Semiconductor scheduling algorithm model configuration system, method, equipment and media

Abstract

This invention relates to the field of semiconductor manufacturing technology, specifically a semiconductor scheduling algorithm model configuration system, method, device, and medium. The system includes a data configuration module, an algorithm configuration module, a parameter parsing module, a task scheduling module, a data acquisition module, a model execution module, and a result processing module. The data configuration module receives user-input data model configuration parameters; the algorithm configuration module receives algorithm model configuration parameters. The parameter parsing module parses the parameters and generates data preprocessing rules and algorithm model instances; the task scheduling module creates scheduling tasks and triggers the process; the data acquisition module obtains raw data from the source data table and performs preprocessing; the model execution module executes the scheduling algorithm based on the algorithm instances; and the result processing module outputs standardized data. This invention achieves decoupling between data and algorithms, improving the system's adaptability and scalability.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor manufacturing technology, and particularly relates to a semiconductor scheduling algorithm model configuration system, method, device and medium. Background Technology

[0002] In the semiconductor manufacturing industry, scheduling algorithms are key tools for optimizing production processes, improving equipment utilization, and shortening delivery cycles. Current semiconductor scheduling technologies typically rely on commercial software or customized solutions. While these systems support dynamic scheduling and equipment load balancing, they suffer from significant shortcomings in algorithm model configuration. Common configuration strategies are limited to allowing users to select certain business objectives (such as completing processing as quickly as possible or prioritizing high-priority tasks) or imposing simple restrictions on material resources (such as specifying machine processing limits). This results in a high degree of coupling between data parameters and the algorithm model, making independent configuration impossible. This coupling leads to repetitive definition of data parameters, poor scalability, and the need to rebuild the model when adding new business scenarios, increasing development costs and operational complexity. Furthermore, existing systems lack global constraint configuration, focusing primarily on resource constraints while neglecting overall optimization goals. Their complex user interfaces also place high technical demands on business personnel, significantly increasing the learning curve. These shortcomings collectively result in insufficient flexibility and low efficiency for semiconductor scheduling systems when adapting to changing business scenarios. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing a semiconductor scheduling algorithm model configuration system, method, device and medium, which solves the high coupling between data parameters and algorithm model, realizes independent parameter configuration and modular reuse, thereby improving the system's adaptability, reducing the difficulty of operation and enhancing its scalability.

[0004] The first aspect of this invention discloses a semiconductor scheduling algorithm model configuration system, comprising:

[0005] The data configuration module is used to receive data model configuration parameters input by the user. The data model configuration parameters include a set of data items, a data item type, and a data item scope. The set of data items is used to specify the data type to be processed, including wafer batch data, machine data, or processing relationship data. The data item type includes a filtering type, a priority type, or a supplementary type. The data item scope is used to calibrate the machine processing mode.

[0006] The algorithm configuration module is used to receive algorithm model configuration parameters input by the user. The algorithm model configuration parameters include target class parameters and constraint class parameters, wherein the target class parameters are used to define the optimization objective and its priority weight, and the constraint class parameters are used to define business constraints and their variable parameters.

[0007] The parameter parsing module, connected to the data configuration module and the algorithm configuration module, is used to parse the data model configuration parameters and the algorithm model configuration parameters to generate data preprocessing rules and algorithm model instances.

[0008] The task scheduling module, connected to the parameter parsing module, is used to create scheduling tasks based on the parsed parameters and trigger the data acquisition operation of the data acquisition module and the scheduling algorithm execution operation of the model execution module, wherein the model execution module executes based on the algorithm model instance generated by the parameter parsing module;

[0009] The data acquisition module, connected to the task scheduling module, is used to acquire raw data from the source data table and filter and transform the raw data according to the data preprocessing rules to generate preprocessed data.

[0010] The model execution module, connected to the data acquisition module and the parameter parsing module, is used to execute the scheduling algorithm based on the algorithm model instance and preprocessed data to generate scheduling results.

[0011] The result processing module is connected to the model execution module and is used to post-process the scheduling results and convert them into standardized output data.

[0012] The data configuration module, algorithm configuration module, parameter parsing module, task scheduling module, data acquisition module, model execution module, and result processing module are linked through parameter passing and event triggering, so that the data model configuration parameters are independent of the algorithm model configuration parameters, but can be dynamically related through shared variables.

[0013] In the above system, the data configuration module is also used to receive machine group configuration parameters, which include machine group name, machine type and operating mode. The machine group is used to define a group of machines with the same processing technology. The parameter parsing module dynamically adjusts the data filtering range in the data preprocessing rules according to the machine group configuration parameters.

