A semiconductor packaging and testing process scheduling optimization scheme
By abstracting factory business rules into constraints and performing scoring optimization, the semiconductor packaging and testing process scheduling is optimized, solving the problem that existing systems cannot take into account both local and global aspects, and realizing efficient and dynamic production plan generation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI GLORYSOFT CO LTD
- Filing Date
- 2022-08-02
- Publication Date
- 2026-05-26
AI Technical Summary
Existing semiconductor packaging and testing process scheduling systems are unable to scientifically and effectively schedule production, cannot take into account both local and global information, resulting in uncontrollable production risks and an inability to predict factory capacity. Furthermore, production targets are mutually constrained and their importance changes over time.
By abstracting factory business rules into constraints, using constraint scoring for optimization, configuring different constraints and scores to generate the optimal solution, combining the selection of sorting machines and testing machines to reduce the machine cut-off rate, and using the first-in-first-out sorting rule to optimize scheduling.
It enables the rapid generation of scheduling plans that meet production targets, reduces manual intervention, improves production efficiency and capacity forecasting capabilities, and meets dynamic production needs at different stages.
Smart Images
Figure CN115239009B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor back-end packaging and testing production, and in particular to a semiconductor packaging and testing process scheduling optimization scheme. Background Technology
[0002] Common practice 1: Export data from ERP / MES, recalculate using Excel based on the existing model, and then have the scheduler adjust and schedule the data based on their experience, or adjust it through meetings / regular reviews.
[0003] Common approach two: Employing a Real-Time Dispatch (RTD) system for production scheduling. RTD uses built-in algorithms to determine reasonable dispatch results for batches to be added to the warehouse. Production Assistants (MAs) simply need to dispatch tasks according to the batch order recommended by the RTD. The rules and algorithms applicable to RTD systems can be broadly categorized into two types: simple dispatch rules based on local information and advanced dispatch rules based on global information. Advanced dispatch rules based on global information often achieve superior control effects compared to simple rules. However, how to balance local and global information, how to determine system status and trends, and how to consider both local and overall interests are the key research focuses and challenges for this type of rule.
[0004] Due to the variability of the production environment in the subsequent testing process, schedulers need to make timely adjustments, resulting in a large workload. Since the process mainly relies on the experience of schedulers, it is difficult to schedule scientifically and effectively and quantify the scheduling results, making it difficult to control production risks. The production status is unclear, and the statistics on factory capacity are relatively lagging, making it impossible to make effective predictions.
[0005] The RTD (Real-Time Delivery) solution automates the factory by configuring dispatch rules for machines to find suitable batches and sending transport instructions to the downstream transport system. However, because the RTD system selects the optimal batch for processing based on the factory's real-time status, it produces an instantaneous optimal solution for the current machine status. This cannot fully consider the factory's overall optimal capacity over a period of time (e.g., three days), nor can it provide reasonable global control and scheduling of limited resources.
[0006] The various objectives that factories focus on, such as order delivery time, shortest product production cycle, and maximum machine utilization, are interconnected and mutually restrictive, and the importance of each objective changes over time. For example, at the beginning of each month, the main purpose of scheduling is to improve machine utilization and achieve a shorter production cycle, while in the later part of the month, the focus may shift to order delivery time. Summary of the Invention
[0007] To address the problems identified in the background art, this application proposes a semiconductor packaging and testing process scheduling optimization scheme, which aims to reduce planning and scheduling time and meet dynamic production targets.
[0008] This application is achieved through the following technical solution:
[0009] A semiconductor packaging and testing process scheduling optimization scheme includes the following steps:
[0010] (1) Select appropriate sorting machine and testing machine: The sorting machine is generally fixed and can process different types of packages depending on the configuration; the testing machine can be moved and the corresponding sorting machine can be selected for connection. The resources of the testing machine are configured inside the testing machine. During testing, the corresponding test board needs to be provided externally according to the requirements of the work order.
[0011] (2) Before scheduling calculation, the data to be included in the scheduling needs to be assembled first. The data needs to be pre-filtered. Based on the work order information, sorting machines and testing machines that cannot be matched are filtered out to reduce the complexity of scheduling calculation.
[0012] (3) Abstract the factory's business rules into constraints, and use constraints to limit the scheduling engine to allocate batches to machines, that is, use constraints to score and find the best;
[0013] (4) Each constraint has a score. The rationality of different solutions is compared by the score. The solution with the higher score represents the better result.
[0014] (5) Set up the solver to find the solution with the highest score among all possible solutions, while the optimal solution is the solution with the highest score encountered in the solution process;
[0015] (6) At different time periods, the factory focuses on different objectives, such as order delivery time, shortest product production cycle, and maximum machine utilization. The choice of different objectives is reflected in the scheduling algorithm by configuring different constraints and scores. Interrelated and restrictive objectives mean the selection and trade-off of different constraints. Different constraint configurations can make the scheduling results achieve different objectives, either single or balanced, which can be configured by the user on the UI interface.
[0016] (7) Constraint type rules are filters, which perform a screening by matching machine and work order information; Score type rules are scoring constraints, which are divided into hard constraints and soft constraints; Sort type rules are sorting conditions, which is a way for the system to optimize the scheduling result without violating constraints.
