AI edge computing platform, method, terminal and system for Fab system

Abstract

The invention provides an AI edge computing platform, method, terminal and system for a Fab system, and the platform is specially designed for the Fab system, is customized and developed according to the business data processing task requirements in each Fab system, is deployed at the edge side of a plurality of machine equipment of a wafer manufacturing plant, and is in communication connection with each machine equipment. The platform can collect and screen process parameter data meeting high-quality wafer standards, perform analysis processing by using a specific algorithm model, dynamically adjust machine equipment process parameters and compensate Fab system indexes according to an analysis result, and optimize machine equipment performance, a production process and the algorithm model. By combining edge calculation and the AI technology, low-delay processing and real-time analysis of semiconductor manufacturing equipment process parameters are achieved, the system processing efficiency and response speed are improved, meanwhile, full-process automatic management of equipment data processing is integrated, the manual intervention requirement is reduced, and the workload of developers is relieved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to an AI edge computing platform, method, terminal and system for a Fab system. Background Art

[0002] With the development of big data and AI technology, the semiconductor industry has begun to enter the application stage of big data and AI technology, combining enterprise business systems and business data in multiple scenarios of AI applications to reduce costs and increase efficiency. With the increasing number of domestic wafer fabs, the efficiency of data processing after the Fab system goes online is required to be higher, and calculations must be performed and corresponding results obtained in the first time. At present, the Fab systems of most domestic wafer manufacturing plants upload massive amounts of data to cloud data centers for processing and return the results to users. Although the Fab system under the traditional cloud computing paradigm can meet the needs of uploading basic business data to cloud data centers for processing, with the continuous development and updating of individual delay-sensitive businesses such as artificial intelligence applications, it is difficult to meet the higher requirements for data volume, latency, and data controllability and security. Summary of the invention

[0003] In view of the shortcomings of the prior art mentioned above, the object of the present invention is to provide an AI edge computing platform, method, terminal and system for a Fab system, so as to solve the technical problem that the prior art is difficult to meet the higher requirements for data volume, latency, controllability and security of data as individual delay-sensitive businesses such as artificial intelligence applications are continuously developed and updated.

[0004] To achieve the above-mentioned purpose and other related purposes, the present invention provides an AI edge computing platform for a Fab system, which is developed based on the business data processing requirements in the Fab system, deployed on the edge side of multiple machine equipment in a wafer manufacturing plant, and communicated with each machine equipment. The platform includes: a machine data acquisition module, which is used to collect process parameter values ​​of each machine equipment according to user parameter requirements, and screen data belonging to a normal wafer production link and meeting high-quality wafer standards; a data analysis module, connected to the machine data acquisition module, and used to analyze and process the data output by the machine data acquisition module using an algorithm model with a Fab system business data processing function selected by the amount of data to be analyzed and processed by the data analysis module to obtain an analysis result; a data optimization module, connected to the data analysis module, and used to dynamically adjust the process parameter values ​​of each machine equipment and compensate for the Fab system index using the analysis result output by the data analysis module, so as to feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment, and use the compensated Fab system index to optimize the model parameters of the algorithm model in the data analysis module; and also used to adjust the type of algorithm model used by the data analysis module based on the amount of data to be processed by the data analysis module.

[0005] In one embodiment of the present invention, the machine data acquisition module includes: a user demand data acquisition unit, which is used to determine user parameter requirements and collect process parameter values ​​of each machine equipment based on the user parameter requirements; wherein the process parameter values ​​include: one or more of the temperature and humidity of the machine chamber, liquid flow rate and gas concentration; a normal wafer production link screening unit, connected to the user demand data acquisition unit, and used to screen the data of the normal wafer production link in the collected data to eliminate the data of the R&D and testing links; a high-quality data screening unit, connected to the normal wafer production link screening unit, and used to screen the data without process problems in the screened normal wafer production link data as data that meets the high quality standards; a data storage unit, connected to the high-quality data screening unit, and used to store the data that meets the high quality standards screened by the high-quality data screening unit.

[0006] In one embodiment of the present invention, the data analysis module is used to perform key parameter analysis on the data output by the machine data acquisition module using a machine learning model or a deep learning model with the business data processing function of each Fab system selected by the amount of data to be analyzed by the data analysis module to obtain corresponding analysis results; wherein, each Fab system includes: YMS system, APC system, FDC system, MES system and SPC system.

[0007] In one embodiment of the present invention, the data analysis module includes: a machine learning model prediction unit, which is used to predict the analysis results based on the data output of the machine data acquisition module through a trained machine learning algorithm model when the amount of data to be analyzed is within the standard data amount range; wherein the machine learning algorithm model adopts a decision tree model; a deep learning model prediction unit, which is used to predict the analysis results based on the data output of the machine data acquisition module through a trained deep learning algorithm model when the amount of data to be analyzed exceeds the standard data amount range.

[0008] In one embodiment of the present invention, the types of tree-buffering models adopted by the machine learning algorithm model include: CatBoost model and XGBoost model; the types of deep learning algorithm models include: CNN and DNN models.

[0009] In one embodiment of the present invention, the data optimization module includes: a parameter adjustment and compensation unit, which is used to dynamically adjust the process parameter values ​​and compensate the Fab system index using the analysis results, and feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment and use the Fab system index to optimize the model parameters of the algorithm model; a model adjustment unit, which is used to monitor the amount of data input by the data analysis module, and control the data analysis module to process the data using a machine learning model when the data amount is within the standard data amount range, and control the data analysis module to process the data using a deep learning algorithm model when the data amount exceeds the standard data amount range.

[0010] In one embodiment of the present invention, the development process of the platform includes: the user determines relevant parameters, and after the communication of the AI ​​edge computing platform and various tests in the traditional development model are completed, the AI ​​computing power required for the AI ​​edge computing platform is determined according to the type and scale of the business data processing tasks required in each Fab system, and the corresponding appropriate edge computing hardware specifications are configured, by deploying the computing, network and storage resources of the cloud server to the edge side of each machine device, and building an algorithm model on the edge server and deploying AI applications.

