Equipment maintenance system applied to generic semiconductor field

Abstract

The invention discloses an equipment maintenance system applied to the field of generic semiconductors, and belongs to the technical field of computers. The system comprises a data module used for obtaining target operation data of target equipment; the large model module is used for acquiring a fault diagnosis result of the target equipment according to the target operation data on the basis of a large language model; and the maintenance early warning module is used for processing the target operation data based on a residual service life prediction model, an overall equipment efficiency improvement model and a health assessment model, and generating an equipment optimization suggestion for the target equipment based on a processing result and the fault diagnosis result. According to the equipment maintenance system applied to the generic semiconductor field disclosed by the invention, the difficulty of carrying out optimization operation on the production equipment is reduced.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of computers, and particularly relates to a device maintenance system applied to the field of general semiconductors. BACKGROUND

[0002] In the field of general semiconductors, such as the field of integrated circuits, solar cells, and semiconductor device materials, the production efficiency of production equipment is generally improved by using a predictive method for overall equipment efficiency and residual service life.

[0003] However, in general, the traditional predictive method lacks explanation in improving the production efficiency of the production equipment. After obtaining the prediction result, the user has difficulty in understanding the reasons and basis behind the prediction result, which leads to an increase in the difficulty of optimizing the operation of the production equipment based on the prediction result. SUMMARY

[0004] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application provides a device maintenance system applied to the field of general semiconductors to reduce the difficulty of improving the operation of the production equipment.

[0005] In a first aspect, the present application provides a device maintenance system applied to the field of general semiconductors, which comprises:

[0006] a data module configured to obtain target operation data of a target device;

[0007] a large model module configured to obtain a fault diagnosis result of the target device based on a large language model and the target operation data;

[0008] a maintenance warning module configured to process the target operation data based on a residual service life prediction model, an overall equipment efficiency improvement model, and a health assessment model, and generate a device optimization suggestion for the target device based on the processed result and the fault diagnosis result.

[0009] According to the device maintenance system applied to the field of general semiconductors of the present application, the maintenance warning module processes the target operation data based on the residual service life prediction model, the overall equipment efficiency improvement model, and the health assessment model, and generates a device optimization suggestion for the target device based on the processed result and the fault diagnosis result generated by the large model module based on the large language model, so as to provide the reasons and basis for device optimization, thereby reducing the difficulty of optimizing the operation of the target device based on the device optimization suggestion.

[0010] According to an embodiment of the present application, the system further comprises a model management module configured to perform performance evaluation on the residual service life prediction model, the overall equipment efficiency improvement model, the health assessment model, and the large language model, and generate a model evaluation report.

[0011] According to an embodiment of the present application, the system further comprises a knowledge base module, the knowledge base module comprising an enhanced knowledge graph in the field of general semiconductors;

[0012] The large model module is specifically configured to, in the case that the target question input by the user is acquired, based on the large language model and the target question, call the knowledge graph to perform thought chain-based retrieval, and acquire the fault diagnosis result.

[0013] According to an embodiment of the present application, the knowledge base module comprises an updating unit; the updating unit is configured to periodically update the knowledge graph and the target response format.

[0014] According to an embodiment of the present application, the large model module comprises:

[0015] The calling management unit is configured to determine a target order based on attributes of each calling request to the large language model, and call the large language model to respond to each calling request based on the target order.

[0016] According to an embodiment of the present application, the large model module further comprises a risk assessment unit; the risk assessment unit is configured to perform risk assessment on the target question input by the user acquired.

[0017] According to an embodiment of the present application, the model management module comprises:

[0018] The effect assessment unit is configured to assess the usage amount and usage effect of the remaining service life prediction model, the overall device efficiency improvement model and the health assessment model.

[0019] According to an embodiment of the present application, the model management module comprises:

[0020] The risk assessment unit is configured to assess the usage amount of the large language model and the risk of calling the large language model.

[0021] According to an embodiment of the present application, the system further comprises:

[0022] The machine learning model module is configured to acquire the remaining service life prediction model, the overall device efficiency improvement model, the health assessment model and the large language model through machine learning.

[0023] According to an embodiment of the present application, the system further comprises a user interaction terminal configured to interact with the user.

[0024] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0025] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description, taken in conjunction with the following drawings in which:

[0026] Figure 1 is one of the structural schematic diagrams of the equipment maintenance system applied to the field of general-purpose semiconductors provided by the embodiments of the present application.

[0027] Figure 2 is another structural schematic diagram of the equipment maintenance system applied to the field of general-purpose semiconductors provided by the embodiments of the present application.

[0028] Figure 3 is a third structural schematic diagram of the equipment maintenance system applied to the field of general-purpose semiconductors provided by the embodiments of the present application.

[0029] Figure 4 is a structural schematic diagram of an electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present application will be described clearly below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art belong to the scope of protection of the present application.

[0031] The terms "first", "second", and the like in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be exchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than that illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally a category, and do not limit the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / ", generally indicates that the front and rear associated objects are in a "or" relationship.

[0032] The application of the equipment maintenance system applied to the field of general-purpose semiconductors, the electronic device and the readable storage medium provided by the embodiments of the present application will be described in detail below in conjunction with the drawings and specific embodiments and their application scenarios.

[0033] The equipment maintenance system applied to the field of general-purpose semiconductors can be applied to a terminal, and can be specifically executed by hardware or software in the terminal.

[0034] The terminal includes, but is not limited to, a portable communication device such as a mobile phone or tablet computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad). It should also be understood that, in some embodiments, the terminal may not be a portable communication device, but a desktop computer having a touch-sensitive surface (e.g., a touch screen display and / or a touch pad).

[0035] In the following embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse, and a joystick.

[0036] The embodiment of the present application provides an equipment maintenance system for application in the pan-semiconductor field. The execution subject of the equipment maintenance system for application in the pan-semiconductor field can be an electronic device or a functional module or functional entity in the electronic device that can implement the equipment maintenance system for application in the pan-semiconductor field. The electronic devices mentioned in the embodiment of the present application include but are not limited to mobile phones, tablet computers, computers, cameras and wearable devices, etc. The equipment maintenance system for application in the pan-semiconductor field provided by the embodiment of the present application is explained below using electronic devices as the execution subject as an example.

[0037] like Figure 1 As shown, the equipment maintenance system applied in the pan-semiconductor field includes a data module 110, a large model module 120 and a maintenance warning module 130. The data module 110 is used to obtain the target operation data of the target equipment; the large model module 120 is used to obtain the fault diagnosis result of the target equipment according to the target operation data based on the large language model; the maintenance warning module 130 is used to process the target operation data based on the remaining service life prediction model, the overall equipment efficiency improvement model and the health assessment model, and generate equipment optimization suggestions for the target equipment based on the processing results and the fault diagnosis results.

