Model training method, dispatching method and device

Abstract

The present application relates to the field of semiconductor manufacturing, and in particular provides a model training method, a dispatching method and equipment. The model training method comprises: obtaining manufacturing environment information of a target wafer; determining a dispatching machine from target machines capable of performing a target process based on the manufacturing environment information and a neural network model; in response to the target wafer completing manufacturing, obtaining a feedback parameter associated with the production efficiency of the dispatching machine; and updating the neural network model according to the feedback parameter and a preset feedback parameter threshold. Since the manufacturing environment information takes into account the current capacity of the machine, the neural network model can greatly improve the capacity of each machine in the manufacturing process based on the dispatching machine predicted based on the manufacturing environment information, effectively avoiding the idle capacity of the production line caused by the accumulation of different stations of products.

Description

Technical Field

[0001] This application relates to the field of semiconductor manufacturing, and in particular to a model training method, a dispatching method, and equipment. Background Technology

[0002] For the semiconductor industry, especially wafer foundries, the capital-intensive nature of their operations necessitates the use of technologically advanced and expensive equipment. Higher equipment utilization efficiency leads to higher output and better meets production line demands. However, excessively high utilization also results in faster equipment depreciation, impacting wafer yield. Therefore, effectively utilizing equipment during daily operations is a crucial aspect of daily wafer production.

[0003] Currently, many companies have introduced dispatching systems to automate the dispatching of wafers. The dispatching strategies obtained by these systems are mainly to ensure that the quality of the wafers themselves is not at risk and to improve the wafer yield, but it is difficult to guarantee the maximum capacity of the wafers. Summary of the Invention

[0004] This application is made in view of the above-mentioned problems. This application provides a model training method, a task assignment method, and an apparatus.

[0005] According to the first aspect of this application, a model training method is provided, comprising:

[0006] Obtain manufacturing environment information for the target wafer, which is the wafer that needs to be manufactured according to the target process in the manufacturing process. The manufacturing environment information is used to indicate the current capacity of the machines that can perform the target process.

[0007] Based on the manufacturing environment information and neural network model, a dispatching machine is determined from the target machines capable of performing the target process, wherein the dispatching machine is used to manufacture the target wafer according to the target process;

[0008] In response to the completion of manufacturing of the target wafer, feedback parameters related to the production efficiency of the dispatching machine are obtained;

[0009] The neural network model is updated based on the feedback parameters and the preset feedback parameter threshold.

[0010] Furthermore, according to the model training method of the first aspect of this application, the manufacturing environment information includes the first queuing time between the target process and the next process of the target process, and the current capacity parameters of the target machine.

[0011] Determining the dispatch machine from the target machine capable of performing the target process based on the manufacturing environment information and neural network model further includes:

[0012] Based on the current production capacity parameters, the neural network model determines the manufacturing time required for the target machine to manufacture the target wafer according to the target process;

[0013] Select at least one first machine from the target machines whose manufacturing time is less than or equal to the first queuing time;

[0014] The dispatching machine is selected from the at least one first machine.

[0015] Furthermore, according to the model training method of the first aspect of this application, the manufacturing environment information includes the current capacity parameters of the target machine, and determining the dispatch machine from the target machines capable of performing the target process based on the manufacturing environment information and the neural network model further includes:

[0016] Based on the current production capacity parameters of the target machine, the production line load balance result of the target machine is determined through the neural network model;

[0017] From the target machines, at least one second machine whose production line load balance result satisfies the first condition is selected;

[0018] The dispatching machine is selected from the at least one second machine.

[0019] Furthermore, according to the model training method of the first aspect of this application, the manufacturing environment information includes the historical end time of manufacturing the target wafer according to the previous process of the target process and the second queuing time between the target process and the previous process, and determining the dispatching machine from the target machines capable of performing the target process based on the manufacturing environment information and the neural network model further includes:

[0020] Based on the historical end time and the second queuing time, determine the start time for manufacturing the target wafer according to the target process;

[0021] The neural network model is used to select at least one third machine from the target machines that is not at full load at the start time;

[0022] The dispatching machine is selected from the at least one third machine.

[0023] Furthermore, according to the model training method of the first aspect of this application, the manufacturing environment information further includes the number of wafers produced between the previous process and the target process, and the selection of at least one third machine from the target machines that is not at full load at the start time using the neural network model further includes:

[0024] Obtain the number of wafers currently being manufactured by the target machine;

[0025] At least one third machine is selected from the target machines whose number of currently manufactured wafers is less than or equal to the quantity difference, wherein the quantity difference is the difference between the number of wafers fed and the number of target wafers.