[0011] To achieve the above-mentioned purpose and other related purposes, the present invention provides an AI edge computing method for a Fab system, which is applied to an AI edge computing platform of the Fab system. The platform is developed based on the business data processing requirements of each Fab system, deployed on the edge side of multiple machine equipment in a wafer manufacturing plant, and communicated with each machine equipment. The method includes: collecting process parameter values ​​of each machine equipment according to user parameter requirements, and screening data belonging to a normal wafer production link and meeting high-quality wafer standards; using an algorithm model with a Fab system business data processing function selected by the amount of data to be analyzed and processed to analyze and process the screened data to obtain analysis results; using the analysis results to dynamically adjust the process parameter values ​​of each machine equipment and compensate for the Fab system index, so as to feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment, and using the compensated Fab system index to optimize the model parameters of the algorithm model.

[0012] To achieve the above-mentioned objectives and other related objectives, the present invention provides an electronic terminal, comprising: one or more memories and one or more processors; the one or more memories are used to store computer programs; the one or more processors are connected to the memories and are used to run the computer programs to execute the AI ​​edge computing method for the Fab system.

[0013] To achieve the above-mentioned purpose and other related purposes, the present invention provides an AI edge computing system, which includes: a cloud server and one or more AI edge computing subsystems; wherein each AI edge computing subsystem includes: a machine group, which is composed of multiple machine equipment of a wafer manufacturing plant; wherein each machine group is obtained by grouping the machine equipment of the wafer manufacturing plant; an AI edge computing platform for a Fab system deployed corresponding to the edge side of the machine group; the AI ​​edge computing platform for the Fab system is communicatively connected between each machine equipment of the corresponding machine group and the cloud server.

[0014] As described above, the present invention is an AI edge computing platform, method, terminal and system for a Fab system, which has the following beneficial effects: The present invention provides an AI edge computing platform designed specifically for a Fab system, which is customized and developed according to the business data processing task requirements in each Fab system, deployed on the edge side of multiple machines and equipment in a wafer manufacturing plant, and communicated with each machine and equipment. The platform can collect and screen process parameter data that meets high-quality wafer standards, analyze and process using a specific algorithm model, dynamically adjust the process parameters of the machine equipment according to the analysis results and compensate for the Fab system index, optimize the performance of the machine equipment and the production process and algorithm model. By combining edge computing with AI technology, the present invention realizes low-latency processing and real-time analysis of the process parameters of semiconductor manufacturing equipment, improves system processing efficiency and response speed, and integrates the full-process automated management of equipment data processing, reduces the need for manual intervention, and reduces the workload of developers. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Shown is a structural diagram of an AI edge computing platform for a Fab system in one embodiment of the present invention.

[0016] Figure 2 Shown is a structural diagram of an AI edge computing platform for a Fab system in one embodiment of the present invention.

[0017] Figure 3 Shown is a flow chart of an AI edge computing method for a Fab system in one embodiment of the present invention.

[0018] Figure 4 Shown is a schematic structural diagram of an electronic terminal in an embodiment of the present invention.

[0019] Figure 5 Shown is a structural diagram of an AI edge computing system in one embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0021] It should be noted that in the following description, reference is made to the accompanying drawings, which describe several embodiments of the present invention. It should be understood that other embodiments may also be used, and that mechanical composition, structure, electrical and operational changes may be made without departing from the spirit and scope of the present invention. The following detailed description should not be considered restrictive, and the scope of the embodiments of the present invention is limited only by the claims of the published patents. The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. Spatially related terms, such as "upper", "lower", "left", "right", "below", "below", "lower", "above", "upper", etc., may be used in the text to facilitate the description of the relationship between an element or feature shown in the figure and another element or feature.

[0022] Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case of "direct connection" but also the case of "indirect connection" by placing other elements therebetween. In addition, when a part is said to "include" a certain constituent element, unless otherwise stated, it does not exclude other constituent elements, but means that other constituent elements may be included.

[0023] The terms first, second and third mentioned herein are used to describe various parts, components, regions, layers and / or segments, but are not limited thereto. These terms are only used to distinguish a certain part, component, region, layer or segment from other parts, components, regions, layers or segments. Therefore, the first part, component, region, layer or segment described below may refer to the second part, component, region, layer or segment within the scope of the present invention.

[0024] Furthermore, as used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless there is an indication to the contrary in the context. It should be further understood that the terms "comprise", "include" indicate the presence of the described features, operations, elements, components, items, kinds, and / or groups, but do not exclude the presence, occurrence or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms "or" and "and / or" used herein are interpreted as inclusive, or mean any one or any combination. Therefore, "A, B or C" or "A, B and / or C" means "any of the following: A; B; C; A and B; A and C; B and C; A, B and C". Exceptions to this definition will only occur when the combination of elements, functions or operations is inherently mutually exclusive in some way.

[0025] Edge computing technology, with its low latency, low power consumption and enhanced privacy and security features, has shown significant advantages in the field of semiconductor manufacturing. It is especially suitable for latency-sensitive businesses, effectively alleviating network congestion and reducing the computing and storage burden of cloud data centers. By sinking cloud resources to the vicinity of terminal devices, edge servers take on more task processing, which not only reduces transmission latency but also reduces computing pressure on the cloud. However, with the increase in data scale and computing complexity, as well as the heterogeneity and limited nature of edge cluster resources, traditional edge computing faces challenges and is difficult to meet the low latency, low energy consumption and automated optimization requirements of equipment processes.