[0038] In actual implementation, the target device can be any device in the semiconductor field, such as a wafer saw, physical vapor deposition equipment, or any other device in the semiconductor field. The target operating data can include the target device's real-time operating parameters and / or evaluation data sets (such as device operating status coefficients, overall device efficiency indicators, and open industrial parameter data sets for the target device).

[0039] In actual execution, the data module may acquire the target operation data of the target device based on the target acquisition device. In some embodiments, the data module may acquire the target operation data of the target device based on the target sensor.

[0040] In some embodiments, the large model module can be used to determine a target order for calling the large language model, and call the large language model based on the target order to obtain a fault diagnosis result for the target device.

[0041] In some embodiments, in the case of obtaining a target question of user input, the large model module can obtain a fault diagnosis result based on the large language model and the target question.

[0042] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can further include a machine learning model module, and the remaining useful life prediction model, the overall equipment efficiency improvement model, the health assessment model, and the large language model can all be obtained based on the machine learning model module.

[0043] In some embodiments, the maintenance warning module can be used to process target operation data based on the remaining useful life prediction model, the overall equipment efficiency improvement model, and the health assessment model. The processing can include overall equipment efficiency parameter prediction, remaining useful life prediction, health assessment (such as abnormal position investigation, fault type determination, etc.). The results of the processing can include fault type information, abnormal position information, remaining useful life prediction results, overall equipment efficiency parameter prediction values, and overall equipment efficiency parameter optimization values, etc.

[0044] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can further include a user interaction terminal, which can be used to display the equipment optimization suggestions for the target equipment generated by the maintenance warning module to the user.

[0045] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can further include a model management module, which can be used to perform performance evaluation on the remaining useful life prediction model, the overall equipment efficiency improvement model, the health assessment model, and the large language model, and send the evaluation results to the machine learning model module, which can update the remaining useful life prediction model, the overall equipment efficiency improvement model, the health assessment model, and the large language model based on the evaluation results.

[0046] According to the equipment maintenance system applied to the field of general semiconductors of the present application, the maintenance warning module processes the target operation data based on the remaining useful life prediction model, the overall equipment efficiency improvement model, and the health assessment model, and generates equipment optimization suggestions for the target equipment based on the results of the processing and the fault diagnosis results generated by the large model module based on the large language model, so as to provide the reasons and basis for performing equipment optimization, thereby reducing the difficulty of optimizing the target equipment based on the equipment optimization suggestions.

[0047] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can further include a model management module for performing performance evaluation on the remaining useful life prediction model, the overall equipment efficiency improvement model, the health assessment model, and the large language model, and generating a model evaluation report.

[0048] In actual implementation, the performance evaluation can include evaluation of model usage quantity and model usage effect. In some embodiments, the model management module can be used to evaluate the remaining service life prediction model, the overall device efficiency improvement model, and the health assessment model in terms of model usage quantity and model usage effect.

[0049] In actual implementation, the performance evaluation can include evaluation of model usage quantity and model usage risk. In some embodiments, the model management module can be used to evaluate the large language model in terms of model usage quantity and model usage risk.

[0050] In some embodiments, the model management module can also be used to send the model evaluation report to the machine learning model module, so that the machine learning model module further updates the remaining service life prediction model, the overall device efficiency improvement model, the health assessment model, and the large language model based on the model evaluation report.

[0051] According to the device maintenance system applied to the field of general semiconductors according to the present application, the usage of the model is counted, the usage frequency and effect of the model are recorded, the performance of the model is evaluated, and the version of the model is managed, so as to ensure the effectiveness and stability of the model in actual application. The large model is evaluated in terms of multiple indexes, so as to ensure the reliability and safety of the large model in actual application.

[0052] In some embodiments, the device maintenance system applied to the field of general semiconductors can further include a knowledge base module, the knowledge base module including an enhanced knowledge graph in the field of general semiconductors; and a large model module, specifically configured to, in a case where a target question input by a user is acquired, call the knowledge graph based on a large language model and the target question to perform retrieval based on a thinking chain and acquire a fault diagnosis result.

[0053] In actual implementation, the knowledge graph can be acquired based on a knowledge document. In some embodiments, the knowledge document input by the user can be acquired through a user interactive terminal, and the knowledge document can be in any form such as text, picture, video, or audio. The knowledge base module can acquire the knowledge document based on a template and a preset format.

[0054] In actual implementation, the knowledge base module can also be used to classify and manage the knowledge document. In some embodiments, in a case where a keyword input by a user is acquired, the knowledge document can be found through full-text retrieval technology according to the keyword.

[0055] In some embodiments, knowledge points can be extracted from the knowledge document, a database, and external resources through an automatic tool to construct the knowledge graph. The optimized knowledge graph structure can also be acquired through the user interactive terminal to ensure the accuracy and integrity of the knowledge graph.

[0056] In some embodiments, a graph database technology that supports efficient access to large-scale data can be used to store knowledge graphs. The graph database has high-performance query and analysis capabilities to ensure that the required knowledge graph can be quickly obtained.

[0057] In some embodiments, the knowledge base module may include a knowledge graph and a query interface corresponding to the knowledge graph. The query interface may be an interface based on target design principles (e.g., RESTful style), and the query interface may support a target protocol (e.g., HTTP protocol). The knowledge graph may be accessed by calling the query interface to ensure the security of the knowledge graph. Furthermore, the query interface may be managed, and its usage, such as activation, deactivation, and update, may be monitored and recorded, generating interface call logs and statistical reports.

[0058] In actual implementation, the target question can be a problem within the broader semiconductor field. The target question can be a question based on user input obtained from the user interaction terminal, and the target question can be used to obtain troubleshooting suggestions for the broader semiconductor field. The target question can be in any theoretically feasible form, such as a field, file, image, or audio, and this application does not impose specific restrictions on this.

[0059] In some embodiments, the large model module can perform a thought chain-based search on the enhanced knowledge graph in the pan-semiconductor field based on the large language model and the target question input by the user, obtain the search results, and obtain the fault diagnosis results based on the search results.

[0060] In some embodiments, the large model module can convert the target problem into multiple sub-problems based on the large language model, and the multiple sub-problems constitute the thinking chain corresponding to the target problem; based on the large language model, each sub-problem is converted into a SPARQL statement for multi-hop query on the enhanced knowledge graph of the pan-semiconductor field; each SPARQL statement is replaced with a synonym to obtain an alternative SPARQL statement group; based on each alternative SPARQL statement group, a beam search is performed on the enhanced knowledge graph of the pan-semiconductor field to obtain retrieval results, and based on multiple retrieval results, a fault diagnosis result is obtained.