[0017] (8) If the first-in-first-out sorting rule is selected, the scheduling system will prioritize the production of batches with longer waiting times for materials.
[0018] In a preferred embodiment, the following factors need to be considered when selecting the sorting machine and the testing machine in step (1): sorting machine model, sorting machine can process package type, testing machine model, testing machine resources, and testing board model.
[0019] As a preferred embodiment, the work order information in step (2) includes the following: the required sorting machine model, packaging type, required test machine model, required test machine resources, required test board model, work order priority, work order entry time, work order product model, customer code, batch quantity, and UPH.
[0020] In a preferred embodiment, UPH is an abbreviation for Unit Per Hour, which refers to the output per hour.
[0021] In a preferred embodiment, the sorting machine is referred to as a setup when switching between different test machines or test boards or changing products. The setup operation takes a certain amount of time, so a low setup rate needs to be ensured to achieve continuous and efficient production in the factory.
[0022] As a preferred embodiment, different constraints may have different weights, and the consequences of breaking one constraint are different from those of breaking another constraint.
[0023] In a preferred embodiment, the time required to switch the test machine type is longer than the time required to switch the test board type. Therefore, the solution obtained by changing the test machine is worse than the solution obtained by changing the test board. In terms of scheduling algorithms, this means that different constraints have different scores.
[0024] As a preferred embodiment, the constraints include two types: hard constraints and soft constraints. Hard constraints refer to those constraints that cannot be violated, that is, situations that are absolutely not allowed in business. A solution that violates a hard constraint is an unusable solution. Soft constraints, in contrast to hard constraints, can be violated. The purpose of setting soft constraints is not to prevent them from being violated, but to quantitatively limit the development direction of the scheduling results.
[0025] As a preferred embodiment, a batch cannot be allocated to different machines at the same time to minimize machine breakage during the production process. Since machine breakage is inevitable during production, fewer machine breakages mean a better solution.
[0026] In a preferred embodiment, the UI interface in step (6) is a constraint configuration interface.
[0027] The design principle of this application is as follows: Business rules are described using constraints, scores are set to represent the weights of these rules, and optimization calculations are performed using constraint scoring. This optimization scheme transforms the scheduling problem of back-end packaging and testing processes into an algorithmic problem of finding a relatively optimal solution. A mature algorithm engine is then used to solve the problem, quickly and effectively yielding the relatively optimal solution. By configuring different constraint rules for different production targets, and selecting the appropriate configuration for scheduling calculations according to requirements, a production plan that meets the requirements can be generated.
[0028] Beneficial effects: This solution can replace manual scheduling, reducing planning and scheduling time; it describes business rules with configurable constraints, and generates different scheduling results through different configuration operations to meet the dynamic production goals of different users at different stages. Attached Figure Description
[0029] Figure 1 This is a schematic diagram of the machine tool structure in one embodiment of the present invention.
[0030] Figure 2 This is a schematic diagram of the scheduling scheme architecture in one embodiment of the present invention.
[0031] Figure 3 This is a schematic diagram of the constraint configuration interface in one embodiment of the present invention. Detailed Implementation
[0032] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings: These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes, but the protection scope of the present invention is not limited to the following embodiments.
[0033] like Figure 1-3 As shown, a semiconductor packaging and testing process scheduling optimization scheme includes the following steps:
[0034] (1) Selecting suitable sorting machines and testing machines: Sorting machines are generally fixed and can process different types of packages depending on the configuration; testing machines can be moved and linked to corresponding sorting machines. The resources of the testing machine are configured inside the testing machine, and the corresponding test boards need to be provided externally according to the requirements of the work order during testing; Step (1) When selecting sorting machines and testing machines, the following factors need to be considered: sorting machine model, package types that the sorting machine can process, testing machine model, testing machine resources, and test board model. When switching between different testing machines or test boards or changing products, it is called machine setup. Machine setup takes a certain amount of time, so a low machine setup rate needs to be ensured to achieve continuous and efficient production in the factory.
[0035] (2) Before scheduling calculation, the data to be included in the scheduling needs to be assembled first. The data needs to be pre-filtered. Based on the work order information, the sorting machines and test machines that cannot be matched are filtered out to reduce the complexity of scheduling calculation. The work order information in step (2) includes the following: the required sorting machine model, packaging type, required test machine model, required test machine resources, required test board model, work order priority, work order entry time, work order product model, customer code, batch quantity, and UPH. UPH is an abbreviation for Unit Per Hour, which refers to the output per hour.
[0036] (3) Abstract the factory's business rules into constraints, and use constraints to limit the scheduling engine to allocate batches to machines, that is, use constraints to score and find the best;
[0037] (4) Each constraint has a score. The rationality of different solutions is compared by the score. The solution with the higher score represents the better result.