[0026] Therefore, the present invention provides an AI edge computing platform for the Fab system, which is specially designed for the Fab system, customized and developed according to the business data processing task requirements in each Fab system, deployed on the edge side of multiple machines and equipment in the wafer manufacturing plant, and communicated with each machine and equipment. The platform can collect and screen process parameter data that meets the high-quality wafer standards, analyze and process using a specific algorithm model, dynamically adjust the process parameters of the machine equipment according to the analysis results and compensate for the Fab system index, optimize the performance of the machine equipment and the production process and algorithm model. By combining edge computing with AI technology, the present invention realizes low-latency processing and real-time analysis of the process parameters of semiconductor manufacturing equipment, improves the system processing efficiency and response speed, and integrates the full-process automated management of equipment data processing, reduces the need for manual intervention, and reduces the workload of developers.

[0027] The following is a detailed description of the embodiments of the present invention with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention. The present invention can be embodied in many different forms and is not limited to the embodiments described herein.

[0028] like Figure 1 A structural schematic diagram of an AI edge computing platform for a Fab system in an embodiment of the present invention is shown.

[0029] During the wafer manufacturing process, the Fab system generates a large amount of business data, which is crucial for real-time monitoring of production status, optimizing production processes, predicting equipment failures, etc. Traditional business data processing methods often rely on cloud servers. With the continuous development and updating of individual latency-sensitive businesses such as artificial intelligence applications, higher requirements are placed on data volume, latency, and data controllability and security. Therefore, developing an AI-based edge computing platform has become an effective way to solve the computing performance problems of data-intensive tasks in the semiconductor manufacturing process.

[0030] The present invention is designed for the Fab system. The Fab system, or Fabrication system, is the abbreviation of a semiconductor manufacturing plant and its manufacturing process. The present invention aims to optimize equipment performance and production processes through artificial intelligence means to improve overall operational efficiency. The Fab systems include: EAP system, YMS system, APC system, FDC system, MES system, and SPC system. These systems each have different business responsibilities and generate a large amount of business data in the process.

[0031] The AI ​​edge computing platform is developed based on the business data processing requirements of each Fab system. The business data processing requirements of each Fab system include the business data processing function requirements of each Fab system. The deployment is on the edge side of multiple machine devices in the wafer manufacturing plant, that is, directly close to the data source. This deployment method can significantly reduce the delay of data transmission and improve the real-time performance of data processing. The platform is connected to each machine device in communication, and the platform includes: The main responsibility of the machine data acquisition module 1 is to accurately collect the process parameter values ​​of each machine equipment in the semiconductor manufacturing plant (Fab system) according to the parameter requirements set by the user. This module is not only responsible for data collection, but also undertakes the important task of data screening. In terms of data collection, the machine data acquisition module 1 captures key process parameters such as temperature, pressure, flow, time, etc. from each machine equipment in real time and accurately according to the user's parameter requirements. These parameters are crucial to understanding the status of the production process and product quality. The machine data acquisition module 1 also needs to strictly screen and filter the collected data. It only screens data that belongs to the normal wafer production link and meets the high-quality wafer standards to ensure the accuracy and effectiveness of subsequent analysis and optimization work. This screening mechanism helps to reduce the interference of noise data and improve the efficiency and accuracy of data processing.

[0032] The data analysis module 2 is connected to the machine data acquisition module 1 and has multiple AI algorithm models with Fab system business data processing task functions built in. The data analysis module 2 uses the AI ​​algorithm model selected by the amount of data that the data analysis module needs to analyze and process to analyze and process the data output by the machine data acquisition module 1 to obtain the analysis results; The data optimization module 3 is connected to the data analysis module 2. The data optimization module 3 uses the analysis results to dynamically adjust the process parameter values ​​of each machine equipment to feed back the adjusted process parameter values ​​to the user. This process is intended to optimize the performance of the machine equipment, improve production efficiency, and ensure that the produced wafers meet high quality standards. By adjusting the process parameters in real time, the data optimization module 3 helps to reduce fluctuations and uncertainties in the production process and improve the stability and controllability of the overall production line. In addition to adjusting the process parameters, the data optimization module 3 can also use the analysis results to compensate the Fab system index to optimize the model parameters of the algorithm model in the data analysis module. This step is to reflect the actual situation in the production process and ensure that the algorithm model in the data analysis module 2 can more accurately reflect the status of the production line. By compensating the Fab system index, it helps to improve the prediction ability and accuracy of the algorithm model, that is, by continuously adjusting and optimizing the model parameters, the data optimization module 3 can ensure that the algorithm model can more accurately predict and explain the data changes in the production process, and improve the efficiency and accuracy of data analysis and optimization. Provide a more reliable basis for subsequent data analysis and optimization. In addition, the data optimization module 3 can also intelligently adjust the type of algorithm model used according to the amount of data to be processed by the data analysis module 2. This feature helps ensure that the data analysis module 2 can maintain efficient and accurate performance at different data scales. By dynamically selecting and optimizing algorithm models, the data optimization module 3 helps improve the flexibility and adaptability of the entire AI edge computing platform.

[0033] The present invention integrates edge computing and artificial intelligence (AI) platforms to optimize the computing performance of data-intensive tasks in edge environments. By intelligently assigning tasks to appropriate AI edge computing platforms and processing edge computing data, low-latency, low-energy consumption, and low-cost transmission of Fab system data is achieved. This solution not only supports low-latency transmission interaction of equipment process parameters, but also uses the model of the AI ​​edge computing platform to analyze key parameters, optimize the process, and provide real-time feedback to users. This innovation has promoted the industrial digitalization and intelligentization of the Fab system, significantly reduced data transmission costs, and provided strong support for the transformation and upgrading of the semiconductor manufacturing industry.

[0034] In one embodiment, if Figure 2 , the machine data acquisition module 1 includes: The user demand data acquisition unit 101 is used to determine the user parameter requirements and collect the process parameter values ​​of each machine equipment based on the user parameter requirements; wherein the process parameter values ​​include: one or more of the temperature and humidity of the machine chamber, the liquid flow rate, and the gas concentration; these parameters are crucial to the wafer manufacturing process because they directly affect the quality and production efficiency of the product. The user demand data acquisition unit 101 can realize the relevant business functions of the EAP system of the Fab system. The EAP system is a system specially designed to realize real-time monitoring and automatic control of the machines in the Fab system.