[0014] In the above system, the algorithm configuration module includes multiple optimization objectives as the target class parameters. Each optimization objective is configured with a priority order and a weight value. The model execution module uses a multi-objective optimization algorithm to solve the problem according to the priority order and weight value. The objective with higher priority is satisfied first, and the weight value is used to adjust the optimization intensity when the priorities are the same.

[0015] In the above system, the constraint parameters in the algorithm configuration module include common constraints and specific constraints. The common constraints are reusable constraints for multiple algorithm models, and the specific constraints are constraints specific to a single algorithm model. Each constraint is configured with variable parameters, including time limit parameters or resource quantity parameters. The parameter parsing module dynamically injects constraint rules into the algorithm model instance according to the variable parameters.

[0016] In the aforementioned system, the common constraints among the constraint parameters include the latest start time constraint, whose variable parameter is the station time constraint duration. The station time constraint duration represents the time constraint that must be followed between stations. The model execution module ensures that the processing start time is no later than the sum of the specified time and the station time constraint duration during the solution process. The specific constraints include the wafer loading and unloading port number constraint in the lithography area, whose variable parameter is the number of wafer loading and unloading ports. The model execution module limits the number of wafer batches processed simultaneously based on the number of wafer loading and unloading ports.

[0017] In the above system, the data configuration module allows users to configure variable names for data items, the algorithm configuration module requires variable names to be assigned values ​​when selecting data items, and the parameter parsing module associates data model configuration parameters with algorithm model configuration parameters through variable names, thereby decoupling data parameters and algorithm models.

[0018] In the above system, the task scheduling module is also configured with a running frequency parameter for periodically triggering scheduling tasks. The task scheduling module dynamically manages scheduling tasks based on the creation and deletion of algorithm model instances. When a new algorithm model instance is added, a corresponding scheduling task is automatically created, and when an algorithm model instance is deleted, the corresponding scheduling task is automatically deleted. The data acquisition module updates the preprocessed data according to the latest configuration parameters each time a task is executed.

[0019] In the above system, the scheduling algorithm used by the model execution module is based on operations research constraint programming. The algorithm model instance is composed of multiple algorithm base classes, each of which represents an abstract business rule, including a target base class or a constraint base class.

[0020] In the above system, the target base class includes priority violation target, minimum overall scheduling time target, or resource utilization target; the constraint base class includes earliest start time constraint or machine processing quantity constraint; and the parameter parsing module dynamically instantiates the target base class or constraint base class according to the user's configuration selection and parameters.

[0021] In the aforementioned system, the result processing module also converts the scheduling results into a data format that the platform can interface with, for visualization or further analysis, thus completing the entire closed loop from configuration to execution.

[0022] The second aspect of this invention discloses a semiconductor scheduling algorithm model configuration method, comprising the following steps:

[0023] S1: Receive data model configuration parameters input by the user. The data model configuration parameters include a set of data items, a type of data item, and a scope of data item. The set of data items is used to specify the data type to be processed, including wafer batch data, machine data, or processing relationship data. The type of data item includes a filtering type, a priority type, or a supplementary type. The scope of data item is used to calibrate the machine processing mode.

[0024] S2: Receive algorithm model configuration parameters input by the user. The algorithm model configuration parameters include target class parameters and constraint class parameters. The target class parameters are used to define the optimization target and its priority weight, and the constraint class parameters are used to define business constraints and their variable parameters.

[0025] S3: Parse the data model configuration parameters and algorithm model configuration parameters to generate data preprocessing rules and algorithm model instances;

[0026] S4: Create a scheduling task based on the parsed parameters, and trigger the data acquisition operation of the data acquisition module and the scheduling algorithm execution operation of the model execution module, wherein the model execution module executes based on the algorithm model instance generated by the parameter parsing module;

[0027] S5: Obtain raw data from the source data table, and filter and transform the raw data according to the data preprocessing rules to generate preprocessed data;

[0028] S6: Execute the scheduling algorithm based on the algorithm model instance and preprocessed data to generate scheduling results;

[0029] S7: Post-process the scheduling results and convert them into standardized output data.

[0030] A third aspect of the present invention discloses an electronic device, comprising: a memory and a processor, wherein the processor and the memory are connected;

[0031] The memory is used to store programs;

[0032] The processor invokes a program stored in the memory to execute the method provided in the second aspect embodiment described above.

[0033] The fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a computer to perform the method provided in the second aspect embodiment.

[0034] Compared with existing technologies, this invention has the following advantages: The data configuration module of this invention allows users to independently configure data model parameters, including data item sets, types, and scope, thereby completely decoupling data parameters from the algorithm model. For example, users can flexibly set data items (such as filtering wafer batch data or machine data) through a graphical interface, avoiding the problems of redundant definition and model binding of data parameters in existing technologies, and improving the system's flexibility to adapt to changes in business scenarios. This decoupling design allows data parameters to be reused in multiple algorithm models, reducing configuration costs and supporting rapid expansion into new business scenarios without secondary development.