[0038] (5) Set up the solver to find the solution with the highest score among all possible solutions, while the optimal solution is the solution with the highest score encountered in the solution process;
[0039] (6) The factory focuses on different objectives at different time periods, such as order delivery time, shortest product production cycle, and maximum machine utilization. The choice of different objectives is reflected in the scheduling algorithm by configuring different constraints and scores. Interrelated and restrictive objectives mean the selection and trade-off of different constraints. Different constraint configurations can make the scheduling results achieve different objectives, either single or balanced. The user configures these on the UI interface, which is the constraint configuration interface.
[0040] (7) Constraint type rules are filters, which perform a screening by matching machine and work order information; Score type rules are scoring constraints, which are divided into hard constraints and soft constraints; Sort type rules are sorting conditions, which is a way for the system to optimize the scheduling result without violating constraints.
[0041] (8) If the first-in-first-out sorting rule is selected, the scheduling system will prioritize the production of batches with longer waiting times for materials.
[0042] Different constraints may have different weights, and the consequences of breaking one constraint are not the same as those of breaking another. The time required to switch to a different test machine type is longer than the time required to switch to a different test board type. Therefore, the solution obtained by changing the test machine is worse than the solution obtained by changing the test board. In terms of scheduling algorithms, this means that different constraints have different scores.
[0043] Constraints include two types: hard constraints and soft constraints. Hard constraints are those that cannot be violated, meaning situations that are absolutely not allowed in business operations. A solution that violates a hard constraint is an unusable solution. Soft constraints, in contrast to hard constraints, can be violated. The purpose of setting soft constraints is not to prevent them from being violated, but to quantitatively limit the development direction of the scheduling results. A batch cannot be assigned to different machines at the same time to minimize machine cuts during the production process. Since machine cuts are inevitable during production, fewer machine cuts indicate a better solution.
[0044] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.
Claims
1. A semiconductor packaging and testing process scheduling optimization scheme, characterized in that, The following steps are included: (1) Select appropriate sorting machine and testing machine: The sorting machine is fixed and can process different types of packages depending on the configuration; the testing machine can be moved and the corresponding sorting machine can be selected for connection. The resources of the testing machine are configured inside the testing machine. During testing, the corresponding test board needs to be provided externally according to the requirements of the work order. (2) Before scheduling calculation, the data to be included in the scheduling needs to be assembled first. The data needs to be pre-filtered. Based on the work order information, the sorting machines and testing machines that cannot be matched are filtered out to reduce the complexity of scheduling calculation. (3) Abstract the business rules of the factory into constraints, and use the constraints to limit the scheduling engine to allocate batches to the machine, that is, use the constraints to score and find the best; (4) Each constraint has a score. The rationality of different solutions is compared by the score. The solution with the higher score represents the better result. (5) The purpose of setting up the solver is to find the solution with the highest score among all possible solutions, while the optimal solution is the solution with the highest score encountered in the solution process; (6) The factory focuses on different objectives in different time periods, such as order delivery period / shortest product production cycle / maximum machine utilization. The selection of different objectives is reflected in the scheduling algorithm by configuring different constraints and scores. Interrelated and restrictive objectives mean different constraints to be selected and weighed. Different constraint configurations can make the scheduling results achieve different objectives, either single or balanced, which can be configured by the user on the UI interface. (7) Constraint type rules are filters, which perform a screening by matching machine and work order information; Score type rules are scoring constraints, which are divided into hard constraints and soft constraints; Sort type rules are sorting conditions, which is a way for the system to optimize the scheduling result without violating constraints. (8) If the first-in-first-out sorting rule is selected, the scheduling system will prioritize the production of batches with longer waiting times for materials.
2. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, When selecting the sorting machine and testing machine in step (1), the following factors need to be considered: Sorting machine model, sorting machine can process packaging types, testing machine model, testing machine resources, and testing board model.
3. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, The work order information in step (2) includes the following: the required sorting machine model, packaging type, required test machine model, required test machine resources, required test board model, work order priority, work order entry time, work order product model, customer code, batch quantity, and hourly output.
4. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, The sorting machine, when switching to different test machines or test boards or changing products, is called a machine switching setup. The machine switching operation takes a certain amount of time, so it is necessary to ensure a low machine switching rate in order to achieve continuous and efficient production in the factory.
5. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, Different constraints may have different weights, and the consequences of breaking one constraint are not the same as those of breaking another.
6. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, The time required to switch from test machine type to test board type is longer than the time required to switch from test board type to test machine. Therefore, the solution obtained by changing the test machine is worse than the solution obtained by changing the test board. In terms of scheduling algorithm, this means that different constraints have different scores.
7. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, The constraints include two types: hard constraints and soft constraints. Hard constraints are those that cannot be violated, that is, situations that are absolutely not allowed in business. A solution that violates a hard constraint is an unusable solution. Soft constraints, in contrast to hard constraints, can be violated. The purpose of setting soft constraints is not to prevent them from being violated, but to quantitatively limit the development direction of the scheduling results.
8. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, A batch should not be assigned to different machines at the same time. We should try to minimize the occurrence of machine breakage during the production process. Since machine breakage is inevitable during production, the fewer machine breakages, the better the solution.
9. The semiconductor packaging and testing process scheduling optimization scheme according to claim 1, characterized in that, The UI interface in step (6) is the constraint configuration interface.