[0026] Furthermore, according to the model training method of the first aspect of this application, selecting the dispatching machine from the at least one first machine includes:

[0027] Calculate the machine utilization rate of the at least one first machine;

[0028] Among the at least one first machine, the machine whose utilization rate meets the second condition is selected as the dispatching machine.

[0029] Furthermore, according to the model training method of the first aspect of this application, the feedback parameters include at least one of the following:

[0030] The actual queuing time for the target process;

[0031] The actual utilization rate of the dispatching machine; or

[0032] The actual production line balance rate of the dispatching machine.

[0033] According to a second aspect of this application, a dispatching method is provided, comprising:

[0034] Obtain the current manufacturing environment information of the current wafer, which is the wafer that needs to be manufactured according to the current process in the current manufacturing process. The current manufacturing environment information is used to indicate the current capacity of the machine that can perform the current process.

[0035] Using the neural network model trained by the model training method described in the first aspect and the current manufacturing environment information, a current dispatching machine is determined from the machines capable of performing the current process, wherein the current dispatching machine is used to manufacture the current wafer according to the current process.

[0036] According to a third aspect of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the first or second aspect.

[0037] According to a fourth aspect of this application, a computer-readable storage medium is provided that stores a computer program / instructions thereon, which, when executed by a processor, implement the steps of the method described in the first or second aspect.

[0038] According to a fifth aspect of this application, a computer program product is provided, comprising a computer program / instructions that, when executed by a processor, implement the steps of the method described in the first or second aspect.

[0039] As will be described in detail below, the model training method according to the embodiments of this application inputs the manufacturing environment information of the target wafer into the neural network model to indicate the current capacity of the machine that can perform the target process. Since the manufacturing environment information takes into account the current capacity of the machine, the neural network model can greatly improve the capacity of each machine in the manufacturing process by predicting the dispatching machine based on the manufacturing environment information, and effectively avoid the idle capacity caused by the accumulation of products at different stations on the production line.

[0040] It should be understood that both the foregoing general description and the following detailed description are exemplary and intended to provide further illustration of the claimed technology. Attached Figure Description

[0041] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the accompanying drawings, the same reference numerals generally represent the same parts or processes.

[0042] Figure 1 This is a flowchart illustrating a model training method according to an embodiment of this application.

[0043] Figure 2 This is a schematic diagram illustrating the number of wafers fed into a process under a certain manufacturing process according to an embodiment of this application.

[0044] Figure 3 This is a schematic diagram illustrating the process of determining the dispatch machine from target machines capable of performing the target process based on manufacturing environment information and a neural network model according to an embodiment of this application.

[0045] Figure 4 This is a flowchart illustrating a dispatching method according to an embodiment of this application.

[0046] Figure 5 This is a structural diagram illustrating a model training apparatus according to an embodiment of this application.

[0047] Figure 6 This is a structural diagram illustrating a dispatching device according to an embodiment of this application.

[0048] Figure 7 This is a hardware block diagram illustrating an electronic device according to an embodiment of this application.

[0049] Figure 8 This is a schematic diagram illustrating a computer program product according to an embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0051] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0052] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0053] In related technologies, the dispatching systems introduced by semiconductor companies mainly consider the following factors when generating dispatching strategies:

[0054] Product Q-time (queue time), product grade, product type, and product size (number of pieces);

[0055] Whether they belong to the same Season Group, whether a process change is required, and whether continuous operation is possible;

[0056] The presence of a buffer or occupied machine space in the hardware determines whether to dispatch goods.

[0057] When multiple machines can be assigned work, are there any preferences set?

[0058] These factors all affect wafer quality, so the dispatching strategy determined based on these parameters is mainly to ensure that the quality of the wafer itself is not at risk and to improve the wafer yield, but it is difficult to guarantee the maximum capacity of the machine.

[0059] To alleviate the technical problems existing in related technologies, embodiments of this application provide a model training method, a dispatching method, and an apparatus. This method inputs the manufacturing environment information of the target wafer into a neural network model, indicating the current capacity of the machines capable of performing the target process. Since the manufacturing environment information takes into account the current capacity of the machines, the neural network model can significantly improve the capacity of each machine in the manufacturing process by predicting the dispatching machines based on this manufacturing environment information, effectively avoiding idle capacity caused by product accumulation at different stations on the production line.

[0060] The following is a description of the terminology used in this application.