[0035] Meanwhile, the algorithm configuration module, by receiving user-input target and constraint parameters, achieves fully productized configurability of the algorithm model. Users can dynamically define optimization objectives (such as minimizing scheduling time) and their priority weights, as well as business constraints (such as machine processing limitations), overcoming the limitations of existing technologies that only support partial target selection or resource constraints. The parameter parsing module further transforms configuration parameters into data preprocessing rules and algorithm model instances. The automated processing reduces manual intervention, lowers the operational threshold, and allows business personnel to easily use the system without requiring extensive technical background.

[0036] Furthermore, the collaborative work of the task scheduling module, data acquisition module, model execution module, and result processing module forms a closed-loop automated scheduling process. The system can automatically create scheduling tasks, acquire and preprocess data, execute algorithms, and post-process results based on configuration parameters. This not only improves scheduling efficiency but also ensures the standardization of output data. Overall, this invention significantly improves the adaptability, scalability, and ease of use of semiconductor scheduling systems through modular decoupling and automated integration, overcoming the shortcomings of high coupling, operational difficulties, and insufficient scalability in the prior art.

[0037] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0038] Figure 1 This is a system module diagram of Embodiment 1 of the present invention.

[0039] Figure 2 This is a flowchart of the method in Embodiment 2 of the present invention. Detailed Implementation

[0040] Example 1

[0041] like Figure 1 As shown, a semiconductor scheduling algorithm model configuration system includes:

[0042] The data configuration module is used to receive data model configuration parameters input by the user. The data model configuration parameters include a set of data items, a data item type, and a data item scope. The set of data items is used to specify the data type to be processed, including wafer batch data, machine data, or processing relationship data. The data item type includes a filtering type, a priority type, or a supplementary type. The data item scope is used to calibrate the machine processing mode.

[0043] The algorithm configuration module is used to receive algorithm model configuration parameters input by the user. The algorithm model configuration parameters include target class parameters and constraint class parameters, wherein the target class parameters are used to define the optimization objective and its priority weight, and the constraint class parameters are used to define business constraints and their variable parameters.

[0044] The parameter parsing module, connected to the data configuration module and the algorithm configuration module, is used to parse the data model configuration parameters and the algorithm model configuration parameters to generate data preprocessing rules and algorithm model instances.

[0045] The task scheduling module, connected to the parameter parsing module, is used to create scheduling tasks based on the parsed parameters and trigger the data acquisition operation of the data acquisition module and the scheduling algorithm execution operation of the model execution module, wherein the model execution module executes based on the algorithm model instance generated by the parameter parsing module;

[0046] The data acquisition module, connected to the task scheduling module, is used to acquire raw data from the source data table and filter and transform the raw data according to the data preprocessing rules to generate preprocessed data.

[0047] The model execution module, connected to the data acquisition module and the parameter parsing module, is used to execute the scheduling algorithm based on the algorithm model instance and preprocessed data to generate scheduling results.

[0048] The result processing module is connected to the model execution module and is used to post-process the scheduling results and convert them into standardized output data.

[0049] The data configuration module, algorithm configuration module, parameter parsing module, task scheduling module, data acquisition module, model execution module, and result processing module are linked through parameter passing and event triggering, so that the data model configuration parameters are independent of the algorithm model configuration parameters, but can be dynamically related through shared variables.

[0050] During implementation, users input data model configuration parameters through a graphical user interface, specifically including the data item set, data item type, and data item scope. For example, in a lithography scheduling scenario, a user adds a new data item, specifying the data item set as type "Lot" to indicate processing wafer batch data; the data item type is selected as a filter type to filter high-priority wafer batches; and the data item scope is set to single-wafer processing mode to match the characteristics of the lithography machine. Users name the data item, such as "High-Priority Lot Filtering," and configure a variable name, such as "priority_level," through a variable name field for subsequent dynamic assignment. During implementation, the system backend stores the parameters as structured data. After the user saves the configuration, the data configuration module generates reusable data rules. Through this parameterized design, data model configuration parameters are stored independently, avoiding direct binding to the algorithm model and enabling flexible adjustment of data input. When the semiconductor plant's business scenario changes, such as adding a lithography process station, users only need to modify the data item parameters without rebuilding the entire model, improving the system's adaptability and configuration efficiency.