[0061] According to the equipment maintenance system for the pan-semiconductor field provided by the embodiments of the present application, the large model module, when obtaining the target question input by the user, calls the knowledge graph to perform a thought chain-based search based on the large language model and the target question, and obtains the fault diagnosis results. By combining the knowledge graph and the large language model, the system can provide fault diagnosis results by combining the rich domain knowledge in the knowledge graph, providing a more comprehensive and in-depth solution. Furthermore, the large language model is used for intelligent analysis and reasoning to generate detailed fault diagnosis reports and maintenance recommendations, improving the accuracy of fault diagnosis and the effectiveness of maintenance recommendations.

[0062] In some embodiments, the knowledge base module includes an updating unit; the updating unit is used to periodically update the knowledge graph and the target response format.

[0063] In actual implementation, the knowledge base module may include preset templates and syntax parsing rules. The preset templates may include templates uploaded by users that can be obtained through the user interaction terminal.

[0064] In some embodiments, the knowledge base module can obtain the equipment usage evaluation values ​​such as the remaining service life prediction value and the overall equipment efficiency parameter prediction value of the target equipment obtained by the maintenance warning module based on the remaining service life prediction model, the overall equipment efficiency improvement model and the health assessment model according to a preset cycle, and update the knowledge document based on the equipment usage evaluation value to ensure the timeliness and accuracy of the knowledge document content.

[0065] In some embodiments, the knowledge base module can send knowledge documents based on a preset frequency via email, notification, or any other theoretically feasible method. For example, the knowledge base module can send knowledge documents based on a preset frequency via email and notification via the user interaction terminal to ensure timely delivery of knowledge information, thereby improving the accuracy of knowledge points.

[0066] In some embodiments, the large model module may include a call management unit configured to determine a target order based on attributes of each call request to the large language model, and call the large language model to respond to each call request based on the target order.

[0067] In some embodiments, the call management unit can be used to execute a request scheduling step, a task assignment step, and a call monitoring step. The request scheduling step is used to assign the target order of calling the large language model according to the priority of different users, the request time, and the importance of the corresponding problem in the call chain. The task assignment step is used to assign the request to the target retrieval-augmented generation (Retrieval-Augmented Generation) text generation framework according to the request type of the user request. For example, for troubleshooting-related requests, they are assigned to a pre-designed troubleshooting RAG text generation framework, and for standard query requests, they are assigned to a pre-designed text retrieval RAG text generation framework to ensure efficient and accurate execution. The call monitoring step is used to monitor the call management process in real time, record call logs, and ensure the reliability and traceability of the call.

[0068] In some embodiments, the large model module can include a model fine-tuning unit for performing a model pre-training step, a model fine-tuning step, and a model optimization step. The model pre-training step is used to preliminarily train the base large model through a large-scale data set to obtain a large language model with general features and basic capabilities. The model fine-tuning step is used to perform refined training according to data of a specific task and field to improve the performance of the large language model in a specific scenario. The model optimization step is used to improve the performance and efficiency of the large language model through parameter adjustment and structure optimization.

[0069] In some embodiments, the large model module can include a knowledge base calling unit for performing a knowledge retrieval step, a knowledge matching step, and a knowledge application step. The knowledge retrieval step is used to quickly find relevant knowledge points from the knowledge base through intelligent search technology. The knowledge matching step is used to match the retrieved knowledge with the current task or problem to ensure that the provided knowledge is the most relevant and useful. The knowledge application step is used to apply the matched knowledge to actual tasks to support automated decision-making and problem solving.

[0070] In some embodiments, the large model module can include an RAG unit for performing a knowledge enhancement generation step, a retrieval enhancement generation step, and a fusion generation step. The knowledge enhancement generation step is used to enhance the accuracy and richness of generated content by calling a knowledge graph in the knowledge base. The retrieval enhancement generation step is used to combine retrieval results and a generation model to provide more accurate answers and content. The fusion generation step is used to integrate the advantages of multiple generation methods to provide high-quality generation results.

[0071] According to the device maintenance system applied to the field of general semiconductors provided by the embodiments of the present application, through the calling management unit in the large model module, the target order is determined based on the attributes of each calling request of the large language model, and the large language model is called based on the target order, so as to avoid the situation that the calculation complexity of the large language model is high, and a large number of user requests in a short period of time will cause hidden dangers to the large model system.

[0072] In some embodiments, the large model module can further include a risk assessment unit for performing risk assessment on the target problem of the user input obtained.

[0073] In some embodiments, the risk assessment unit is configured to perform a risk identification step, a risk analysis step, and a risk control step. The risk identification step is configured to identify whether the user question has offensive or inappropriate content through natural language processing and sentiment analysis techniques. The risk analysis step is configured to perform a detailed analysis of the identified potential risks to assess their potential impact and probability of occurrence. The risk control step is configured to ensure the compliance and security of the answer content by developing and implementing control measures, reducing or eliminating the negative impact of potential risks, and ensuring the security of the system.

[0074] In some embodiments, the model management module can include an effect evaluation unit configured to evaluate the usage amount and usage effect of the remaining useful life prediction model, the overall device efficiency improvement model, and the health assessment model.

[0075] In actual implementation, the effect evaluation unit can be configured to perform model usage evaluation on the remaining useful life prediction model, the overall device efficiency improvement model, and the health assessment model. In some embodiments, the effect evaluation unit is configured to perform a model call statistics process, a usage frequency analysis process, and a resource consumption monitoring process. The model call statistics process is configured to record and count the number of calls and the usage duration of the above-mentioned models to obtain detailed usage reports. The usage frequency analysis process is configured to analyze the usage frequency of the above-mentioned models and optimize the model resource configuration for high-frequency usage models. The resource consumption monitoring process is configured to monitor the consumption of the above-mentioned models in terms of computing resources and storage resources to ensure reasonable use and optimized configuration of resources.

[0076] In actual implementation, the effect evaluation unit can also be configured to perform model usage effect evaluation on the remaining useful life prediction model, the overall device efficiency improvement model, and the health assessment model. In some embodiments, the effect evaluation unit is configured to perform a model performance evaluation process, a model effect monitoring process, and an evaluation report generation process. The model performance evaluation process is configured to evaluate the performance and effect of the model based on performance indicators such as accuracy, recall rate, F1 value, etc. The model effect monitoring process is configured to monitor the performance of the model in actual application in real time, identify and warn of a decrease in model effect. The evaluation report generation process is configured to generate a model evaluation report periodically, which includes detailed performance indicators and improvement suggestions to help operation and maintenance personnel continuously optimize the model.

