heuristic function of the task which is not scheduled after the starting time and the ending time of the task are determined, and reordering the tasks, thereby updating the upper and lower bound ranges of the execution time of the total construction period. The above-described process is repeated until the execution time of the total construction period becomes a certain value.
After the above-mentioned process is completed, the minimum value of the schedule periods of the three production plans can be obtained, and the task arrangement meeting the maximum utilization efficiency can be obtained. In the subsequent work, the output and task arrangement of the flow are put back into the actual processing production line, so that an advantageous plan is provided for the operation and processing arrangement of the actual production line corresponding to the three production plans, and the working efficiency and delivery capacity of the whole high-multilayer ceramic package substrate production line are improved.
Through multiple experiments, the effect of the method provided by the invention on actual production scheduling is compared with various solving methods, and the results are shown in the following table:
| the application | Linear programming | Simulated annealing | First-come-first-serve | Short job priority | Sequential scheduling | |
| General construction period (Tian) | 31 | 41 | 36 | 45 | 42 | 50 |
Claims (1)
1. The multi-constraint automatic scheduling method for the production line of the chip high multilayer ceramic package substrate is characterized by comprising the following steps of:
step 1: collecting production data of a chip high multilayer ceramic package substrate production line and constructing structural data based on the production data;
step 2: modeling the structured data obtained in the step 1;
after the structured data is constructed, modeling the data in the ceramic substrate production line by adopting variable modeling, constraint modeling and target modeling methods, so that the data is abstracted;
step 3: searching the modeling result in the step 2 based on depth priority and a heuristic method;
searching the modeled variables based on depth priority and a heuristic method, wherein the searching is to narrow the range of the upper and lower bounds of the variables and the optimization targets until the optimization targets are converged in a feasible domain on the basis of meeting the constraint modeling feasibility of the step 2;
after the search is completed, the minimum construction period of the scheduling plan after the modeling of the actual production plan can be obtained, and the scheduling scheme for realizing the minimum construction period can be realized;
step 1, collecting production data of a chip high multilayer ceramic package substrate production line: collecting production plan, procedure, layer work piece and workshop resource entity information of a production line from an upstream manufacturing execution system;
constructing structured data of different entities through information extraction and relationship abstraction, simultaneously encoding the structured data by using a JSON data format, and giving different hierarchical relationships and nested information;
the structured data obtained after encoding comprises: production plan information at the same moment, wherein each production plan comprises a plurality of working procedures, each working procedure has corresponding workshop resources, and the workshops have corresponding equipment and personnel to finish the working procedure; in each process step, the hierarchical parts to be processed are explicitly mentioned;
step 2, modeling the variables; let i denote the production plan number, j denote a certain process in the production plan, m denote the process task number in the process, and define the following variables:
start time: s is S i,j,m The method comprises the steps of carrying out a first treatment on the surface of the End time: ei ,j,m ；
Duration of time: d (D) i,j,m ＝(S i,j,m ,E i,j,m ,d)；
And (3) processing a workshop: r is R i,j,m ；
Daily shop hours:
wherein the start time S i,j,m And end time E i,j,m Are all at any point in time within the executable time range, duration D i,j,m Is a triplet comprising the start time and end time of the executable time range and the duration constant d, R of the workpiece processing calculated in advance according to the workshop capacity i,j,m Representing the process plant to which the workpiece is assigned,representing the daily working time of the corresponding workshop;
step 2, modeling the constraint; in order to perform data characterization on hard limiting conditions in the actual production process, respectively designing No-overlapping constraint, processing-order constraint and Maximum-time constraint;