[0051] Next, the algorithm configuration module receives the algorithm model configuration parameters input by the user, including target parameters and constraint parameters. Target parameters define the optimization objectives and their priority weights. For example, in lithography scheduling, the user selects the priority violation objective as the first optimization objective, setting its priority order to 1 and its weight to 0.7, and selects the minimization of overall scheduling time objective as the second optimization objective, setting its priority order to 2 and its weight to 0.3. Constraint parameters define business constraints and their variable parameters. The user selects common constraints such as the latest start time constraint, with its variable parameter, the site time constraint, set to 1 hour, meaning the processing start time cannot be later than the planned time plus 1 hour. Simultaneously, the user selects specific constraints such as the number of wafer loading / unloading ports in the lithography area, with its variable parameter, the number of wafer loading / unloading ports, set to 4, meaning the number of wafer batches processed simultaneously by the machine is limited by the number of ports. In implementation, the algorithm configuration module provides drop-down lists and numerical input boxes for user interaction; the parameters are parsed and stored as algorithm instance templates. By breaking down the algorithm model into configurable units of objectives and constraints, business personnel can customize optimization strategies without programming, reducing operational complexity and supporting the accurate execution of multi-objective optimization algorithms, thereby improving the rationality of scheduling results.

[0052] The parameter parsing module connects with the data configuration module and the algorithm configuration module. It parses user-input parameters and generates data preprocessing rules and algorithm model instances. In the lithography area example, the parameter parsing module reads the variable name "priority_level" from the data item "high-priority Lot filtering" and the priority setting in the algorithm objective to generate data preprocessing rules: during the data acquisition phase, Lot data with higher variable values ​​are prioritized for filtering; simultaneously, constraint rules are dynamically injected into the algorithm model instance based on constraint parameters such as site time constraint duration, forming an executable constraint programming model. In implementation, the parameter parsing module uses a parsing engine to map configuration parameters to internal rule objects, ensuring that data parameters and algorithm parameters are associated through shared variables. This parsing mechanism enables dynamic linkage between data model configuration parameters and algorithm model configuration parameters, avoiding redundant parameter definitions, ensuring configuration consistency, and improving the system's ability to handle complex scenarios.

[0053] The task scheduling module creates scheduled tasks based on the output of the parameter parsing module and configures the running frequency parameters, such as triggering once every 5 minutes. When a user goes online with the lithography area scheduling model, the task scheduling module periodically starts the task, triggering the data acquisition module to execute. The data acquisition module obtains raw data from source data tables (such as the Lot information table, machine information table, and product line information table) and filters and transforms it according to data preprocessing rules. For example, in the lithography area scenario, the rules require filtering the Lot data that is about to arrive at the lithography site and applying priority filtering to generate preprocessed data such as a high-priority Lot list and an available machine list. In implementation, the data acquisition module uses SQL queries or API calls to obtain data and performs data cleaning and transformation. Through task scheduling and real-time data updates, the scheduling process is automated, reducing manual intervention and improving system response speed and production efficiency.

[0054] The model execution module executes the scheduling algorithm based on the algorithm model instance and preprocessed data, employing an operations research-based constraint programming method for solution. In the lithography area example, the algorithm model instance combines priority violation objectives, minimizing scheduling time objectives, and lithography area-specific constraints. The model execution module uses multi-objective optimization algorithms (such as linear programming or heuristic algorithms) to calculate the optimal scheduling scheme, ensuring that high-priority lots are processed first and that the machine port quantity limit is adhered to. After solving, the scheduling results are generated, including processing order, start time, and resource allocation table. In implementation, the model execution module calls the algorithm library to perform calculations and handle constraint conflicts. Through the combination of modular algorithm base classes, flexible application of business rules is achieved, improving scheduling accuracy and resource utilization, while supporting rapid iterative optimization.

[0055] The result processing module performs post - processing on the scheduling results generated by the model execution module, converting them into a data format that can be docked with the platform, such as JSON or XML format. In the lithography area scenario, the result processing module maps the scheduling results into a standardized data structure for visualization or further analysis, such as generating Gantt charts or production reports. When implemented, the result processing module uses a data converter to normalize the original results and outputs them to the user interface or external systems. Through the standardization of the result format, a closed - loop management from configuration to execution is achieved, supporting data - driven decision optimization and enhancing the system's integrability.

[0056] The entire system realizes linkage through parameter passing and event triggering between modules. For example, the variable name in the data configuration module and the parameter value in the algorithm configuration module are dynamically associated through shared variables. In the lithography area scheduling embodiment, the user configures the data item variable name "priority_level" and selects the variable assignment "high" in the algorithm model, and the parameter parsing module ensures that data screening is synchronized with the algorithm target. Through this decoupled design, the data model configuration parameters are independent of the algorithm model configuration parameters, but can dynamically adapt to business changes, achieving a high degree of modularity and scalability of the system, significantly reducing the coupling degree, enabling semiconductor factories to quickly adapt to new requirements and improving the overall operation efficiency.