[0061] Wafer manufacturing processes refer to a series of technological steps that transform semiconductor materials into electronic devices. A wafer is a thin, round silicon wafer and is the foundation of semiconductor devices. The goal of wafer manufacturing processes is to transfer circuit patterns onto the wafer and form the structure of electronic devices through a series of processing steps.

[0062] Queue time (Q-time) refers to the time in semiconductor manufacturing where a wafer waits for the next process to begin after completing one process. It affects production efficiency, quality, and delivery time.

[0063] Wafer per hour (WPH) is a metric for measuring production line efficiency, indicating the number of wafers a production line can produce within a specific time frame. Improving WPH is key to increasing production efficiency and enhancing competitiveness, and this requires optimizing processes and enhancing equipment performance.

[0064] Overall equipment effectiveness (OEE) refers to the ratio of an equipment's actual production capacity to its theoretical production capacity.

[0065] Manufacturing Execution System (MES) is used to coordinate various production stages, monitor production activities in real time, and provide detailed production data to improve the execution efficiency of the entire production process.

[0066] To facilitate understanding of this embodiment, a model training method disclosed in this application will first be described in detail. The execution subject of the model training method provided in this application embodiment is generally an electronic device with certain computing power, such as a terminal device, a server, or other processing device. This electronic device may be equipped with a Model Execution System (MES), which executes the model training method in this application embodiment. In some possible implementations, the model training method can be implemented by a processor calling computer-readable instructions stored in memory.

[0067] See Figure 1 The diagram shown is a flowchart of a model training method provided in an embodiment of this application. The method includes the following steps:

[0068] Step 101: Obtain the manufacturing environment information of the target wafer. The target wafer is the wafer that needs to be manufactured according to the target process in the manufacturing process. The manufacturing environment information is used to indicate the current capacity of the machines that can perform the target process.

[0069] In this embodiment, the target process is any one of the manufacturing processes, and the target process is the process in which the target wafer will be manufactured.

[0070] In one or more embodiments, manufacturing environment information can be specifically categorized as follows:

[0071] Machine information, wafer information, operation time information, and the number of wafers fed into each process in the manufacturing process.

[0072] The machine information includes the machine number, model, quantity, and current capacity parameters. The current capacity parameters include, but are not limited to, the machine's current WPH.

[0073] Wafer information includes the wafer number, type, and quantity of the target wafer.

[0074] The job time information includes the Q-time of two adjacent steps in the manufacturing process, the historical end time of the previous step in manufacturing the target wafer, and the manufacturing time required for the machine to manufacture the target wafer according to the target step based on the machine's current capacity forecast.

[0075] The wafer feed count for a process refers to the maximum number of wafers allowed to be fed when the Q-time of two adjacent processes is met. As an example, see... Figure 2 This diagram illustrates the number of wafers fed into a specific manufacturing process. For example... Figure 2As shown, the manufacturing process includes four steps: A, B, C, and D. The Q-time between step A and step C is 24 hours. To meet this Q-time, the maximum number of wafers allowed to be placed between step A and step C is 1500. In other words, if the number of wafers placed between step A and step C exceeds 1500, the Q-time between step A and step C will exceed 24 hours, and the quality of the manufactured wafers will not meet the requirements.

[0076] It should be noted that although manufacturing environment information can be categorized as described above, the specific manufacturing environment information obtained when executing the model training method of this application embodiment can be either all of the above-mentioned manufacturing environment information or only a portion thereof. For example, only the first queuing time between the target process and the next process after the target process, and the current capacity parameters of the target machine, are obtained. Or, only the second queuing time between the target process and the previous process, and the historical end time of the previous process of manufacturing the target wafer, are obtained.

[0077] The previous process, the next process, and the target process are all processes in the same manufacturing process. The previous process and the next process are adjacent to the target process. The previous process is the process that occurs before the target process, and the next process is the process that occurs after the target process.

[0078] Step 102: Determine the dispatching machine from the target machines capable of performing the target process based on manufacturing environment information and neural network model, wherein the dispatching machine is used to manufacture the target wafer according to the target process.

[0079] Depending on the specific parameters included in the manufacturing environment information, the following methods are provided for determining the dispatching machine.

[0080] The first method obtains manufacturing environment information including the first queuing time between the target process and the next process after the target process, as well as the current capacity parameters of the target machine.

[0081] Determining the dispatching machine from the target machine capable of performing the target process based on manufacturing environment information and neural network models may further include the following steps:

[0082] Based on current capacity parameters, a neural network model is used to determine the manufacturing time required for the target machine to manufacture the target wafer according to the target process;

[0083] Select at least one first machine whose manufacturing time is less than or equal to the first queuing time from the target machines;

[0084] Select a dispatch machine from at least one first machine.