[0035] The normal wafer production link screening unit 102 is connected to the user demand data collection unit 101, and is used to use an effective data cleaning method to screen the data of the normal wafer production link in the collected data to eliminate the data of the R&D test link; for example, the data of the normal wafer production link is screened based on the data characteristics of the normal wafer production link and the R&D test link. Specifically, the data of the normal wafer production link can be screened according to the production plan and test plan time, the data related identifiers of the production equipment and the test equipment, and the format of the production data and the R&D test data.

[0036] The high-quality data screening unit 103 is connected to the normal wafer production link screening unit 102, and is used to use a preprocessing method to screen the data without process problems in the screened normal wafer production link data as data that meets the high-quality standards; for example, the data with process problems are screened by identifying failure factors. The failure factor is a key factor that can cause process problems in the wafer production link. For example, process problems that are close to scrap status and have large fluctuations can be judged based on defect density, defect size, defect type, etc.

[0037] The data storage unit 104 is connected to the high-quality data screening unit 103 and is used to store data that meets the high-quality standard and is screened by the high-quality data screening unit.

[0038] In one embodiment, the data analysis module 2 is an advanced artificial intelligence processing center integrated in the platform, which deploys high-performance computing resources and advanced algorithm models, and can receive and process the data of each Fab system in real time. Through deep learning and machine learning technology, it analyzes and processes data and performs regression prediction, and feeds back optimized solutions to different users, helping users to optimize equipment performance and adjust production plans in real time, and comprehensively manage production data.

[0039] The data analysis module 2 uses a machine learning model or a deep learning model with the business data processing function of each Fab system to analyze the key parameters of the data output by the machine data acquisition module to obtain the corresponding analysis results. For example, the key parameters are key parameters such as production progress, equipment status, material consumption, etc. The data analysis module 2 uses a model selected according to the amount of data to be analyzed to process the data. The Fab systems mentioned here include the YMS system, APC system, FDC system, MES system and SPC system, and each system plays an important role in semiconductor production.

[0040] YMS system: Yield Management System is a system that integrates data management, data analysis and professional tools. It is mainly used in semiconductor manufacturing, packaging testing and other processes, especially in the mass production stage, to help engineers greatly improve data analysis efficiency, quickly analyze one or more types of data, and find the key points to improve yield, thereby promoting the stability and controllability of the entire production process, improving product quality and reducing enterprise costs.

[0041] APC system: Advanced Process Control, which uses mathematical models and real-time data to accurately control the production process to improve product quality and production efficiency.

[0042] FDC system: Fault Detection and Classification system, used to monitor equipment status in real time and diagnose faults.

[0043] MES system: Manufacturing Execution System is an indispensable part of semiconductor production. It is responsible for converting production plans into specific production operations and monitoring production progress and equipment status in real time.

[0044] SPC system: Statistical Process Control (SPC) is a method of monitoring and controlling product quality by collecting and analyzing data from the production process. Based on statistical principles, it helps companies to detect abnormal fluctuations in the production process in real time by monitoring and predicting key parameters in the production process, so that they can take corresponding measures to adjust and optimize.

[0045] In one embodiment, if Figure 2 , the data analysis module 2 includes: The machine learning model prediction unit 201 is used when the processing data is within the standard data volume range. The trained decision tree model is used to output accurate prediction analysis results based on the data output by the machine data acquisition module. The decision tree model displays the decision process through a tree structure, where each internal node represents an attribute test, each branch represents a test result, and the leaf node corresponds to the final prediction result. This model has a clear structure, is easy to understand, and performs well when processing structured data.

[0046] The deep learning model prediction unit 202, when the amount of processed data exceeds the standard range, uses the trained deep learning algorithm model, also based on the data of the machine data acquisition module, to perform prediction analysis and output the predicted analysis results. Deep learning algorithm models have significant advantages in processing large-scale, high-dimensional data. They can automatically extract complex features in the data and achieve high-precision prediction through multi-layer nonlinear transformations.

[0047] By integrating the machine learning model prediction unit and the deep learning model prediction unit, the data analysis module can flexibly respond to data processing needs of different scales and provide strong support for key parameter analysis in the wafer production process.

[0048] In one embodiment, the types of tree-buffering models used by the machine learning algorithm model include: CatBoost model and XGBoost model; CatBoost is a decision tree algorithm based on gradient boosting. It performs well in dealing with classification and regression problems, and is particularly suitable for preliminary data analysis and model building. It can automatically process categorical features and reduce the risk of overfitting through a unique algorithm. CatBoost introduces a symmetrical tree structure to make the model more stable during training and improve the accuracy of predictions. In the wafer production process, the CatBoost model can be used to predict key indicators such as equipment failure rate and product yield. Because it can automatically process categorical features, it is very suitable for dealing with complex scenarios containing multiple types of data.

[0049] XGBoost is another powerful gradient boosting algorithm that performs well when processing large-scale data and can handle missing values ​​and categorical features. XGBoost improves the generalization ability of the model by optimizing the loss function and regularization term, while supporting parallel computing, which greatly improves the training speed. The XGBoost model can be used for a variety of prediction tasks in wafer production, such as predicting the production cycle of products, the maintenance cycle of equipment, etc. Due to its fast training speed and high accuracy, it is very suitable for processing large-scale data sets.

[0050] The deep learning algorithm model adopts a CNN or DNN model; CNN is a convolutional neural network model, and its network structure has excellent performance in image processing. Through structures such as convolutional layers, pooling layers, and fully connected layers, CNN can automatically extract features and perform tasks such as classification or regression. CNN has a wide range of applications in image processing, video analysis, and other fields. In the wafer production process, the amount of data that needs to be analyzed is usually large, and CNN can quickly process this data and output prediction results. This helps to detect problems in a timely manner and take appropriate measures, thereby improving production efficiency and product quality.