[0077] In some embodiments, the model management module can also include a risk assessment unit configured to evaluate the usage amount of the large language model and the risk of calling the large language model.

[0078] In actual execution, the model management module can further include a risk assessment unit for model usage assessment of the large language model. In some embodiments, the risk assessment unit is used to perform a large model call statistics process, a large model resource consumption process, and a large model performance analysis process. Among them, the large model call statistics part records the number of calls and usage of the large model, providing detailed usage statistics. The large model resource consumption part monitors the consumption of the large model on computing and storage resources, optimizing resource allocation and management. The large model performance analysis part evaluates the effect and performance of the large model through performance indicators and user feedback, and provides optimization suggestions.

[0079] In actual execution, the risk assessment unit is also used to call the risk assessment of the large language model. In some embodiments, the risk assessment unit is used to perform a risk identification process, a risk assessment process, and a risk control process. Among them, the risk identification process is used to identify potential risks and problems by monitoring and analyzing the running data of the large language model. The risk assessment process is used to analyze the identified risks in detail to assess their impact and probability. The risk control process is used to develop and implement risk control measures to reduce or eliminate risks in the operation of the large language model, ensuring the safety and stability of the system.

[0080] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can further include a machine learning model module for obtaining a remaining useful life prediction model, an overall equipment efficiency improvement model, a health assessment model, and a large language model through machine learning.

[0081] In actual execution, the machine learning model module can include a data preprocessing unit for performing data standardization steps, feature engineering steps, and data enhancement steps. The data standardization step is used to normalize or standardize the original data to ensure data consistency and comparability. The feature engineering step is used to improve the prediction ability of the model by automatically extracting and selecting key features. The data enhancement step is used to generate more training samples using data enhancement techniques to improve the generalization ability of the model.

[0082] In actual execution, the machine learning model module can include an offline training unit that integrates EfficientNet, stream model, Transformer network, transfer learning network, KNN model, and BERT model, etc. to provide rich algorithm selection. The offline training unit can be used to perform automated training steps and high-performance computing steps. The automated training step is used to automatically perform the model training process to reduce human intervention and improve training efficiency. The high-performance computing step is used to use distributed computing and GPU acceleration technology to accelerate the model training process and improve training speed and efficiency.

[0083] In actual implementation, the machine learning model module can include a model verification unit for performing a cross-validation step, a model indicator evaluation step, and a comparative analysis step. The cross-validation step is used to evaluate the performance of the model based on cross-validation techniques to ensure the stability and reliability of the model. The model indicator evaluation step is used to comprehensively evaluate the model effect based on various evaluation indicators such as accuracy, recall rate, F1 value, and ROC curve. The comparative analysis step is used to compare different models to help users select the best model.

[0084] In actual implementation, the machine learning model module can include a model library for performing model storage and management, model calling interface management, and model document generation. Model storage and management are used to establish a unified model library to store and manage trained models and support model version control. Model calling interface management is used to provide API interfaces to support fast calling and deployment of models and facilitate integration into business systems. Model document generation is used to generate model description documents that detail the training process, parameter settings, and performance indicators of the model to facilitate user understanding and use of the model.

[0085] In actual implementation, the machine learning model module can include a learning unit for performing a model online updating step, an adaptive learning step, and a user feedback learning step. The model online updating step is used for online updating and iteration of the model to ensure its adaptability in real-time data environments. The adaptive learning step is used to automatically adjust model parameters based on new data and feedback to improve the prediction accuracy of the model. User feedback learning is used to continuously optimize the model based on user feedback to improve user experience and system performance.

[0086] According to the device maintenance system applied to the field of general semiconductors provided by the embodiments of the present application, through data preprocessing, model training, model verification, model library management, and continuous learning, data preprocessing includes data format conversion and feature engineering to ensure that the data is suitable for model training and verification. Model training uses EfficientNet, KNN, and other models for training to improve the accuracy of the model using advanced machine learning algorithms. Model verification is performed through cross-validation and multiple evaluation indicators to ensure model performance and reliability. Model library management and storage of different versions of the model support fast loading and deployment of the model. Continuous learning realizes online updating and adaptive adjustment of the model to improve its adaptability and accuracy in actual applications.

[0087] In some embodiments, the device maintenance system applied to the field of general semiconductors can further include a user interaction terminal for interacting with users.

[0088] In actual implementation, the user interaction terminal can include a visualization unit, which is configured to perform data display, trend analysis, and interface interaction. The data display is configured to visually display the running state of the equipment, OEE value, RUL prediction result, health score and other information through charts, dashboards and the like. The trend analysis is configured to analyze the running law and future trend of the equipment based on historical trends and prediction analysis of data. The interactive interface is based on a friendly user interface, supports user-defined views and reports, and improves the usability and practicality of the system.

[0089] In actual implementation, the user interaction terminal can include an interface unit, which includes an API interface, a third-party system integration channel, and a device compatibility channel. The API interface can be a standardized API interface, supporting system function extension and integration. The third-party system integration channel is configured to integrate with other systems and services, enhancing the interoperability of the system. The device compatibility channel is configured to ensure the compatibility of the system with various devices, supporting data acquisition and control of multiple devices.

[0090] In actual implementation, the user interaction terminal can include a report generation unit, which is configured to perform a periodic report generation step, a custom report template step, and a report automatic sending step. The periodic report generation step is configured to automatically generate reports on system operation and model usage according to a preset time interval. The custom report template step is configured to support user-defined report formats and contents, meeting individual needs. The report automatic sending step is configured to automatically send the generated reports to relevant users and systems through email, notifications and the like, ensuring timely information delivery.

[0091] In actual implementation, the user interaction terminal can include an optimization suggestion unit, which is configured to generate optimization suggestions, executable operation suggestions, and optimization suggestion feedback based on data analysis. The optimization suggestion unit can analyze system and model data to provide suggestions for optimizing system and model performance. The optimization suggestion unit can also convert optimization suggestions into specific operation plans to guide maintenance personnel to implement optimization measures. The optimization suggestion unit can also collect and analyze the implementation effects of optimization measures to continuously improve and optimize system performance.

[0092] According to the device maintenance system applied to the field of general semiconductors provided by the embodiments of the present application, the user interaction terminal is used to interact with the user, thereby enhancing the user's experience.

[0093] In order to better understand the device maintenance system applied to the field of general semiconductors provided by the embodiments of the present application, further explanation is made as follows, and it should be understood that the following discussion is only exemplary.