[0085] In this embodiment, the neural network model can estimate the manufacturing time required for the target machine to manufacture the target wafer according to the target process based on the current capacity parameters of the target machine, so as to use at least the time to select at least one first machine from the target machines.

[0086] In one embodiment, if the acquired manufacturing environment information includes only the first queuing time and the current capacity parameters of the target machine as shown in the first manner, and does not include parameters shown in other manners, a machine can be randomly selected from at least one first machine as the dispatching machine.

[0087] In another embodiment, when the acquired manufacturing environment information includes parameters shown in several other ways in addition to the first queuing time and the current capacity parameters of the target machine as shown in the first method, the dispatching machine can be selected from at least one first machine by referring to the parameters given in the other methods.

[0088] In yet another embodiment, an implementation scheme for selecting a dispatch machine from at least one first machine based on machine utilization is also provided, which specifically includes the following steps:

[0089] Calculate the machine utilization rate of at least one first machine;

[0090] Select the machine whose utilization rate meets the second condition from at least one first machine as the dispatch machine.

[0091] The second condition includes, but is not limited to, the highest machine utilization rate, or the machine utilization rate being greater than or equal to a preset machine utilization rate threshold. This embodiment does not specifically limit this.

[0092] The second approach involves manufacturing environment information, including the current production capacity parameters of the target machine.

[0093] Determining the dispatching machine from the target machine capable of performing the target process based on manufacturing environment information and neural network models may include the following steps:

[0094] Based on the current capacity of the target machine, the production line load balance result of the target machine is determined through the neural network model.

[0095] From the target machines, select at least one second machine whose production line load balance result satisfies the first condition;

[0096] Obtain a dispatch machine from at least one second machine.

[0097] In this embodiment, the production line load balance result meeting the first condition includes, but is not limited to, the production line load balance result being the highest, or the production line load balance result being greater than the production line load balance result threshold, etc. This embodiment does not make specific limitations on this.

[0098] In one example, when selecting at least one second machine, the production line load balance results of the target machines can be sorted from high to low or from low to high. Then, based on the sorting results, the N target machines with the highest production line load balance results are selected as the second machines.

[0099] In one example, there are 3 target machines, denoted as target machine 1, target machine 2 and target machine 3. The result of sorting the production line load balance of these three target machines is target machine 1 > target machine 2 > target machine 3. If the preset number of second machines obtained by screening is 2, then target machine 1 and target machine 2 can be used as second machines.

[0100] In another example, when selecting at least one second machine, the difference between the production line load balance result of each target machine and the preset production line load balance result threshold can be calculated, and the target machine corresponding to the largest M production line load balance result differences can be selected as the second machine.

[0101] In one embodiment, if the acquired manufacturing environment information includes only the current capacity parameters of the target machine shown in the second manner, and does not include parameters shown in other manners, a machine can be randomly selected from at least one second machine as the dispatching machine.

[0102] In another embodiment, when the acquired manufacturing environment information includes not only the current capacity parameters of the target machine shown in the second manner, but also parameters shown in other manners, the dispatching machine can be selected from at least one second machine by referring to the parameters given in the other manners.

[0103] As an example, if the acquired manufacturing environment information also includes the first queuing time from the first approach, then selecting dispatch machines from at least one second machine may include the following steps:

[0104] Based on the current capacity parameters of each second machine, the manufacturing time required for at least one second machine to manufacture the target wafer according to the target process is determined through a neural network model.

[0105] Select a target second machine from at least one second machine whose manufacturing time is less than or equal to the first queuing time;

[0106] Select a dispatch machine from the target second machine.

[0107] In one embodiment, if the acquired manufacturing environment information includes only the first queuing time and the current capacity parameters of the target machine shown in the first and second methods, a machine can be randomly selected from the target second machine as the dispatching machine.

[0108] In another embodiment, when the acquired manufacturing environment information includes parameters shown in several other ways in addition to the first queuing time and the current capacity parameters of the target machine, the dispatching machine can be selected from the target second machine by referring to the parameters given in the other ways.

[0109] The third approach involves manufacturing environment information including the historical end time of manufacturing the target wafer according to the previous process of the target process, as well as the second queuing time between the target process and the previous process.

[0110] Determining the dispatching machine from the target machine capable of performing the target process based on manufacturing environment information and neural network models may further include the following steps:

[0111] Based on the historical end time and the second queue time, determine the start time for manufacturing the target wafer according to the target process.