[0051] DNN is a computational model that simulates the structure and function of the human brain neural network, with powerful feature learning and nonlinear processing capabilities. Through multiple layers of nonlinear hidden layers, it can approximate complex functions and achieve a universal approximation effect. DNN can automatically extract useful features from raw data, greatly improving the generalization ability of the model. In the wafer production process, DNN can be used for advanced tasks such as predicting product performance and optimizing production parameters.

[0052] In one embodiment, if Figure 2 , the data optimization module 3 includes: Parameter adjustment compensation unit 301, based on the analysis results, can adjust the key process parameters in the wafer production process, such as temperature, pressure, time, etc., in real time and accurately. The adjusted process parameter values ​​will be fed back to the user in real time, so that the user can understand the production status in time and make further optimization and adjustments as needed. This helps to improve the transparency and efficiency of production management. In addition, the model parameters of the Fab system indexation algorithm model are compensated based on the analysis results. Through this optimization, the accuracy and reliability of the algorithm model can be further improved, thereby ensuring that the process parameter adjustment in the production process is more accurate and effective.

[0053] The parameter adjustment and compensation unit 301 can quickly adjust the process parameter values ​​based on the real-time analysis results. This immediate response capability helps to prevent potential problems before they occur, thereby reducing the triggering of OCap alarms. The unit not only adjusts the process parameters, but also compensates the Fab system index. For example, in the APC system, the compensated system index is the index in front of each input process parameter in the algorithm equation (which can be understood as a slope, and its size range is 0~1). This precise compensation can ensure that the production environment operates in the best state and reduce OCap alarms caused by system fluctuations or instability. The unit 301 can ensure that the equipment maintains a stable operating state during the production process by real-time monitoring and adjusting the process parameter values ​​and compensating the Fab system index. This helps to reduce equipment failures and downtime, and improve production efficiency and product quality. In addition to ensuring the stability of equipment operation, the unit 301 can also further improve the stability of equipment operation by continuously optimizing the algorithm model and process parameter values. This helps to reduce volatility and uncertainty in the production process and improve the controllability and predictability of production. The wafer production process may be affected by a variety of external factors, such as temperature changes, humidity fluctuations, power supply fluctuations, etc. The unit 301 can offset the impact of these external factors on the equipment to a certain extent by dynamically adjusting the process parameter values ​​and compensating the Fab system index, thereby ensuring the stability and controllability of the production process.

[0054] The model adjustment unit 302 is used to monitor the amount of data input by the data analysis module, and control the data analysis module 2 to process the data using the machine learning model when the data amount is within the standard data amount range, and control the data analysis module 2 to process the data using the deep learning algorithm model when the data amount exceeds the standard data amount range; The model adjustment unit 302 can make intelligent decisions based on the real-time changes in the data volume and select the appropriate algorithm model for processing. This decision-making method can ensure the efficiency and accuracy of data processing. The unit 302 can flexibly adjust the standard data volume range according to the changes in system processing capabilities and business needs. This highly adaptive feature enables the system to cope with data processing tasks of different scales and complexities.

[0055] In one embodiment, the development process of the platform includes: Users determine the relevant parameters of the AI ​​edge computing platform based on their own needs. These parameters may include data processing capabilities, real-time requirements, security requirements, etc. Based on user needs, platform developers set corresponding parameters in the system to lay the foundation for subsequent system development and configuration.

[0056] Ensure that the AI ​​edge computing platform can communicate stably and efficiently. Before introducing AI technology, the system is tested through traditional development models, namely local simulation environment testing, actual simulation environment testing, and actual production environment testing to ensure the stability and reliability of basic functions.

[0057] Analyze the required AI computing power requirements based on the type (such as image recognition, data analysis, prediction, etc.) and scale (such as data volume, processing speed, etc.) of business data processing tasks in each Fab system.

[0058] According to the AI ​​computing power requirements, select appropriate edge computing hardware specifications, including servers, storage devices, network devices, etc. Based on the hardware selection results, configure the specification parameters of the edge computing hardware, such as processor model, memory capacity, storage type, etc.

[0059] Deploy the computing, network and storage resources of the cloud server to the edge side of each machine device to obtain the corresponding edge server, realize the localization of data, that is, store and process data in the edge area, and reduce network transmission delay.

[0060] Build an algorithm model suitable for the Fab system on the edge server and deploy AI applications to provide real-time data processing and optimization services on the edge side. Specifically, the machine data acquisition module collects process parameter values ​​for processing business data tasks of each Fab system from each machine device according to user parameter requirements, and selects data that belongs to the normal wafer production link and meets the high-quality wafer standards; the data is initially selected for training and testing in the machine learning model. As the amount of data increases (for example, when the amount of data exceeds 10,000), consider adjusting the model and using deep learning algorithms instead. These deep learning algorithms have stronger expression and generalization capabilities when processing large-scale, high-dimensional data. Based on in-depth data mining and analysis, we built a system model that can accurately reflect the complex relationship between equipment process and parameters. Through this model, we optimized the equipment process and parameters, significantly improving the equipment process capability and product yield. At the same time, we also optimized the system model itself, enhancing its accuracy and robustness to ensure that it can better adapt to the actual production environment. Finally, we successfully built an algorithm model designed for the Fab system on the edge server and deployed AI applications. The deployed edge servers can receive and process data from various Fab systems in real time, and use AI algorithms to achieve real-time analysis and optimization, thereby greatly improving the real-time and accuracy of data processing.

[0061] The entire development process is centered on user needs to ensure that the AI ​​edge computing platform can meet the actual needs of users. By deploying cloud server resources to the edge side, data transmission latency is reduced and data processing efficiency is improved. According to the characteristics and scale of business data processing tasks, edge computing hardware specifications and cloud server resources are flexibly configured to achieve optimal resource allocation. With the development of business and the advancement of technology, the AI ​​edge computing platform can be easily expanded and upgraded to meet higher performance requirements.