[0094] As Figure 2As shown, the present application provides a device maintenance system applied to the field of general semiconductors, which comprises a data module 210, a large model module 220, a maintenance early warning module 230, a model management module 240, a machine learning model module 250, a user interaction terminal 260, and a knowledge base module 270.

[0095] In some embodiments, as shown, the data module can comprise a data acquisition unit for accessing data sources, collecting sensor data, collecting system logs, and importing user data. Data source access obtains real-time data or historical data by establishing connections with various data sources, including but not limited to databases, API interfaces, third-party services, and local file systems. Sensor data acquisition obtains data on environment, status, performance, etc. from various sensor devices and performs preliminary processing. System log collection is used to collect and analyze system operation logs, including system events, error logs, and access logs, etc. to facilitate system status monitoring and fault diagnosis. User data import is used to import data based on interface or batch file upload methods, supporting multiple data formats to meet user-defined data analysis and processing needs. Figure 3 In some embodiments, the data module can comprise a data cleaning unit for data preprocessing and data label annotation. Data preprocessing is used to convert the collected data to a consistent and complete format, handle missing values and outliers, and ensure data consistency and integrity. Data label annotation is used to define and annotate labels for training data, supporting automatic annotation and manual annotation to facilitate machine learning model training and verification.

[0096] In some embodiments, the data module can further comprise a data pool for raw data storage, cleaned data storage, and historical data storage. Raw data storage is used to save the unprocessed raw data collected from various data sources. Cleaned data storage is used to store data that has been preprocessed and cleaned to ensure data quality for subsequent analysis and modeling. Historical data storage is used to record the history and changes of data, supporting indexing and traceability management of historical data, which can include but is not limited to a data warehouse based on the DVC version control framework to ensure traceability and effectiveness of version management.

[0097]

[0098] ​In some embodiments, the data module can also include a data security and privacy unit for performing data access control, critical data encryption, and data de-identification processing. Data access control is used to restrict authorized user access and operation of sensitive data. Critical data encryption is used to encrypt important data in storage and transmission to prevent unauthorized access and leakage. Data de-identification processing is used to protect the privacy of local learning data, which can use techniques such as differential privacy to ensure that personal information is not exposed. In addition, the data security and privacy unit also includes data security logging for monitoring and recording data access and operation behavior to ensure traceability of data operations. The data module 210 also includes but is not limited to the specifications of GB / T 35273-2020 related to data privacy and security regulations to ensure the legality and compliance of data processing.

[0099] In some embodiments, the maintenance and early warning module can include an overall device efficiency improvement unit for performing data collection, calculation and analysis, and result display steps. The data collection step is used to obtain device running status, production efficiency, and quality data from various sensors and system logs. The calculation and analysis step is used to calculate the OEE value of the device based on the collected data, analyze the efficiency of each link in the production process of the device, and obtain recommended controllable parameter suggestions based on the trained KNN module. The result display step is used to visually display the calculation results, trends, and parameter suggestions of OEE through visualization tools, helping operation and maintenance personnel understand the device running status and timely discover and solve problems.

[0100] In some embodiments, the maintenance and early warning module can include a remaining useful life unit for performing data preprocessing, model prediction, and result interpretation steps. The data preprocessing step is used to clean and feature extract the collected device running data to ensure data quality. The model prediction step is used to predict the remaining useful life of the device based on the trained Transformer network and transfer learning model. The result interpretation step is used for explainable analysis of the prediction results to help operation and maintenance personnel understand the running status and expected life of the device and develop corresponding maintenance plans.

[0101] In some embodiments, the maintenance and early warning module can include a health assessment unit for data collection, health score generation, and health report generation. Data collection is used to obtain device running data from device sensors, system logs, and user feedback. Health score generation scores the health status of the device based on the collected data to provide a comprehensive health assessment result. Health report generation is used to generate a detailed device health report containing health scores, evaluation basis, and maintenance suggestions to help operation and maintenance personnel comprehensively understand the health status of the device.

[0102] In some embodiments, the maintenance early warning module can include a large model fault diagnosis unit for performing fault detection, fault location, and fault analysis. Fault detection performs real-time analysis on device operation data through fine-tuned domain large language models to detect potential faults. Fault location accurately locates the location and cause of the fault in combination with the running state and historical data of the device. Fault analysis provides a detailed analysis report of the fault, including the cause of the fault, the scope of the impact, and the solution, helping operation and maintenance personnel to quickly troubleshoot and restore normal operation of the device.

[0103] In some embodiments, the maintenance early warning module can include a real-time monitoring unit for performing real-time data stream processing, anomaly detection and response, instant notification and alarm. Real-time data stream processing is used to collect and analyze device operation data in real time to ensure the timeliness of the data. Anomaly detection and response detect anomalies in device operation through intelligent algorithms and respond quickly to take appropriate measures. Instant notification and alarm are used to send notifications and alarm information to operation and maintenance personnel immediately upon detecting anomalies or faults to ensure that problems can be handled in a timely manner, reducing device downtime and production losses.

[0104] In some embodiments, the large model module is configured to obtain a fault diagnosis result of a target device based on a large language model and target operation data.

[0105] In some embodiments, the device maintenance system applied to the field of generic semiconductors can further include a knowledge base module including an enhanced knowledge graph in the field of generic semiconductors, and a large model module configured to, based on a large language model and a target problem input by a user, call the knowledge graph to perform thought chain-based retrieval and obtain a fault diagnosis result.

[0106] In actual implementation, the knowledge graph can be obtained based on knowledge documents. In some embodiments, the knowledge documents input by the user can be obtained through a user interaction terminal, and the knowledge documents can be in any form such as text, picture, video, or audio. The knowledge base module can obtain the knowledge documents based on templates and preset formats.

[0107] In actual implementation, the knowledge base module can also be used to classify and manage the knowledge documents. In some embodiments, based on a keyword input by the user, the knowledge documents can be found through full-text retrieval technology according to the keyword.

[0108] In some embodiments, knowledge points can be extracted from knowledge documents, databases, and external resources through automated tools to construct the knowledge graph. Optimized knowledge graph structures can also be obtained through a user interaction terminal to ensure the accuracy and integrity of the knowledge graph.

[0109] In some embodiments, the knowledge graph can be stored by using a graph database technology supporting efficient access of large-scale data, the graph database has high-performance query and analysis capabilities, ensuring that the required knowledge graph can be quickly obtained.

[0110] In some embodiments, the knowledge base module can include a knowledge graph and a query interface corresponding to the knowledge graph, the query interface can be an interface based on target design principles (such as RESTful style), and the query interface can support target protocols (such as HTTP protocol). The knowledge graph can be accessed by calling the query interface, ensuring the security of the knowledge graph. In addition, the query interface can be managed, and the use cases such as enabling, disabling, and updating of the query interface can be monitored and recorded to generate interface call logs and statistical reports.