[0112] Use a neural network model to select at least one third machine that is not at full load at the start time from the target machines;

[0113] Select a dispatch machine from at least one third machine.

[0114] The start time for manufacturing the target wafer according to the target process can be obtained by summing the historical end time and the second queue time.

[0115] In this embodiment, depending on the actual operating conditions of the target equipment, different target equipment may have different operating states at the start time. These operating states determine whether the target equipment can manufacture the target wafer according to the target process at the start time. For example, at the start time, some target equipment may be in an idle state, while others may be in a fully loaded state. Target equipment in an idle state can manufacture the target wafer according to the target process at the start time, while target equipment in a fully loaded state cannot manufacture the target wafer because all of its capacity has been used to manufacture other wafers.

[0116] In one embodiment, if the acquired manufacturing environment information includes only the second queuing time and historical end time shown in the third manner, and does not include parameters shown in other manners, a machine can be randomly selected from at least one third machine as the dispatching machine.

[0117] In another embodiment, when the acquired manufacturing environment information includes parameters shown in several other ways in addition to the second queuing time and historical end time shown in the third way, the dispatching machine can be selected from at least one third machine by referring to the parameters given in the other ways.

[0118] As an example, if the acquired manufacturing environment information also includes the current capacity parameters of the target machine, then selecting a dispatch machine from at least one third machine may include the following steps:

[0119] Based on the current production capacity parameters of each third machine, the production line load balance result of at least one third machine is determined through the neural network model.

[0120] From at least one third machine, select multiple target third machines with the highest production line load balancing results;

[0121] The dispatching machine is selected from multiple target third machines.

[0122] In one or more alternative embodiments, to improve wafer yield and ensure wafer quality, when the manufacturing environment information also includes the number of wafer starts between the previous process and the target process, using a neural network model to screen at least one third machine from the target machines that is not at full capacity at the start time may include the following steps:

[0123] Obtain the number of wafers currently being manufactured by the target machine;

[0124] Select at least one third machine from the target machines whose number of currently manufactured wafers is less than or equal to the quantity difference, where the quantity difference is the difference between the number of wafers fed and the number of target wafers.

[0125] In this embodiment, the number of wafers fed between the previous process and the target process is affected by the Q-time between the two processes. With a fixed Q-time, the number of wafers fed between the previous process and the target process is also a constant value. If the sum of the number of wafers currently being manufactured on the target machine and the target wafer count exceeds the number of wafers fed between the previous process and the target process, there are too many wafers between them. This makes it difficult to guarantee the manufacturing time of some wafers, thus affecting wafer quality or yield. Therefore, the number of wafers currently being manufactured on the third machine selected from the target machine should be less than or equal to the difference calculated based on the number of wafers fed and the target wafer count.

[0126] In one or more embodiments, when the manufacturing environment information includes a first queuing time between the target process and the next process after the target process, the current capacity parameters of the target machine, the historical end time of manufacturing the target wafer according to the previous process of the target process, a second queuing time between the target process and the previous process, and the number of wafer starts between the previous process and the target process, such as Figure 3 As shown, step 102 may include the following steps:

[0127] Step 301: Based on the historical end time and the second queuing time, determine the start time for manufacturing the target wafer according to the target process; and obtain the number of wafers currently being manufactured by the target machine.

[0128] Step 302: Select at least one third machine from the target machines whose number of wafers currently being manufactured is less than or equal to the quantity difference, wherein the quantity difference is the difference between the number of wafers fed and the number of target wafers;

[0129] Step 303: Based on the current production capacity parameters, determine the manufacturing time required for at least one third machine to manufacture the target wafer according to the target process using a neural network model;

[0130] Step 304: Select at least one target third machine from at least one third machine whose manufacturing time is less than or equal to the first queuing time;

[0131] Step 305: Based on the current capacity parameters of at least one target third machine, calculate the production line load balance result of at least one target third machine using a neural network model;

[0132] Step 306: From at least one target third machine, select at least one machine to be dispatched with the highest production line load balance result;

[0133] Step 307: Calculate the machine utilization rate of at least one machine to be dispatched;

[0134] Step 308: Select the machine with the highest utilization rate among at least one machine to be assigned as the machine to be assigned.

[0135] In one or more embodiments, any neural network model in the related art that can be executed according to the above implementation of the screening and dispatching machine as defined in this application can be used as the neural network model for step 102. For example, an LSTM (Long Short-Term Memory) model.

[0136] Step 103: In response to the completion of manufacturing of the target wafer, obtain feedback parameters related to the production efficiency of the dispatching machine.