[0057] In one embodiment, the data configuration module is further configured to receive the machine group configuration parameters, where the machine group configuration parameters include the machine group name, machine type, and operation mode. The machine group is used to define a set of machine combinations with the same processing technology, and the parameter parsing module dynamically adjusts the data screening range in the data pre - processing rules according to the machine group configuration parameters.

[0058] When implemented, in the lithography area scheduling scenario, the user inputs the machine group configuration parameters through the data configuration module. For example, the machine group name is set to "IMM lithography machine group", the machine type is specified as a lithography machine, and the operation mode is selected as the single - boat mode to define a set of machine combinations with the same processing technology. When implemented, the data configuration module provides a form interface for the user to input these parameters, and the system background stores the parameters as machine group objects; the parameter parsing module is connected to the data configuration module. After parsing the machine group configuration parameters, it dynamically generates data pre - processing rules. For example, in the data acquisition stage, the rule requires only screening the machine data belonging to the IMM lithography machine group and excluding other types of machines. Through this fine - tuning of the machine group parameters, precise control of the data screening range is achieved, ensuring the optimization of the scheduling model for a specific production area, avoiding data redundancy, and improving processing efficiency; in the lithography area example, when the user adds a similar machine group, only the parameters need to be adjusted without modifying the core rules, enhancing the system's reusability and adaptability.

[0059] In one embodiment, the target class parameter in the algorithm configuration module includes multiple optimization objectives, each of which is configured with a priority order and a weight value. The model execution module uses a multi-objective optimization algorithm to solve the problem according to the priority order and weight value, wherein objectives with higher priority are satisfied first, and the weight value is used to adjust the optimization intensity when priorities are the same.

[0060] In implementation, during scheduling in the lithography area, users define multiple optimization objectives through the algorithm configuration module. For example, they can select a priority violation objective as the first optimization objective, setting its priority order to 1 and its weight to 0.7, and select minimizing the overall scheduling time as the second optimization objective, with a priority order of 2 and a weight to 0.3. In practice, the algorithm configuration module uses a graphical interface that allows users to drag and drop the order of objectives and input weight values; the parameters are parsed into a list of objectives. The model execution module executes multi-objective optimization algorithms based on these parameters, such as using weighted sum or priority methods. During the solution process, higher-priority objectives are prioritized (e.g., ensuring high-priority lots are processed first), and the optimization intensity is adjusted by weight values ​​when priorities are equal (e.g., objectives with higher weights receive more resource allocation). Through refined configuration of multi-objective parameters, flexible optimization of the scheduling algorithm is achieved, enabling users to dynamically adjust the optimization direction according to business needs, improving the accuracy and practicality of the scheduling results. In the lithography area scenario, this configuration method avoids the limitations of a single objective and supports comprehensive optimization in complex production environments.

[0061] In one embodiment, in the algorithm configuration module, the constraint parameters include common constraints and specific constraints. The common constraints are reusable constraints for multiple algorithm models, and the specific constraints are constraints specific to a single algorithm model. Each constraint is configured with variable parameters, including time limit parameters or resource quantity parameters. The parameter parsing module dynamically injects constraint rules into the algorithm model instance according to the variable parameters.

[0062] During implementation, in the lithography area scheduling, users select constraint parameters through the algorithm configuration module. These include common constraints such as the latest start time constraint, with its variable parameter, the site time constraint duration, set to 1 hour; and specific constraints such as the number of wafer loading / unloading ports in the lithography area, with its variable parameter, the number of wafer loading / unloading ports, set to 4. In practice, the algorithm configuration module categorizes constraints into common and specific libraries. Users select the desired constraints and input variable parameter values. The parameter parsing module parses these parameters and dynamically injects constraint rules into the algorithm model instance. For example, it maps the site time constraint duration to a time limit rule and the number of ports to a resource limit rule. Through constraint classification and parameterized design, modular management of constraint rules is achieved, allowing common constraints to be reused across multiple models (e.g., the latest start time constraint also applies to the diffusion area), while specific constraints focus on local needs, reducing configuration complexity. In the lithography area example, this design ensures comprehensive coverage of business constraints, improving the system's scalability and maintainability.

[0063] In one embodiment, the common constraints in the constraint class parameters include the latest start time constraint, whose variable parameter is the site time constraint duration. The site time constraint duration represents the time constraint that must be followed between sites. The model execution module ensures that the processing start time is no later than the sum of the specified time and the site time constraint duration when solving the problem. The specific constraints include the wafer loading and unloading port number constraint in the lithography area, whose variable parameter is the number of wafer loading and unloading ports. The model execution module limits the number of wafer batches processed simultaneously based on the number of wafer loading and unloading ports.