[0111] In actual execution, the target problem can be a problem in the field of general semiconductors. The target problem can be a problem based on user input obtained through a user interaction terminal, and the target problem can be used to obtain a processing suggestion for a fault in the field of general semiconductors. The target problem can be in any theoretically feasible form such as a field, a file, an image, or an audio, and the present application does not make specific limitations thereto.

[0112] In some embodiments, the large model module can be based on a large language model, and based on a target problem input by a user, perform a thought chain-based search on an enhanced knowledge graph of the field of general semiconductors to obtain a search result, and based on the search result, obtain a fault diagnosis result.

[0113] In some embodiments, the large model module can be based on a large language model, and convert a target problem into a plurality of sub-problems, the plurality of sub-problems constituting a thought chain corresponding to the target problem; based on the large language model, convert each sub-problem into a SPARQL statement for performing a multi-hop query on an enhanced knowledge graph of the field of general semiconductors; respectively perform synonym replacement on each SPARQL statement to obtain a group of alternative SPARQL statements; respectively based on each group of alternative SPARQL statements, perform a beam search on the enhanced knowledge graph of the field of general semiconductors to obtain a search result, and based on a plurality of search results, obtain a fault diagnosis result.

[0114] In some embodiments, the knowledge base module can further include an updating unit; the updating unit is used to periodically update the knowledge graph and the target response format.

[0115] In actual execution, the knowledge base module can include a preset template and a syntax analysis rule, and the preset template can include a template uploaded by a user through a user interaction terminal.

[0116] In some embodiments, the knowledge base module can obtain the remaining service life prediction value of the target device and the overall device efficiency parameter prediction value and the like device usage evaluation value obtained by the maintenance early warning module based on the remaining service life prediction model, the overall device efficiency improvement model and the health assessment model according to a preset period, and update the knowledge document based on the device usage evaluation value, to ensure the timeliness and accuracy of the content of the knowledge document.

[0117] In some embodiments, the knowledge base module can send the knowledge document through any theoretically feasible way such as mail and notification based on a preset frequency. For example, the knowledge base module sends the knowledge document in the form of mail and notification through the user interaction terminal based on a preset frequency, to ensure the timely delivery of knowledge information, thereby improving the accuracy of knowledge points.

[0118] In some embodiments, the large model module can include a call management unit, which is configured to determine a target order based on the attributes of each call request to the large language model, and call the large language model to respond to each call request based on the target order.

[0119] In some embodiments, the call management unit can be used to perform a request scheduling step, a task allocation step and a call monitoring step. The request scheduling step is used to allocate the target order of calling the large language model in the call chain according to the priority of different users, the request time and the importance of the corresponding problem. The task allocation step is used to allocate the request to a target Retrieval-Augmented Generation (RAG) text generation framework according to the request type of the user request, for example, allocating the request related to fault repair to a pre-designed fault repair RAG text generation framework, and allocating the request for specification query to a pre-designed text retrieval RAG text generation framework, to ensure efficient and accurate execution. The call monitoring step is used to monitor the call management process in real time, record the call log, and ensure the reliability and traceability of the call.

[0120] In some embodiments, the large model module can include a model fine-tuning unit, which is configured to perform a model pre-training step, a model fine-tuning step and a model optimization step. The model pre-training step is used to preliminarily train the base large model through a large-scale data set, to obtain a large language model with general features and basic capabilities. The model fine-tuning step is used to perform refined training according to the data of specific tasks and fields, to improve the performance of the large language model in specific scenarios. The model optimization step is used to improve the performance and efficiency of the large language model through parameter adjustment and structure optimization.

[0121] In some embodiments, the large model module can include a knowledge base calling unit for performing a knowledge retrieval step, a knowledge matching step, and a knowledge application step. The knowledge retrieval step is used to quickly find relevant knowledge points from the knowledge base through intelligent search technology. The knowledge matching step is used to match the retrieved knowledge with the current task or problem, ensuring that the provided knowledge is the most relevant and useful. The knowledge application step is used to apply the matched knowledge to the actual task, supporting automated decision-making and problem-solving.

[0122] In some embodiments, the large model module can include an RAG unit for performing a knowledge enhancement generation step, a retrieval enhancement generation step, and a fusion generation step. The knowledge enhancement generation step is used to enhance the accuracy and richness of the generated content by calling the knowledge graph in the knowledge base. The retrieval enhancement generation step is used to combine retrieval results and generation models to provide more accurate answers and content. The fusion generation step is used to integrate the advantages of multiple generation methods to provide high-quality generation results.

[0123] In some embodiments, the large model module can also include a risk assessment unit for performing a risk assessment on the target problem of the obtained user input.

[0124] In some embodiments, the risk assessment unit is used to perform a risk identification step, a risk analysis step, and a risk control step. The risk identification step is used to identify whether the user's question has offensive or inappropriate content through natural language processing and sentiment analysis techniques. The risk analysis step is used to analyze the potential risks identified in detail, evaluate their possible impact and probability of occurrence. The risk control step is used to ensure the compliance and security of the answer content by developing and implementing control measures, reducing or eliminating the negative impact of potential risks, and ensuring the safety of the system.

[0125] In some embodiments, the device maintenance system applied to the field of general semiconductors can also include a model management module for performing performance evaluation on the remaining useful life prediction model, the overall device efficiency improvement model, the health assessment model, and the large language model, and generating a model evaluation report.

[0126] In actual execution, performance evaluation can include evaluation of model usage and model usage effect. In some embodiments, the model management module can be used to evaluate the model usage and model usage effect of the remaining useful life prediction model, the overall device efficiency improvement model, and the health assessment model.

[0127] In actual implementation, the performance evaluation can include evaluation of model usage amount and model usage risk. In some embodiments, the model management module can be used for model usage amount and model usage risk evaluation of the large language model.

[0128] In some embodiments, the model management module can also be used to send the model evaluation report to the machine learning model module, so that the machine learning model module further updates the remaining service life prediction model, the overall device efficiency improvement model, the health assessment model and the large language model based on the model evaluation report.

[0129] In some embodiments, the model management module can also include a risk assessment unit; the risk assessment unit is used for risk assessment on the target problem of the obtained user input.

[0130] In some embodiments, the risk assessment unit is used to perform a risk identification step, a risk analysis step and a risk control step. The risk identification step is used to identify whether the user question has offensive or inappropriate content through natural language processing and sentiment analysis technology. The risk analysis step is used to analyze the potential risks identified in detail, evaluate the possible impact and probability of occurrence. The risk control step is used to ensure the compliance and safety of the answer content by formulating and implementing control measures, reducing or eliminating the negative impact of potential risks, and ensuring the safety of the system.