the No-overlapping constraint represents that the same workshop does not allow other products to be arranged for processing when processing a certain product, namely, the same workshop can only allocate one task at the same time; if the j-th process task m of the ith production plan and the l-th process task n of the kth production plan are allocated to the same shop for processing, their time periods cannot be overlapped; expressed in terms of the following constraints:
the Processing-order constraint represents the constraint of the sequence of different working procedures of the same layer, and the working procedures of a certain layer of workpieces to be processed must be executed according to a specified sequence; for the mth layer of the work piece, if process j precedes process l, then the ending time of j cannot be later than the starting time of l, expressed by the following constraint:
E i，j,m ≤S i,l,m
the Maximum-time is designed for some machining procedures of the combined set, the machining procedures must not be started until the machining of other layers is completed, and if there is a procedure j, the procedure must be started after the machining of other procedure sets L and layer number sets M is completed, the constraint is defined as:
step 2, modeling the target;
defining a shortest target of total execution period:
defining the highest target of the unit-day full load rate of the workshop:
the step 3 is specifically as follows:
step 3.1: the pre-solving process comprises the following steps: after variable modeling, constraint modeling and target modeling are obtained, firstly determining a numerical range after modeling according to the capacity and the production quantity of an actual production line, wherein the numerical range is divided by assuming that all production schedules are sequentially arranged without considering any constraint condition;
step 3.2: boundary search process: the boundary search needs to search nodes with nearest feasible distances to the optimization target, which are calculated according to a heuristic search method, by continuously calculating the starting time and the ending time of the new nodes and updating the optimal value of the target;
step 3.2 specifically comprises the following steps:
step S1: a node is represented as a processing task m of a process j of a production plan i, i.e. triples (i, j, m), the set of all possible triples constituting a state space represented as: s= { (i, j, m) }, i.e. all processing tasks of all procedures in the production process; the task queues to be solved next are expressed as: preferential solving queue q= { (i, j, m) }, which holds all task nodes that are to be next determined for start and end times;
step S2, initializing a priority solving queue Q in the step S1, and putting a first node of a state space S into the queue;
step S3: to determine the order of the search process in the state set, a heuristic function H (i, j, m) is defined to evaluate the potential advantage of each node, taking the interval between the end time of the completed process and the start time of the next process as part of the heuristic function:
f(i,j,m)＝max(0,S i，j,m -E i，j-1,m S i，j,m -E i,j，m-1 )
wherein E is i,j-1,m And E is i,j,m-1 Representing the latest end time of the last working procedure of the same layer of processing workpieces and the latest end time of the last working piece in the same working procedure respectively;
in addition, in order to better select the next candidate node, the work efficiency of the workshop related to the production line needs to be considered, and the ratio of the daily work time of the workshop to the task duration of the node is taken as another part of the heuristic function:
the value of the heuristic function represents the reciprocal of the utilization rate of the working time of the current working procedure, and the lower the proportion is, the more the workshop time is fully utilized by the current processing task;
thus, the final heuristic function is:
H(i,j,m)＝f(i,j,m)+α·g(i,j,m)
wherein α represents a balance parameter for balancing the influence factors of the two sub-items;
calculating the priority according to the value of the heuristic function, wherein the lower the value of the heuristic function is, the highest priority is proved, and the node with the highest priority is taken out from the queue;
step S4: expanding the node according to Processing-order constraint, namely selecting a new possible node from a state space S= { (i, j, m) }, calculating the value of a heuristic function of the new possible node, and updating the values of other heuristic functions which do not perform the task of arranging in the state space;
