Search

KR-20260065378-A - METHOD AND SYSTEM FOR LLM-BASED REAL-TIME MANUFACTURING DATA INTEGRATION ANALYSIS AND DEFECT PREDICTION

KR20260065378AKR 20260065378 AKR20260065378 AKR 20260065378AKR-20260065378-A

Abstract

A method and system for real-time manufacturing data integration analysis and defect prediction based on LLM are disclosed. Through a Large Language Model (LM), insights can be derived by analyzing manufacturing-related data, including incoming inspection, material requirement planning (MRP), production processes, equipment status, and quality control.

Inventors

  • 양진홍

Assignees

  • 인제대학교 산학협력단

Dates

Publication Date
20260508
Application Date
20241101

Claims (15)

  1. At least one processor implemented to execute readable instructions on a computer device Includes, The above-mentioned at least one processor is, Deriving insights by analyzing manufacturing-related data, including incoming inspection, material requirement planning (MRP), production processes, equipment status, and quality control, through a large language model (LLM). A computer device characterized by
  2. In paragraph 1, The above-mentioned at least one processor is, Performing data cleansing, normalization, and standardization on source data corresponding to manufacturing-related data A computer device characterized by
  3. In paragraph 1, The above-mentioned at least one processor is, Integrated storage of structured and unstructured data through a data lake architecture A computer device characterized by
  4. In paragraph 1, The above-mentioned at least one processor is, Analyzing multimodal data as manufacturing-related data using an LLM specialized in the manufacturing domain A computer device characterized by
  5. In paragraph 1, The above-mentioned at least one processor is, Contextualizing analysis results through LLM and transforming them into actionable insights A computer device characterized by
  6. In paragraph 1, The above-mentioned at least one processor is, Visualizing analysis results and insights on manufacturing-related data and providing them through a web-based dashboard A computer device characterized by
  7. In paragraph 1, The above-mentioned at least one processor is, Normalize the data input as manufacturing-related data, and Performing validation and preprocessing of input data A computer device characterized by
  8. In paragraph 1, The above-mentioned at least one processor is, For text data, data encoding is performed using a BERT (bidirectional encoder representations from transformers) based encoder, and For numerical data, data encoding is performed using an MLP (multi-layer perceptron) or a 1D CNN (convolutional neural network), and For image data, performing data encoding using a vision transformer or CNN. A computer device characterized by
  9. In paragraph 1, The above-mentioned at least one processor is, Generating an integrated data representation of manufacturing-related data through learning the relationships between modalities A computer device characterized by
  10. In paragraph 1, The above-mentioned at least one processor is, Combining rule-based reasoning and LLM-based reasoning using a manufacturing expertise database A computer device characterized by
  11. In paragraph 1, The above-mentioned at least one processor is, Using explainable AI technology to provide the interpretation and summary of LLM-based inference results as a basis for decision-making A computer device characterized by
  12. In paragraph 1, The above-mentioned at least one processor is, Performing model retraining and parameter optimization for LLM based on new training data and feedback A computer device characterized by
  13. In paragraph 1, The above-mentioned at least one processor is, In LLM training, performing model fine-tuning for a specific task based on the performance metrics of the current model. A computer device characterized by
  14. In paragraph 1, The above-mentioned at least one processor is, Performing hyperparameter tuning through grid search and Bayesian optimization in LLM training A computer device characterized by
  15. In a method for analyzing LLM-based manufacturing data of a computer device comprising at least one processor, A step of analyzing manufacturing-related data, including incoming inspection, material requirements planning (MRP), production process, equipment status, and quality control, through LLM by at least one processor; and A step of converting LLM-based analysis results into actionable insights by the above-mentioned at least one processor and providing the LLM-based analysis results and insights as visualized data. LLM-based manufacturing data analysis method including

Description

Method and System for LLM-Based Real-Time Manufacturing Data Integration Analysis and Defect Prediction The following description concerns a technology that predicts defects by analyzing manufacturing data. The manufacturing industry is an industry that produces products using raw materials, undergoing a continuous process in which various processes are performed sequentially, and the outputs of each process are mixed with one another or the state of the output of a specific process changes before being supplied to subsequent processes. With the recent advancement of IT technology, Smart Factory technology is emerging. This technology utilizes artificial intelligence and big data to analyze desired products, plan and design customized goods, and automatically produce them through optimized processes using IoT and automated robots, thereby controlling product receipts and orders in real time. A Smart Factory is an intelligent production plant that applies ICT-based digital automation solutions to production processes such as design and development, manufacturing, and distribution and logistics, encompassing various technologies capable of improving productivity, quality, and customer satisfaction. A Manufacturing Execution System (MES) refers to a system that manages all production activities in a smart factory, from the commencement of production based on orders to the quality inspection, shipment, and sale of finished products. It is an integrated production management system that establishes a high-quality, profit-oriented production system by collecting, aggregating, analyzing, and monitoring various information from the production site—such as production performance, worker activity, equipment operation, and product quality data—in real time, and controlling production processes. In other words, by digitizing all processes related to production—from order placement to production planning, design, and manufacturing—as well as material status, shipment, and quality control—into IoT-based digital data and managing them accurately in real time, it improves productivity and can provide objective indicators of improvement to meet customer demands for enhanced quality. As an example of smart factory technology, Korean Published Patent Application No. 10-2018-0104919 discloses a technology that supports global and automatic interoperability between information systems (MES, ERP) and field systems (Sensors, Actuators, Field Devices) used in smart factories. FIG. 1 is a block diagram illustrating an example of the internal configuration of a computer device in an embodiment of the present invention. FIG. 2 illustrates an example of a manufacturing data analysis environment in an embodiment of the present invention. FIG. 3 illustrates the overall architecture of an LLM-based manufacturing data analysis system in one embodiment of the present invention. Figure 4 illustrates the detailed configuration of an LLM-based analysis engine and a model learning optimization module in an embodiment of the present invention. FIGS. 5 and 6 illustrate the detailed operation of an LLM-based analysis engine in an embodiment of the present invention. FIGS. 7 and 8 illustrate the detailed operation of a model learning optimization module in an embodiment of the present invention. Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. Embodiments of the present invention relate to a technology for predicting defects by analyzing manufacturing data. Embodiments including those specifically disclosed in this specification can identify the causes of defects in real time by utilizing LLM to comprehensively analyze various manufacturing-related data, such as incoming inspection, material requirements planning (MRP), production processes, equipment status, and quality control. An LLM-based manufacturing data analysis system according to embodiments of the present invention may be implemented by at least one computer device, and an LLM-based manufacturing data analysis method according to embodiments of the present invention may be performed through at least one computer device included in the LLM-based manufacturing data analysis system. At this time, a computer program according to one embodiment of the present invention may be installed and run on the computer device, and the computer device may perform an LLM-based manufacturing data analysis method according to embodiments of the present invention under the control of the run computer program. The above-described computer program may be stored on a computer-readable recording medium to be combined with the computer device to execute the LLM-based manufacturing data analysis method on the computer. FIG. 1 is a block diagram illustrating an example of a computer device according to an embodiment of the present invention. For example, an LLM-based manufacturing data analysis system according to embodiments of the present inv