[0064] In implementation, in the lithography area scheduling scenario, when configuring the latest start time constraint, the user sets the variable parameter site time constraint duration to 1 hour, meaning the processing start time cannot be later than the planned time plus 1 hour. Simultaneously, a constraint on the number of wafer loading / unloading ports in the lithography area is configured, setting the variable parameter number of wafer loading / unloading ports to 4, indicating that the number of wafer batches processed simultaneously by the machine is limited by the number of ports. During implementation, the parameter parsing module converts these variable parameters into specific rules in the algorithm model instance. The model execution module ensures that the processing time conforms to the site constraints and limits the number of ports used during the solution process. Through the parameterization of specific constraint instances, accurate modeling of specific semiconductor manufacturing scenarios is achieved, enabling the scheduling algorithm to comply with actual business rules. In the lithography area example, this implementation avoids time conflicts and resource overload, improving the reliability and efficiency of the production plan.

[0065] In one embodiment, the data configuration module also allows users to configure variable names for data items, the algorithm configuration module requires the variable names to be assigned values ​​when selecting data items, and the parameter parsing module associates the data model configuration parameters with the algorithm model configuration parameters through the variable names, thereby decoupling the data parameters and the algorithm model.

[0066] In implementation, during scheduling in the lithography area, users configure variable names for data items through the data configuration module. For example, they can set the variable name "priority_level" for the data item "high-priority Lot filtering." The algorithm configuration module requires assigning values ​​to variable names when selecting data items, such as assigning "high" to "priority_level." In practice, the data configuration module provides a variable name field for user input, and the parameter parsing module reads the variable name and assigned value, dynamically linking the data model and algorithm model. For example, in the data preprocessing stage, high-priority Lot data is filtered based on the assigned value. This variable name mechanism decouples data parameters and algorithm models, allowing users to independently configure data input without binding to algorithm logic. In the lithography area scenario, this design supports dynamic parameter adjustment, reduces repetitive configuration work, and improves system flexibility and user experience.

[0067] In one embodiment, the task scheduling module is further configured with a running frequency parameter for periodically triggering scheduling tasks. The task scheduling module dynamically manages scheduling tasks based on the creation and deletion of algorithm model instances, wherein a corresponding scheduling task is automatically created when a new algorithm model instance is added, and a corresponding scheduling task is automatically deleted when an algorithm model instance is deleted. The data acquisition module updates the preprocessed data according to the latest configuration parameters each time a task is executed.

[0068] In implementation, during scheduling in the lithography area, users set the execution frequency parameters through the task scheduling module, such as triggering a scheduled task every 5 minutes. Based on these parameters, the task scheduling module creates periodic tasks, triggering the data acquisition module to retrieve the latest data from the source data table. In practice, the task scheduling module uses a timer mechanism to start tasks, and the data acquisition module updates preprocessed data according to the latest configuration parameters each time it executes, such as refreshing the Lot status and machine availability. By configuring the execution frequency, the scheduled tasks are executed automatically and periodically, ensuring data real-time performance. In the lithography area example, this periodic scheduling avoids frequent manual triggering, improving system response speed and production efficiency.

[0069] In one embodiment, the scheduling algorithm used by the model execution module is based on operations research constraint planning. The algorithm model instance is composed of multiple algorithm base classes, each of which represents an abstract business rule, including a target base class or a constraint base class.

[0070] In implementation, during scheduling in the lithography area, the model execution module uses a constraint programming algorithm. The algorithm model instance is composed of multiple basic algorithm classes, such as target classes like priority violation targets and constraint classes like earliest start time constraints. During implementation, the parameter parsing module instantiates these basic classes according to user configuration, and the model execution module combines them for solving. Through the modular combination of algorithm basic classes, the scheduling algorithm achieves high configurability, supporting the flexible application of complex business rules. In the lithography area scenario, this design improves the algorithm's scalability and solution accuracy.

[0071] In one embodiment, the target base class includes a priority violation target, a target for minimizing the overall scheduling duration, or a resource utilization target; the constraint base class includes an earliest start time constraint or a machine processing quantity constraint; and the parameter parsing module dynamically instantiates the target base class or constraint base class based on the user-configured selection and parameters.

[0072] During implementation, in the lithography area scheduling, users select target base classes (such as priority violation targets) and constraint base classes (such as earliest start time constraints) through the algorithm configuration module. The parameter parsing module dynamically instantiates these classes based on user configuration, for example, setting parameters for priority violation targets. In implementation, the base classes represent abstract business rules, which are then integrated into the algorithm model after instantiation. Through dynamic instantiation, precise mapping of business rules is achieved, enabling the system to quickly adapt to changing requirements. In the lithography area example, this implementation enhances the model's practicality and adaptability.