[0062] Similar to the principle of the above embodiment, the present invention provides an AI edge computing method for a Fab system.

[0063] The following provides specific embodiments in conjunction with the accompanying drawings: like Figure 3 A flowchart of an AI edge computing method for a Fab system in an embodiment of the present invention is shown.

[0064] The AI ​​edge computing platform applied to the Fab system is developed based on the business data processing requirements of each Fab system, deployed on the edge side of multiple machine devices in the wafer manufacturing plant, and communicated with each machine device. The method includes: Step S1: Collect the process parameter values ​​of each machine equipment according to the user's parameter requirements, and select the data that belongs to the normal wafer production process and meets the high-quality wafer standards.

[0065] In one embodiment, step S1 includes: Determine user parameter requirements, and collect process parameter values ​​of each machine device based on the user parameter requirements; wherein the process parameter values ​​include: one or more of the temperature and humidity of the machine cavity, liquid flow rate, and gas concentration; Filter the data collected from the normal wafer production process to eliminate the data from the R&D testing process; The data without process problems are selected from the data of normal wafer production process as data meeting high quality standards; The data that meets the high quality standard screened by the high quality data screening unit is stored.

[0066] Step S2: Analyze and process the screened data using an algorithm model with Fab system business data processing function selected according to the amount of data to be analyzed and processed to obtain analysis results.

[0067] In one embodiment, step S2 includes: A machine learning model or deep learning model with the business data processing function of each Fab system selected by the data volume to be analyzed by the data analysis module is used to perform key parameter analysis on the data output by the machine data acquisition module to obtain corresponding analysis results; wherein, each Fab system includes: YMS system, APC system, FDC system, MES system and SPC system.

[0068] In one embodiment, the selected data is analyzed and processed using an algorithm model having a Fab system business data processing function selected according to the amount of data to be analyzed and processed, and the analysis results obtained include: When the amount of data to be analyzed is within the standard data amount range, the trained machine learning algorithm model outputs the predicted analysis results according to the data output by the machine data acquisition module; wherein the machine learning algorithm model adopts a decision tree model; When the amount of data to be analyzed exceeds the standard data amount range, the trained deep learning algorithm model outputs the predicted analysis results based on the data output by the machine data acquisition module.

[0069] In one embodiment, the types of tree-buffering models adopted by the machine learning algorithm model include: CatBoost model and XGBoost model; the types of deep learning algorithm models include: CNN or DNN model.

[0070] Step S3: dynamically adjust the process parameter values ​​of each machine equipment and compensate the Fab system index using the analysis results, so as to feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment, and use the compensated Fab system index to optimize the model parameters of the algorithm model.

[0071] In one embodiment, step S3 includes: Dynamically adjust the process parameter values ​​and compensate the Fab system index using the analysis results, and feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment and use the Fab system index to optimize the model parameters of the algorithm model; Monitor the amount of data that needs to be processed, and control the use of machine learning models to process the data when the data amount is within the standard data amount range, and control the use of deep learning algorithm models to process the data when the data amount exceeds the standard data amount range.

[0072] In one embodiment, the development process of the AI ​​edge computing platform for the Fab system includes: the user determines relevant parameters, and after the communication of the AI ​​edge computing platform and various tests in the traditional development model are completed, the AI ​​computing power required for the AI ​​edge computing platform is determined according to the type and scale of the business data processing tasks required in each Fab system, and the corresponding appropriate edge computing hardware specifications are configured, by deploying the computing, network and storage resources of the cloud server to the edge side of each machine device, and building an algorithm model on the edge server and deploying AI applications.

[0073] Since the implementation principle of the AI ​​edge computing method for the Fab system has been described in the previous embodiments, it will not be repeated here.

[0074] The AI ​​edge computing platform for the Fab system provided in the embodiment of the present invention can be implemented on the terminal side or the server side. As for the hardware structure of the electronic terminal, please refer to Figure 4 , is an optional hardware structure diagram of an AI edge computing terminal 1000 for a Fab system provided in an embodiment of the present invention. The terminal 1000 may be a mobile phone, a computer device, a tablet device, a personal digital processing device, a factory background processing device, etc. The terminal 1000 includes: at least one processor 1001, a memory 1002, at least one network interface 10010 and a user interface 1009. The various components in the device are coupled together through a bus system 1005. It can be understood that the bus system 1005 is used to realize connection and communication between these components. In addition to the data bus, the bus system 1005 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, in Figure 4 In the specification, various buses are labeled as bus systems.

[0075] The user interface 1009 may include a display, a keyboard, a mouse, a trackball, a click gun, keys, buttons, a touch pad or a touch screen.

[0076] It is understood that the memory 1002 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), which is used as an external cache. By way of exemplary but not limiting explanation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM). The memory described in the embodiments of the present invention is intended to include but is not limited to these and any other suitable categories of memory.

[0077] The memory 1002 in the embodiment of the present invention is used to store various categories of data to support the operation of the terminal 1000. Examples of these data include: any executable program for operating on the terminal 1000, such as an operating system 10021 and an application 10022; the operating system 10021 includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application 10022 may include various applications, such as a media player (MediaPlayer), a browser (Browser), etc., for implementing various application services. The AI ​​edge computing platform for the Fab system provided by the embodiment of the present invention may be included in the application 10022.

[0078] The method disclosed in the above embodiment of the present invention can be applied to the processor 1001, or implemented by the processor 1001. The processor 1001 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 1001 or the instruction in the form of software. The above processor 1001 may be a general processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 1001 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present invention. The general processor 1001 can be a microprocessor or any conventional processor, etc. In combination with the steps of the accessory optimization method provided in the embodiment of the present invention, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0079] In an exemplary embodiment, the terminal 1000 may be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD) to execute the aforementioned method.