step S5: adding newly generated possible nodes, namely selected possible nodes, in a state space S= { (i, j, m) } into a priority solving queue Q, and arranging the nodes according to the value of a heuristic function from small to large;
step S6: computing slave q= { (iStart and end times of nodes fetched in j, m) } updating the target Z 1 、Z 2 And narrow the boundary;
repeating the process of step S3-step S6 until the queue is empty or the target Z 1 、Z 2 Is equal to the upper and lower bounds of (c).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311684093.XA CN117391423B (en) | 2023-12-11 | 2023-12-11 | Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311684093.XA CN117391423B (en) | 2023-12-11 | 2023-12-11 | Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117391423A CN117391423A (en) | 2024-01-12 |
| CN117391423B true CN117391423B (en) | 2024-03-22 |
Family
ID=89463458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311684093.XA Active CN117391423B (en) | 2023-12-11 | 2023-12-11 | Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117391423B (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000010448A (en) * | 1998-07-31 | 2000-02-15 | 이구택 | Automatic schedule modeling method in cold mill pcm line |
| JP2009075631A (en) * | 2007-09-18 | 2009-04-09 | Kobe Steel Ltd | Production schedule preparation method and system therefor |
| CN105204472A (en) * | 2015-09-30 | 2015-12-30 | 河南科技大学 | Single-piece discrete type production operation scheduling optimization method |
| CN107944695A (en) * | 2017-11-21 | 2018-04-20 | 电子科技大学 | One kind is based on the probabilistic chaining key setting buffers method of production |
| CN111507523A (en) * | 2020-04-16 | 2020-08-07 | 浙江财经大学 | Cable production scheduling optimization method based on reinforcement learning |
| CN112001619A (en) * | 2020-08-20 | 2020-11-27 | 北京京东振世信息技术有限公司 | Production scheduling method and device, computer storage medium and electronic equipment |
| CN112132546A (en) * | 2020-09-25 | 2020-12-25 | 杉数科技(北京)有限公司 | Method and device for scheduling production |
| CN112465451A (en) * | 2020-11-19 | 2021-03-09 | 北京航空航天大学 | Optimization method for project parallel and multi-resource supporting scheme |
| CN112947319A (en) * | 2019-12-11 | 2021-06-11 | 上海宝信软件股份有限公司 | Batch scheduling optimization method and system for multiple production lines in cold rolling area of iron and steel enterprise |
| CN113705866A (en) * | 2021-08-16 | 2021-11-26 | 成都飞机工业（集团）有限责任公司 | Scheduling optimization method and system based on resource-constrained project scheduling problem model |
| CN114154821A (en) * | 2021-11-22 | 2022-03-08 | 厦门深度赋智科技有限公司 | Intelligent scheduling dynamic scheduling method based on deep reinforcement learning |
| CN114298423A (en) * | 2021-12-30 | 2022-04-08 | 北京深度奇点科技有限公司 | APS dynamic production plan scheduling algorithm |
| CN114462770A (en) * | 2021-12-27 | 2022-05-10 | 埃克斯工业(广东)有限公司 | Semiconductor production scheduling method, system and computer readable storage medium |
| CN114995329A (en) * | 2022-05-09 | 2022-09-02 | 扬州大学 | Chip encapsulation measurement linear planning and scheduling method based on multi-machine dynamic scheduling |
| CN116128273A (en) * | 2023-04-17 | 2023-05-16 | 清云小筑(北京)创新技术有限公司 | Automatic scheduling method and device for working procedure considering variable construction period and dynamic resource constraint |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7292904B2 (en) * | 2003-10-31 | 2007-11-06 | International Business Machines Corporation | Method for sizing production lot starts within a linear system programming environment |
| US20230057537A1 (en) * | 2021-08-06 | 2023-02-23 | At&T Intellectual Property I, L.P. | Facilitating striated random restart during allocation of resources using a heuristic approach |
-
2023
- 2023-12-11 CN CN202311684093.XA patent/CN117391423B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000010448A (en) * | 1998-07-31 | 2000-02-15 | 이구택 | Automatic schedule modeling method in cold mill pcm line |
| JP2009075631A (en) * | 2007-09-18 | 2009-04-09 | Kobe Steel Ltd | Production schedule preparation method and system therefor |
| CN105204472A (en) * | 2015-09-30 | 2015-12-30 | 河南科技大学 | Single-piece discrete type production operation scheduling optimization method |
| CN107944695A (en) * | 2017-11-21 | 2018-04-20 | 电子科技大学 | One kind is based on the probabilistic chaining key setting buffers method of production |
| CN112947319A (en) * | 2019-12-11 | 2021-06-11 | 上海宝信软件股份有限公司 | Batch scheduling optimization method and system for multiple production lines in cold rolling area of iron and steel enterprise |
| CN111507523A (en) * | 2020-04-16 | 2020-08-07 | 浙江财经大学 | Cable production scheduling optimization method based on reinforcement learning |