[0137] In one or more embodiments, the feedback parameter includes at least one of the following:

[0138] The actual queuing time for the target process;

[0139] The actual utilization rate of the dispatching machine; or

[0140] The actual production line balance rate of the dispatching machine.

[0141] The actual queuing time for the target process includes the actual queuing time between the target process and the previous process, as well as the actual queuing time between the target process and the next process.

[0142] Step 104: Update the neural network model based on the feedback parameters and the preset feedback parameter thresholds.

[0143] In this embodiment, feedback parameters and preset feedback parameter thresholds can be used to calculate the model loss, and the model loss can be used to optimize the parameters of the neural network model until the neural network model converges.

[0144] In this embodiment, each feedback parameter has a corresponding feedback parameter threshold. For example, the feedback parameter threshold corresponding to the actual queuing time between the target process and the previous process is the Q-time between the target process and the previous process. Similarly, the feedback parameter threshold corresponding to the actual queuing time between the target process and the next process is the Q-time between the target process and the next process. As another example, the feedback parameter threshold corresponding to the actual machine utilization rate of the dispatching machine is the theoretical machine utilization rate of the dispatching machine. Likewise, the feedback parameter threshold corresponding to the actual production line balance rate of the dispatching machine is the theoretical production line balance rate of the dispatching machine.

[0145] It should be understood that after optimizing the parameters of the neural network model using model loss, if the neural network model does not converge, it should be iterated multiple times, i.e., returned to step 101 multiple times. Since the machine's capacity is affected by parameters such as the machine model and time point, the manufacturing environment information obtained each time step 101 is returned will be different. Specifically, different models of machines can provide different capacities for the same manufacturing process. At each point in time, the machine's operating conditions may be limited by hardware, software, or other factors, causing the available capacity to change in real time.

[0146] In this embodiment, in order to improve the convergence speed of the model, if the model loss calculated in the previous iteration indicates that the feedback parameter is better than the feedback parameter threshold, and the manufacturing environment information obtained in the current iteration is exactly the same as that in the previous iteration, it can be directly determined that the dispatching machine used in the current iteration is the same as that used in the previous iteration.

[0147] In the solution provided in this embodiment, the manufacturing environment information of the target wafer input to the neural network model indicates the current capacity of the machine that can perform the target process. Since the manufacturing environment information takes into account the current capacity of the machine, the neural network model can greatly improve the capacity of each machine in the manufacturing process by predicting the dispatching machine based on the manufacturing environment information, and effectively avoid the idle capacity caused by the accumulation of products at different stations on the production line.

[0148] This application also provides a dispatching method, the execution subject of which is generally an electronic device with a certain computing power, such as a terminal device, a server, or other processing device. The electronic device may be equipped with a Model Execution System (MES), which executes the model training method described in this application. In some possible implementations, the model training method can be implemented by a processor calling computer-readable instructions stored in memory.

[0149] like Figure 4 As shown, the method may include the following steps:

[0150] Step 401: Obtain the current manufacturing environment information of the current wafer. The current wafer is the wafer that needs to be manufactured according to the current process in the current manufacturing process. The current manufacturing environment information is used to indicate the current capacity of the machine that can perform the current process.

[0151] Step 402: Using the neural network model trained by the model training method in the foregoing embodiment and the current manufacturing environment information, determine the current dispatching machine from the machines capable of performing the current process, wherein the current dispatching machine is used to manufacture the current wafer according to the current process.

[0152] The process of selecting the current dispatching machine in step 402 is the same as the principle of selecting the dispatching machine in step 102 in the previous embodiment. To avoid redundancy, it will not be described in detail here.

[0153] This application also provides a model training apparatus for executing the model training method provided in any of the above embodiments. Figure 5 As shown, the device includes:

[0154] The first acquisition module 51 is used to acquire manufacturing environment information of the target wafer, wherein the target wafer is the wafer that needs to be manufactured according to the target process in the manufacturing process, and the manufacturing environment information is used to indicate the current capacity of the machine that can perform the target process.

[0155] The first determining module 52 is used to determine a dispatching machine from the target machines capable of performing the target process based on the manufacturing environment information and the neural network model, wherein the dispatching machine is used to manufacture the target wafer according to the target process;

[0156] The module 53 obtains feedback parameters related to the production efficiency of the dispatching machine in response to the completion of manufacturing of the target wafer;

[0157] The update module 54 is used to update the neural network model based on the feedback parameters and the preset feedback parameter threshold.