[0080] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0081] In the embodiments provided in the present application, the computer readable and writable storage medium may include a read-only memory, a random access memory, an EEPROM, a CD-ROM or other optical disk storage device, a disk storage device or other magnetic storage device, a flash memory, a USB flash drive, a mobile hard disk, or any other medium that can be used to store the desired program code in the form of an instruction or data structure and can be accessed by a computer. In addition, any connection can be appropriately referred to as a computer-readable medium. For example, if the instruction is sent from a website, a server or other remote source using a coaxial cable, an optical fiber cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, optical fiber cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. However, it should be understood that computer readable and writable storage media and data storage media do not include connections, carriers, signals, or other temporary media, but are intended to be non-temporary, tangible storage media. Disk and disc, as used in this application, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers.

[0082] like Figure 5 A structural schematic diagram of an AI edge computing system in an embodiment of the present invention is shown.

[0083] The AI ​​edge computing system is composed of a cloud server and one or more AI edge computing subsystems. Each AI edge computing subsystem is further subdivided into a machine group and a corresponding AI edge computing platform. The machine group consists of multiple machine devices that are geographically close or functionally related in a wafer manufacturing plant, and the AI ​​edge computing platform is deployed near these machine devices to realize the edge side of each machine device deployed in this machine group, and the communication connection between each machine device of the corresponding machine group and the cloud server serves as a communication bridge between each machine device and the cloud server.

[0084] Among them, each AI edge computing subsystem includes: For each AI edge computing subsystem, the AI ​​edge computing platform of each AI edge computing subsystem is responsible for data collection and analysis of each machine equipment in the machine group within the system; the AI ​​edge computing platform collects the process parameter values ​​of each machine equipment in the machine group within the system according to user parameter requirements, and screens the data that belongs to the normal wafer production link and meets the high-quality wafer standards; then uses the algorithm model with Fab system business data processing function selected by the amount of data to be analyzed and processed to analyze and process the screened data to obtain analysis results; uses the analysis results to dynamically adjust the process parameter values ​​of each machine equipment and compensate the Fab system index, so as to feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment, and uses the compensated Fab system index to optimize the model parameters of the algorithm model.

[0085] Each AI edge computing platform used in the Fab system can transmit the platform's data to the cloud server as needed, allowing the cloud server to perform backup or other analysis and processing.

[0086] Through distributed deployment, the AI ​​edge computing platform of this system can process the data generated by the machine equipment nearby without long-distance transmission to the cloud, thereby significantly reducing the delay of data transmission and ensuring the response speed of application scenarios with high real-time requirements. And because the need to transmit large amounts of raw data to the cloud server is reduced, the data transmission cost and network load are reduced. In addition, for the situation where there are many machines and equipment in the wafer manufacturing plant and the amount of data is large, through the distributed deployment platform, multiple AI edge computing platforms can process data from different machines and equipment at the same time, realize parallel computing, and improve the overall data processing speed. By reasonably allocating tasks to different edge computing platforms, it is possible to avoid overloading a single platform, achieve load balancing, and improve system stability and reliability. The edge computing platform can also dynamically adjust computing resources according to actual needs, optimize resource utilization, and reduce energy consumption and operating costs.

[0087] Compared with the prior art, the present invention has the following advantages: 1. Through the deep integration of edge computing and AI technology, the present invention realizes real-time analysis and optimization of process parameters of semiconductor manufacturing equipment, significantly improving system processing efficiency and response speed. The AI ​​model intelligently identifies and processes multi-type equipment data, quickly performs low-latency optimization on edge devices, greatly reduces cloud transmission costs, and enhances equipment stability and production efficiency. At the same time, using AI edge computing technology, the present invention performs low-latency processing and real-time analysis of equipment data, intelligently parses and optimizes data, automatically extracts machine data and standardizes it to the edge system, thereby optimizing equipment processes, achieving efficient system operation, and further improving the level of factory automation.

[0088] 2. The present invention integrates the full process management of equipment data processing, and realizes automated operation from data collection, analysis to optimization feedback, which greatly reduces the need for manual intervention and reduces the workload of developers. Developers only need to write a small amount of code and adjust the model to quickly complete the deployment of the system. This standardized processing method is widely applicable to various types of semiconductor equipment, and can help various types of wafer fabs achieve process optimization and promote production lines to achieve efficient operation and intelligent management in a short time.

[0089] 3. The present invention effectively monitors the process and equipment operation status through the positive supply and feedback of data, optimizes the equipment process parameters in real time, and makes dynamic adjustments according to production needs to improve product yield. The system integrates functional modules such as data processing, model reasoning, and decision feedback through the edge computing platform to ensure the efficiency and accuracy of equipment operation, especially when dealing with complex production scenarios, showing strong adaptability.

[0090] 4. The present invention has strong integration capabilities, covering design, development, testing and other resources, and realizes intelligent management of the entire process from data collection to optimization feedback. The resources integrated by this technology ensure the low-latency response and high-performance data processing capabilities of the system, and significantly reduce the cost of data transmission to the cloud during the industrial digital transformation of the Fab system.

[0091] In summary, the AI ​​edge computing platform, method, terminal and system for the Fab system of the present invention are specially designed for the Fab system, customized and developed according to the business data processing task requirements in each Fab system, deployed on the edge side of multiple machine equipment in the wafer manufacturing plant, and communicated with each machine equipment. The platform can collect and screen process parameter data that meets the high-quality wafer standards, analyze and process using a specific algorithm model, dynamically adjust the process parameters of the machine equipment according to the analysis results and compensate for the Fab system index, optimize the performance of the machine equipment and the production process and algorithm model. By combining edge computing with AI technology, the present invention realizes low-latency processing and real-time analysis of the process parameters of semiconductor manufacturing equipment, improves the system processing efficiency and response speed, and integrates the full-process automated management of equipment data processing, reduces the need for manual intervention, and reduces the workload of developers. Therefore, the present invention effectively overcomes the various shortcomings in the prior art and has a high industrial utilization value.