[0073] In one embodiment, the result processing module also converts the scheduling results into a platform-compatible data format for visualization or further analysis, completing the entire configuration-to-execution closed loop.

[0074] In implementation, during the lithography area scheduling, after the model execution module generates the scheduling results, the result processing module converts them into JSON or XML format for visualization or analysis. Specifically, the result processing module uses a data converter to standardize the results before outputting them to the platform interface. This standardization of the result format achieves a closed loop from configuration to execution, supporting data-driven decision optimization. In the lithography area scenario, this processing improves the usability of the results and the system's integrability.

[0075] Example 2

[0076] like Figure 2 As shown, a semiconductor scheduling algorithm model configuration method includes the following steps:

[0077] S1: Receive data model configuration parameters input by the user. The data model configuration parameters include a set of data items, a type of data item, and a scope of data item. The set of data items is used to specify the data type to be processed, including wafer batch data, machine data, or processing relationship data. The type of data item includes a filtering type, a priority type, or a supplementary type. The scope of data item is used to calibrate the machine processing mode.

[0078] S2: Receive algorithm model configuration parameters input by the user. The algorithm model configuration parameters include target class parameters and constraint class parameters. The target class parameters are used to define the optimization target and its priority weight, and the constraint class parameters are used to define business constraints and their variable parameters.

[0079] S3: Parse the data model configuration parameters and algorithm model configuration parameters to generate data preprocessing rules and algorithm model instances;

[0080] S4: Create a scheduling task based on the parsed parameters, and trigger the data acquisition operation of the data acquisition module and the scheduling algorithm execution operation of the model execution module, wherein the model execution module executes based on the algorithm model instance generated by the parameter parsing module;

[0081] S5: Obtain raw data from the source data table, and filter and transform the raw data according to the data preprocessing rules to generate preprocessed data;

[0082] S6: Execute the scheduling algorithm based on the algorithm model instance and preprocessed data to generate scheduling results;

[0083] S7: Post-process the scheduling results and convert them into standardized output data.

[0084] The semiconductor scheduling algorithm model configuration method provided in this embodiment has the same implementation principle and technical effect as the system embodiment in Embodiment 1. For the sake of brevity, any parts not mentioned in the method embodiment can be referred to the corresponding content in Embodiment 1.

[0085] Example 3

[0086] A computer-readable storage medium having a computer program stored thereon, the computer program being executed by a computer to perform the semiconductor scheduling algorithm model configuration method described in Embodiment 2 above.

[0087] Example 4

[0088] An electronic device includes: a memory and a processor, wherein the processor and the memory are connected;

[0089] The memory is used to store programs;

[0090] The processor calls a program stored in the memory to execute a semiconductor scheduling algorithm model configuration method as described in Embodiment 2.

[0091] It should be noted that the electronic device mentioned may be, but is not limited to, personal computers (PCs), tablet computers, mobile internet devices (MIDs), etc.

[0092] It should be noted that processors, memory, and other components that may be present in electronic devices are electrically connected to each other, directly or indirectly, to enable data transmission or interaction. For example, processors, memory, and other components may be electrically connected to each other via one or more communication buses or signal lines.

[0093] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0094] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0095] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0096] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, laptop, server, mobile phone, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any simple modifications, alterations, or equivalent structural changes made to the above embodiments based on the technical essence of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A semiconductor scheduling algorithm model configuration system, characterized by, The application relates to a data processing system, comprising: a data configuration module for receiving user-inputted data model configuration parameters, the data model configuration parameters comprising a data item set, a data item type and a data item action range, wherein the data item set is used for specifying a data type to be processed, the data item type comprises a screening type, a priority type or a supplementary type, and the data item action range is used for marking a machine processing mode; an algorithm configuration module for receiving user-inputted algorithm model configuration parameters, the algorithm model configuration parameters comprising target class parameters and constraint class parameters, wherein the target class parameters are used for defining an optimization target and a priority weight thereof, and the constraint class parameters are used for defining a business constraint and a variable parameter thereof; a parameter analysis module connected with the data configuration module and the algorithm configuration module, and used for analyzing the data model configuration parameters and the algorithm model configuration parameters, and generating a data preprocessing rule and an algorithm model instance; a task scheduling module connected with the parameter analysis module, and used for creating a scheduling task based on the analyzed parameters, and triggering a data acquisition operation of a data acquisition module and a scheduling algorithm execution operation of a model execution module, wherein the model execution module is executed based on the algorithm model instance generated by the parameter analysis module; the data acquisition module connected with the task scheduling module, and used for acquiring original data from a source data table, and screening and converting the original data according to the data preprocessing rule to generate preprocessed data; the model execution module connected with the data acquisition module and the parameter analysis module, and used for executing a scheduling algorithm according to the algorithm model instance and the preprocessed data to generate a scheduling result; a result processing module connected with the model execution module, and used for post-processing the scheduling result to convert the scheduling result into standard output data.