[0158] In one or more embodiments, the manufacturing environment information includes a first queuing time between the target process and the next process after the target process, and the current capacity parameters of the target machine.

[0159] The first determining module 52 is used for:

[0160] Based on the current production capacity parameters, the neural network model determines the manufacturing time required for the target machine to manufacture the target wafer according to the target process;

[0161] Select at least one first machine from the target machines whose manufacturing time is less than or equal to the first queuing time;

[0162] The dispatching machine is selected from the at least one first machine.

[0163] In one or more embodiments, the manufacturing environment information includes the current production capacity parameters of the target machine.

[0164] The first determining module 52 is used for:

[0165] Based on the current production capacity parameters of the target machine, the production line load balance result of the target machine is calculated through the neural network model;

[0166] From the target machines, at least one second machine whose production line load balance result satisfies the first condition is selected;

[0167] The dispatching machine is selected from the at least one second machine.

[0168] In one or more embodiments, the manufacturing environment information includes the historical end time of manufacturing the target wafer according to the previous process of the target process and a second queuing time between the target process and the previous process;

[0169] The first determining module 52 is used for:

[0170] Based on the historical end time and the second queuing time, determine the start time for manufacturing the target wafer according to the target process;

[0171] The neural network model is used to select at least one third machine from the target machines that is not at full load at the start time;

[0172] The dispatching machine is selected from the at least one third machine.

[0173] In one or more embodiments, the manufacturing environment information also includes the number of wafers fed between the previous process and the target process;

[0174] The first determining module 52 is used for:

[0175] Obtain the number of wafers currently being manufactured by the target machine;

[0176] At least one third machine is selected from the target machines whose number of currently manufactured wafers is less than or equal to the quantity difference, wherein the quantity difference is the difference between the number of wafers fed and the number of target wafers.

[0177] In one or more embodiments, the first determining module 52 is configured to:

[0178] Calculate the machine utilization rate of the at least one first machine;

[0179] Among the at least one first machine, the machine whose utilization rate meets the second condition is selected as the dispatching machine.

[0180] In one or more embodiments, the feedback parameter includes at least one of the following:

[0181] The actual queuing time for the target process;

[0182] The actual utilization rate of the dispatching machine; or

[0183] The actual production line balance rate of the dispatching machine.

[0184] The model training device and the model training method provided in this application are based on the same inventive concept and have the same beneficial effects as the methods they adopt, run or implement.

[0185] This application also provides a dispatching device for executing the dispatching method provided in any of the above embodiments. Figure 6 As shown, the device includes:

[0186] The second acquisition module 61 is used to acquire the current manufacturing environment information of the current wafer, wherein the current wafer is the wafer that needs to be manufactured according to the current process in the current manufacturing process, and the current manufacturing environment information is used to indicate the current capacity of the machine that can perform the current process;

[0187] The second determining module 62 is used to select a current dispatching machine from the machines capable of performing the current process using the neural network model trained by the model training method described in the foregoing embodiment and the current manufacturing environment information, wherein the current dispatching machine is used to manufacture the current wafer according to the current process.

[0188] This application also provides an electronic device for executing the above-described model training method or dispatching method. Please refer to... Figure 7It illustrates a schematic diagram of an electronic device provided by some embodiments of this application. For example... Figure 7 As shown, the electronic device 7 includes: a processor 700, a memory 701, a bus 702, and a communication interface 703. The processor 700, the communication interface 703, and the memory 701 are connected via the bus 702. The memory 701 stores a computer program that can run on the processor 700. When the processor 700 runs the computer program, it executes the model training method or dispatching method provided in any of the foregoing embodiments of this application.

[0189] The memory 701 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this device network element and at least one other network element is achieved through at least one communication interface 703 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0190] Bus 702 can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory 701 is used to store programs. After receiving an execution instruction, the processor 700 executes the program. The model training method or dispatching method disclosed in any of the foregoing embodiments of this application can be applied to the processor 700, or implemented by the processor 700.

[0191] The processor 700 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 700 or by instructions in software form. The processor 700 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 701. Processor 700 reads the information in memory 701 and, in conjunction with its hardware, completes the steps of the above method.

[0192] The electronic device provided in this application embodiment and the model training method or dispatching method provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they adopt, operate or implement.

[0193] This application also provides a computer-readable storage medium corresponding to the model training method or dispatching method provided in the foregoing embodiments. The computer-readable storage medium is an optical disc, on which a computer program (i.e., a computer program product) is stored. When the computer program is run by a processor, it executes the model training method or dispatching method provided in any of the foregoing embodiments.