[0131] In some embodiments, the model management module can include an effect evaluation unit, which is used to evaluate the usage amount and effect of the remaining service life prediction model, the overall device efficiency improvement model and the health assessment model.

[0132] In actual implementation, the effect evaluation unit can be used for model usage amount evaluation of the remaining service life prediction model, the overall device efficiency improvement model and the health assessment model. In some embodiments, the effect evaluation unit is used for model call statistics process, usage frequency analysis process and resource consumption monitoring process. Among them, the model call statistics process is used to record and count the number of calls and the usage time of the above models, and obtain detailed usage reports. The usage frequency analysis process is used to analyze the usage frequency of the above models, and optimize the model resource configuration of the high-frequency usage models. The resource consumption monitoring process is used to monitor the consumption of the above models on computing resources and storage resources, to ensure the reasonable use and optimized configuration of resources.

[0133] In actual implementation, the effect evaluation unit can also be used to evaluate the model usage effect of the remaining useful life prediction model, the overall device efficiency improvement model, and the health evaluation model. In some embodiments, the effect evaluation unit is used to perform a model performance evaluation process, a model effect monitoring process, and an evaluation report generation process. The model performance evaluation process is used to evaluate the performance and effect of the model based on performance indicators such as accuracy, recall rate, F1 value, etc. The model effect monitoring process is used to monitor the performance of the model in actual application in real time, identify and warn of a decrease in model effect. The evaluation report generation process is used to generate a model evaluation report periodically, and the model evaluation report includes detailed performance indicators and improvement suggestions to help operation and maintenance personnel continuously optimize the model.

[0134] In some embodiments, the model management module can also include a risk evaluation unit for evaluating the usage amount of the large language model and the risk of calling the large language model.

[0135] In actual implementation, the model management module can also include a risk evaluation unit, and the risk evaluation unit is used to evaluate the usage amount of the large language model. In some embodiments, the risk evaluation unit is used to perform a large model calling statistics process, a large model resource consumption process, and a large model performance analysis process. The large model calling statistics part records the calling times and usage of the large model, and provides detailed usage statistics data. The large model resource consumption part monitors the consumption of the large model on computing and storage resources, optimizes resource allocation and management. The large model performance analysis part evaluates the effect and performance of the large model through performance indicators and user feedback, and provides optimization suggestions.

[0136] In actual implementation, the risk evaluation unit is also used to evaluate the risk of calling the large language model. In some embodiments, the risk evaluation unit is used to perform a risk identification process, a risk evaluation process, and a risk control process. The risk identification process is used to identify potential risks and problems by monitoring and analyzing the running data of the large language model. The risk evaluation process is used to analyze the identified risks in detail to evaluate their impact and probability. The risk control process is used to develop and implement risk control measures to reduce or eliminate risks in the running of the large language model, and to ensure the safety and stability of the system.

[0137] In some embodiments, the device maintenance system applied to the field of general semiconductors can also include a machine learning model module for obtaining the remaining useful life prediction model, the overall device efficiency improvement model, the health evaluation model, and the large language model through machine learning.

[0138] In actual implementation, the machine learning model module can include a data preprocessing unit for performing data standardization steps, feature engineering steps, and data augmentation steps. The data standardization steps are used to normalize or standardize the original data, ensuring the consistency and comparability of the data. The feature engineering steps are used to improve the predictive ability of the model by automatically extracting and selecting key features. The data augmentation steps are used to generate more training samples using data augmentation techniques to improve the generalization ability of the model.

[0139] In actual implementation, the machine learning model module can include an offline training unit that integrates multiple models such as EfficientNet, stream model, Transformer network, transfer learning network, KNN model, and BERT model, providing rich algorithm selection. The offline training unit can be used to perform automated training steps and high-performance computing steps. The automated training steps are used to automate the model training process, reducing manual intervention and improving training efficiency. The high-performance computing steps are used to accelerate the model training process using distributed computing and GPU acceleration techniques, improving training speed and efficiency.

[0140] In actual implementation, the machine learning model module can include a model verification unit for performing cross-validation steps, model index evaluation steps, and comparative analysis steps. The cross-validation steps are used to evaluate the performance of the model based on cross-validation techniques, ensuring the stability and reliability of the model. The model index evaluation steps are used to comprehensively evaluate the model effect based on multiple evaluation indicators such as accuracy, recall rate, F1 value, and ROC curve. The comparative analysis steps are used to compare and analyze different models, helping users select the best model.

[0141] In actual implementation, the machine learning model module can include a model library for performing model storage and management, model calling interface management, and model document generation. Model storage and management are used to establish a unified model library to store and manage trained models, supporting model version control. Model calling interface management is used to provide API interfaces to support fast calling and deployment of models, facilitating integration into business systems. Model document generation is used to generate model specification documents, detailing the training process, parameter settings, and performance indicators of the model, facilitating user understanding and use of the model.

[0142] In actual implementation, the machine learning model module can include a learning unit for performing a model online updating step, an adaptive learning step, and a user feedback learning step. The model online updating step is used for model online updating and iteration, ensuring the adaptability of the model in real-time data environments. The adaptive learning step is used to automatically adjust model parameters based on new data and feedback, improving the prediction accuracy of the model. User feedback learning is used to continuously optimize the model based on user feedback, improving user experience and system performance.

[0143] In some embodiments, the equipment maintenance system applied to the field of general semiconductors can also include a user interaction terminal for interacting with users.

[0144] In actual implementation, the user interaction terminal can include a visualization unit for performing data display, trend analysis, and interface interaction. Data display is used to visually display device running status, OEE value, RUL prediction results, health score, and other information through charts, dashboards, and other means. Trend analysis is used to analyze the running rules and future trends of the device based on historical trends and predictive analysis of data. The interactive interface is based on a friendly user interface that supports user-defined views and reports, improving the ease of use and practicality of the system.

[0145] In actual implementation, the user interaction terminal can include an interface unit including an API interface, a third-party system integration channel, and a device compatibility channel. The API interface can be a standardized API interface that supports system function extension and integration. The third-party system integration channel is used for integration with other systems and services to enhance system interoperability. The device compatibility channel is used to ensure system compatibility with various devices, supporting data collection and control of multiple devices.

[0146] In actual implementation, the user interaction terminal can include a report generation unit for performing a regular report generation step, a custom report template step, and a report automatic sending step. The regular report generation step is used to automatically generate system operation and model usage reports according to a pre-set time interval. The custom report template step is used to support user-defined report formats and content to meet individual needs. The report automatic sending step is used to automatically send generated reports to relevant users and systems through email, notifications, and other means, ensuring timely information delivery.