| CN112001619A (en) * | 2020-08-20 | 2020-11-27 | 北京京东振世信息技术有限公司 | Production scheduling method and device, computer storage medium and electronic equipment |
| CN112132546A (en) * | 2020-09-25 | 2020-12-25 | 杉数科技(北京)有限公司 | Method and device for scheduling production |
| CN112465451A (en) * | 2020-11-19 | 2021-03-09 | 北京航空航天大学 | Optimization method for project parallel and multi-resource supporting scheme |
| CN113705866A (en) * | 2021-08-16 | 2021-11-26 | 成都飞机工业（集团）有限责任公司 | Scheduling optimization method and system based on resource-constrained project scheduling problem model |
| CN114154821A (en) * | 2021-11-22 | 2022-03-08 | 厦门深度赋智科技有限公司 | Intelligent scheduling dynamic scheduling method based on deep reinforcement learning |
| CN114462770A (en) * | 2021-12-27 | 2022-05-10 | 埃克斯工业(广东)有限公司 | Semiconductor production scheduling method, system and computer readable storage medium |
| CN114298423A (en) * | 2021-12-30 | 2022-04-08 | 北京深度奇点科技有限公司 | APS dynamic production plan scheduling algorithm |
| CN114995329A (en) * | 2022-05-09 | 2022-09-02 | 扬州大学 | Chip encapsulation measurement linear planning and scheduling method based on multi-machine dynamic scheduling |
| CN116128273A (en) * | 2023-04-17 | 2023-05-16 | 清云小筑(北京)创新技术有限公司 | Automatic scheduling method and device for working procedure considering variable construction period and dynamic resource constraint |
Non-Patent Citations (2)
| Title |
|---|
| 具有交货时间窗约束的无等待流水车间调度模型与算法;王晶;王伟玲;;中国机械工程(第19期);全文 * |
| 考虑班组负荷均衡的智能制造车间工序级作业任务排程模型研究;周琪森;林杰;白翱;;制造业自动化(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117391423A (en) | 2024-01-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | An improved particle swarm optimization algorithm for dynamic job shop scheduling problems with random job arrivals | |
| Guo et al. | A digital twin-based flexible cellular manufacturing for optimization of air conditioner line | |
| Guo et al. | Optimisation of integrated process planning and scheduling using a particle swarm optimisation approach | |
| Tang et al. | Flexible job-shop scheduling with tolerated time interval and limited starting time interval based on hybrid discrete PSO-SA: An application from a casting workshop | |
| Shen et al. | Mathematical modeling and multi-objective evolutionary algorithms applied to dynamic flexible job shop scheduling problems | |
| Gomes et al. | Reactive scheduling in a make-to-order flexible job shop with re-entrant process and assembly: a mathematical programming approach | |
| CN113705866B (en) | Scheduling optimization method and system based on resource-constrained project scheduling problem model | |
| JPH09153090A (en) | Method and device for forming working process production plan | |
| CN105974891B (en) | A kind of mold production process self-adaptation control method based on dynamic billboard | |
| CN111259314B (en) | Production scheduling method for distributed concrete prefabricated part flow shop | |
| CN114565247B (en) | Workshop scheduling method, device and system based on deep reinforcement learning | |
| CN112465451A (en) | Optimization method for project parallel and multi-resource supporting scheme | |
| Shah et al. | Colored Petri net model for significant reduction of invariants in flexible manufacturing systems | |
| Hosseinian et al. | An energy-efficient mathematical model for the resource-constrained project scheduling problem: an evolutionary algorithm | |
| CN117391423B (en) | Multi-constraint automatic scheduling method for chip high multilayer ceramic package substrate production line | |
| CN116009419A (en) | Virtual reconstruction and simulation operation method and system for complex equipment manufacturing process | |
| CN112861371B (en) | Steel industry cloud production scheduling method based on edge computing | |
| Peters et al. | Resource optimization in business processes | |
| Vinod et al. | Development and analysis of scheduling decision rules for a dynamic flexible job shop production system: a simulation study | |
| Shukla et al. | SimEvents-based discrete-event simulation modelling and performance analysis for dynamic job-shop manufacturing system | |
| Salehi et al. | A novel model for production optimization with stochastic rework and failure-prone job shop schedule problem via hybrid simulation–heuristic optimization | |
| Zhongyi et al. | Automated planning and scheduling system for the composite component manufacturing workshop | |
| CN116755397B (en) | Multi-machine collaborative task scheduling method based on graph convolution strategy gradient | |
| Huang | Target cascading optimization for flexible distributed shop scheduling problems with multi-objectives | |
| Zikos et al. | Efficient location-aware scheduling of maintenance tasks in shop floors |
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 |