[0194] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0195] The computer-readable storage medium provided in the above embodiments of this application is based on the same inventive concept as the model training method or dispatching method provided in the embodiments of this application, and has the same beneficial effects as the methods adopted, run or implemented by the application stored therein.

[0196] This application also provides a computer program product; please refer to [reference needed]. Figure 8 The computer program product 800 carries program code, namely computer program 801. The instructions included in the computer program 801 can be used to execute the steps of the model training method or the dispatching method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.

[0197] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0198] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0199] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0200] Additionally, as used herein, the “or” used in a list of items beginning with “at least one” indicates a separate list, such that a list of, for example, “at least one of A, B, or C” means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word “exemplary” does not imply that the described example is preferred or better than other examples.

[0201] It should also be noted that in the system and method of this application, the components or processes can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions of this application.

[0202] Various changes, substitutions, and modifications can be made to the technology described herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0203] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0204] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A model training method, characterized in that, include: Obtain manufacturing environment information for the target wafer, which is the wafer that needs to be manufactured according to the target process in the manufacturing process. The manufacturing environment information is used to indicate the current capacity of the machines that can perform the target process. Based on the manufacturing environment information and neural network model, a dispatching machine is determined from the target machines capable of performing the target process, wherein the dispatching machine is used to manufacture the target wafer according to the target process; In response to the completion of manufacturing of the target wafer, feedback parameters related to the production efficiency of the dispatching machine are obtained; The neural network model is updated based on the feedback parameters and the preset feedback parameter threshold.

2. The method according to claim 1, characterized in that, The manufacturing environment information includes the first queuing time between the target process and the next process after the target process, and the current capacity parameters of the target machine. Determining the dispatch machine from the target machine capable of performing the target process based on the manufacturing environment information and neural network model further includes: Based on the current production capacity parameters, the neural network model determines the manufacturing time required for the target machine to manufacture the target wafer according to the target process; Select at least one first machine from the target machines whose manufacturing time is less than or equal to the first queuing time; The dispatching machine is selected from the at least one first machine.

3. The method according to claim 1, characterized in that, The manufacturing environment information includes the current capacity parameters of the target machine, and determining the dispatch machine from the target machines capable of performing the target process based on the manufacturing environment information and a neural network model further includes: Based on the current production capacity parameters of the target machine, the production line load balance result of the target machine is determined through the neural network model; From the target machines, at least one second machine whose production line load balance result satisfies the first condition is selected; The dispatching machine is selected from the at least one second machine.

4. The method according to any one of claims 1-3, characterized in that, The manufacturing environment information includes the historical end time of manufacturing the target wafer according to the previous process of the target process and the second queuing time between the target process and the previous process. Determining the dispatch machine from the target machines capable of performing the target process based on the manufacturing environment information and a neural network model further includes: Based on the historical end time and the second queuing time, determine the start time for manufacturing the target wafer according to the target process; The neural network model is used to select at least one third machine from the target machines that is not at full load at the start time; The dispatching machine is selected from the at least one third machine.

5. The method according to claim 4, characterized in that, The manufacturing environment information also includes the number of wafer starts between the previous process and the target process, and further includes using the neural network model to screen at least one third machine from the target machines that is not at full capacity at the start time: Obtain the number of wafers currently being manufactured by the target machine; At least one third machine is selected from the target machines whose number of currently manufactured wafers is less than or equal to the quantity difference, wherein the quantity difference is the difference between the number of wafers fed and the number of target wafers.

6. The method according to claim 2, characterized in that, Selecting the dispatching machine from the at least one first machine further includes: Calculate the machine utilization rate of the at least one first machine; Among the at least one first machine, the machine whose utilization rate meets the second condition is selected as the dispatching machine.

7. The method according to any one of claims 1-3, characterized in that, The feedback parameters include at least one of the following: The actual queuing time for the target process; The actual utilization rate of the dispatching machine; or The actual production line balance rate of the dispatching machine.

8. A method for dispatching work, characterized in that, include: Obtain the current manufacturing environment information of the current wafer, which is the wafer that needs to be manufactured according to the current process in the current manufacturing process. The current manufacturing environment information is used to indicate the current capacity of the machine that can perform the current process. Using the neural network model trained by the model training method according to any one of claims 1-7 and the current manufacturing environment information, a current dispatching machine is determined from the machines capable of performing the current process, wherein the current dispatching machine is used to manufacture the current wafer according to the current process.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the model training method according to any one of claims 1-7 or the dispatching method according to claim 8.

Images