2. The semiconductor scheduling algorithm model configuration system of claim 1, wherein, The data configuration module is further used for receiving machine group configuration parameters, the machine group configuration parameters comprising a machine group name, a machine type and a running mode, wherein the machine group is used for defining a combination of a plurality of machines with the same processing technology, and the parameter analysis module dynamically adjusts a data screening range in the data preprocessing rule according to the machine group configuration parameters.

3. The semiconductor scheduling algorithm model configuration system of claim 1, wherein, In the algorithm configuration module, the target class parameters comprise a plurality of optimization targets, each optimization target is configured with a priority order and a weight value, the model execution module adopts a multi-target optimization algorithm for solving according to the priority order and the weight value, wherein a target with a high priority is preferentially met, and the weight value is used for adjusting an optimization intensity when the priorities are the same.

4. The semiconductor scheduling algorithm model configuration system of claim 1, wherein, In the algorithm configuration module, the constraint class parameters comprise common constraints and specific constraints, the common constraints are reusable constraints for a plurality of algorithm models, and the specific constraints are constraints exclusively belonging to a single algorithm model, wherein each constraint is configured with a variable parameter, the variable parameter comprises a time limit parameter or a resource quantity parameter, and the parameter analysis module dynamically injects a constraint rule into the algorithm model instance according to the variable parameter.

5. The semiconductor scheduling algorithm model configuration system of claim 4, wherein, The common constraint in the constraint class parameter includes a latest start time constraint, and a variable parameter of the latest start time constraint is a site time constraint duration, the site time constraint duration represents a time constraint that must be observed between sites, and the model execution module ensures that the processing start time is not later than the sum of the specified time and the site time constraint duration; the specific constraint includes a lithography area wafer loading and unloading port quantity constraint, and a variable parameter of the lithography area wafer loading and unloading port quantity constraint is a wafer loading and unloading port number, and the model execution module limits the number of wafer batches processed simultaneously according to the wafer loading and unloading port number.

6. The semiconductor scheduling algorithm model configuration system of claim 1, wherein, The data configuration module also allows a user to configure a variable name for a data item, the algorithm configuration module requires the variable name to be assigned when selecting the data item, and the parameter analysis module associates the data model configuration parameter and the algorithm model configuration parameter through the variable name.

7. The semiconductor scheduling algorithmic model configuration system of claim 1, wherein, The task scheduling module is also configured with a running frequency parameter for triggering the scheduling task periodically, and the task scheduling module dynamically manages the scheduling task according to the creation and deletion of the algorithm model instance, wherein a corresponding scheduling task is automatically created when a new algorithm model instance is added, and a corresponding scheduling task is automatically deleted when an algorithm model instance is deleted; the data acquisition module updates the pre-processing data according to the latest configuration parameter each time the task is executed.

8. A method of configuring a model of a semiconductor scheduling algorithm, the method comprising: receiving a plurality of parameters of the model; and determining a configuration of the model based on the plurality of parameters. The method comprises the following steps: S1: receiving a data model configuration parameter input by a user, the data model configuration parameter comprising a data item set, a data item type and a data item action range, wherein the data item set is used to specify a data type to be processed, including wafer batch data, machine data or processing relationship data, the data item type comprises a screening type, a priority type or a supplement type, and the data item action range is used to mark a machine processing mode; S2: receiving an algorithm model configuration parameter input by a user, the algorithm model configuration parameter comprising a target class parameter and a constraint class parameter, wherein the target class parameter is used to define an optimization target and a priority weight thereof, and the constraint class parameter is used to define a business constraint and a variable parameter thereof; S3: analyzing the data model configuration parameter and the algorithm model configuration parameter to generate a data pre-processing rule and an algorithm model instance; S4: creating a scheduling task based on the analyzed parameter, and triggering a data acquisition operation of a data acquisition module and a scheduling algorithm execution operation of a model execution module, wherein the model execution module is executed based on the algorithm model instance generated by a parameter analysis module; S5: acquiring original data from a source data table, and screening and converting the original data according to the data pre-processing rule to generate pre-processing data; S6: executing a scheduling algorithm according to the algorithm model instance and the pre-processing data to generate a scheduling result; S7: post-processing the scheduling result to convert it into standard output data.

9. An electronic device, comprising: The method comprises: a memory and a processor, the processor and the memory are connected; the memory is used to store a program; the processor calls the program stored in the memory to execute the method in claim 8.

10. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is run by a computer to execute the method in claim 8.

Images