[0147] In actual implementation, the user interface can include an optimization suggestion unit, which generates optimization suggestions, executable action suggestions, and optimization suggestion feedback based on data analysis. The optimization suggestion unit analyzes system and model data to provide recommendations for optimizing system and model performance. The optimization suggestion unit can also translate optimization suggestions into specific action plans to guide operations and maintenance personnel in implementing optimization measures. The optimization suggestion unit can also collect and analyze the effectiveness of optimization measures to continuously improve and optimize system performance.

[0148] The equipment maintenance system applied to the pan-semiconductor field in the embodiment of the present application can be deployed in an electronic device, or it can be a component in the electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or it can be other devices other than a terminal. For example, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, a car-mounted electronic device, a mobile Internet device (Mobile Internet Device, MID), an augmented reality (augmented reality, AR) / virtual reality (virtual reality, VR) device, a robot, a wearable device, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a netbook or a personal digital assistant (personal digital assistant, PDA), etc. It can also be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., and the embodiment of the present application does not make specific limitations.

[0149] The device maintenance system for the pan-semiconductor field in the embodiments of the present application can be deployed on a device having an operating system. The operating system can be a Microsoft (Windows) operating system, an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiments of the present application.

[0150] In some embodiments, as Figure 4 As shown, an embodiment of the present application also provides an electronic device 400, including a processor 401, a memory 402, and a computer program stored in the memory 402 and executable on the processor 401. When the program is executed by the processor 401, the above-mentioned embodiment of the equipment maintenance system applied to the pan-semiconductor field is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0151] It should be noted that the electronic device in the embodiments of the present application includes the mobile electronic device and the non-mobile electronic device described above.

[0152] The embodiments of the present application further provide a non-transitory computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the above-mentioned device maintenance system applied to the field of general semiconductor and achieve the same technical effects. To avoid repetition, details are not described herein.

[0153] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes a computer readable storage medium, such as a computer readable only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0154] The embodiments of the present application further provide a computer program product, which includes a computer program. The computer program is executed by a processor to implement the above-mentioned device maintenance system applied to the field of general semiconductor.

[0155] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes a computer readable storage medium, such as a computer readable only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0156] The embodiments of the present application further provide a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is configured to run a program or an instruction to implement the above-mentioned device maintenance system applied to the field of general semiconductor and achieve the same technical effects. To avoid repetition, details are not described herein.

[0157] It should be understood that the chip mentioned in the embodiments of the present application can also be referred to as a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0158] It should be noted that, in the present document, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises a", "comprising", or "comprises" does not, without more constraints, preclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element. Additionally, it should be noted that the terms "one embodiment", "some embodiments", "certain embodiments", "certain examples", or "some examples" as used in the present document are intended to refer to one or more embodiments or examples that do not necessarily have to cover all embodiments or examples of the present application. In other words, use of the above terms does not necessarily refer to the same embodiment or example. Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0159] From the above description of the embodiments, it is apparent that the above-described method of the embodiments can be realized by means of software and general-purpose hardware platforms, of course, but in many cases, the former is a better embodiment. Based on such an understanding, the technical solutions of the present application can be embodied in the form of a computer software product, which is stored in a storage medium (such as a ROM / RAM, a magnetic disk, or an optical disk) and includes a plurality of instructions for causing a terminal (which can be a mobile phone, a computer, a server, or a network device) to perform the methods described in the various embodiments of the present application.

[0160] The embodiments of the present application are described above in conjunction with the drawings, but the present application is not limited to the above-described specific embodiments, which are merely illustrative rather than restrictive, and a person of ordinary skill in the art can make many forms without departing from the scope of the present application and the scope of protection of the claims.

[0161] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "certain embodiments", "an example", "a specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in any one or more embodiments or examples.

[0162] While the embodiments of the application have been shown and described, it is to be understood that the embodiments can be varied, modified, substituted and changed by those skilled in the art without departing from the principles and spirit of the application, the scope of which is defined by the claims and their equivalents.

Claims

1. An equipment maintenance system applied in the pan-semiconductor field, characterized in that: include: Data module, used to obtain target operation data of target device; A large model module, configured to obtain a fault diagnosis result of the target device based on the target operation data based on the large language model; The maintenance warning module is used to process the target operation data based on the remaining service life prediction model, the overall equipment efficiency improvement model and the health assessment model, and generate equipment optimization suggestions for the target equipment based on the processing results and the fault diagnosis results.

2. The equipment maintenance system applied to the pan-semiconductor field according to claim 1, characterized in that: The system also includes a model management module for performing performance evaluation on the remaining service life prediction model, the overall equipment efficiency improvement model, the health assessment model and the large language model, and generating a model evaluation report.

3. The equipment maintenance system applied to the pan-semiconductor field according to claim 1, characterized in that: The system further includes a knowledge base module, wherein the knowledge base module includes an enhanced knowledge graph of the pan-semiconductor field; The large model module is specifically used to call the knowledge graph to perform a thought chain-based retrieval based on the large language model and the target question when obtaining the target question input by the user, so as to obtain the fault diagnosis result.

4. The equipment maintenance system applied to the pan-semiconductor field according to claim 3, characterized in that: The knowledge base module includes an updating unit; the updating unit is used to periodically update the knowledge graph and target response format.

5. The equipment maintenance system applied to the pan-semiconductor field according to claim 1, characterized in that: The large model module includes: The call management unit is configured to determine a target order based on attributes of each call request to the large language model, and call the large language model to respond to each call request based on the target order.

6. The equipment maintenance system applied to the pan-semiconductor field according to claim 1, characterized in that: The large model module also includes a risk assessment unit; the risk assessment unit is used to perform risk assessment on the target problem input by the user.

7. The equipment maintenance system applied to the pan-semiconductor field according to claim 2, characterized in that: The model management module includes: The effect evaluation unit is used to evaluate the usage and effect of the remaining service life prediction model, the overall equipment efficiency improvement model and the health assessment model.

8. The equipment maintenance system applied to the pan-semiconductor field according to claim 2, characterized in that: The model management module includes: The risk assessment unit is configured to assess the usage of the large language model and the risk of calling the large language model.

9. The equipment maintenance system applied to the pan-semiconductor field according to claim 1, characterized in that: The system further comprises: The machine learning model module is used to obtain the remaining service life prediction model, the overall equipment efficiency improvement model, the health assessment model and the large language model through machine learning.

10. The equipment maintenance system applied to the pan-semiconductor field according to any one of claims 1 to 9, characterized in that: The system further comprises: User interaction terminal, used to interact with users.

Images