[0092] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. An AI edge computing platform for Fab system, characterized in that: Developed based on the business data processing requirements of each Fab system, deployed on the edge of multiple machines in the wafer manufacturing plant, and connected to each machine, the platform includes: The machine data acquisition module is used to collect the process parameter values ​​of each machine equipment according to the user's parameter requirements, and screen the data that belongs to the normal wafer production link and meets the high-quality wafer standards; A data analysis module, connected to the machine data acquisition module, used to analyze and process the data output by the machine data acquisition module using an algorithm model with Fab system business data processing function selected according to the amount of data to be analyzed and processed by the data analysis module to obtain an analysis result; A data optimization module is connected to the data analysis module and is used to dynamically adjust the process parameter values ​​of each machine equipment and compensate the Fab system index using the analysis results output by the data analysis module, so as to feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment, and use the compensated Fab system index to optimize the model parameters of the algorithm model in the data analysis module; it is also used to adjust the type of algorithm model used by the data analysis module based on the amount of data that the data analysis module needs to process.

2. The AI ​​edge computing platform for Fab system according to claim 1, characterized in that: The machine data acquisition module includes: A user demand data collection unit is used to determine the user parameter requirements and collect the process parameter values ​​of each machine equipment based on the user parameter requirements; wherein the process parameter values ​​include: one or more of the temperature and humidity of the machine cavity, the liquid flow rate and the gas concentration; A normal wafer production link screening unit, connected to the user demand data collection unit, is used to screen the data of the normal wafer production link in the collected data to exclude the data of the R&D test link; A high-quality data screening unit, connected to the normal wafer production link screening unit, is used to screen the data without process problems from the screened normal wafer production link data as data that meets the high-quality standard; The data storage unit is connected to the high-quality data screening unit and is used to store the data that meets the high-quality standard and is screened by the high-quality data screening unit.

3. The AI ​​edge computing platform for Fab system according to claim 1, characterized in that: The data analysis module is used to perform key parameter analysis on the data output by the machine data acquisition module by using a machine learning model or deep learning model with the business data processing function of each Fab system selected according to the amount of data to be analyzed by the data analysis module to obtain corresponding analysis results; wherein, each Fab system includes: YMS system, APC system, FDC system, MES system and SPC system.

4. The AI ​​edge computing platform for Fab system according to claim 3, characterized in that: The data analysis module includes: A machine learning model prediction unit, used to output a predicted analysis result based on the data output by the machine data acquisition module through a trained machine learning algorithm model when the amount of data to be analyzed is within the standard data amount range; wherein the machine learning algorithm model adopts a decision tree model; The deep learning model prediction unit is used to output the predicted analysis results based on the data output by the machine data acquisition module through the trained deep learning algorithm model when the amount of data to be analyzed exceeds the standard data amount range.

5. The AI ​​edge computing platform for Fab system according to claim 4, characterized in that: The types of tree-buffering models used by the machine learning algorithm model include: CatBoost model and XGBoost model; the types of deep learning algorithm models include: CNN and DNN models.

6. The AI ​​edge computing platform for Fab system according to claim 4, characterized in that: The data optimization module includes: A parameter adjustment and compensation unit, used to dynamically adjust the process parameter values ​​and compensate the Fab system index using the analysis results, and feed back the adjusted process parameter values ​​to the user to optimize the performance and production process of the machine equipment and optimize the model parameters of the algorithm model using the Fab system index; The model adjustment unit is used to monitor the amount of data input by the data analysis module, and control the data analysis module to process the data using a machine learning model when the data amount is within a standard data amount range, and control the data analysis module to process the data using a deep learning algorithm model when the data amount exceeds the standard data amount range.

7. The AI ​​edge computing platform for Fab system according to claim 1, characterized in that: The development process of the platform includes: the user determines the relevant parameters, and after the communication of the AI ​​edge computing platform and various tests in the traditional development model are completed, the AI ​​computing power required for the AI ​​edge computing platform is determined according to the type and scale of the business data processing tasks required in each Fab system, and the corresponding appropriate edge computing hardware specifications are configured, by deploying the computing, network and storage resources of the cloud server to the edge side of each machine device, and building an algorithm model on the edge server and deploying AI applications.

8. An AI edge computing method for a Fab system, characterized in that: The AI ​​edge computing platform applied to the Fab system is developed based on the business data processing requirements of each Fab system, deployed on the edge side of multiple machine devices in the wafer manufacturing plant, and communicated with each machine device. The method includes: Collect the process parameter values ​​of each machine equipment according to the user's parameter requirements, and select the data that belongs to the normal wafer production process and meets the high-quality wafer standards; Analyze and process the screened data using an algorithm model with Fab system business data processing function selected according to the amount of data to be analyzed and processed to obtain analysis results; The analysis results are used to dynamically adjust the process parameter values ​​of each machine equipment and compensate the Fab system index, so that the adjusted process parameter values ​​are fed back to the user to optimize the performance and production process of the machine equipment, and the compensated Fab system index is used to optimize the model parameters of the algorithm model.

9. An electronic terminal, characterized in that: include: one or more memories and one or more processors; The one or more memories are used to store computer programs; The one or more processors, connected to the memory, are configured to run the computer program to perform the method according to claim 8.

10. An AI edge computing system, characterized in that: The system includes: a cloud server and one or more AI edge computing subsystems; Among them, each AI edge computing subsystem includes: A tool group, consisting of a plurality of tool equipments in a wafer manufacturing plant; wherein each tool group is obtained by grouping tool equipments in a wafer manufacturing plant; An AI edge computing platform for the Fab system as described in claims 1 to 7 is deployed correspondingly on the edge side of the machine group; the AI ​​edge computing platform for the Fab system is communicatively connected between each machine device of the corresponding machine group and the cloud server.

Images