모바일 메뉴 닫기
 
제목
[BK세미나] 7/2(수) Erhan Budak(Sabanci University) "Application of Physics Informed Machine Learning Approach for Modeling,
작성일
2025.06.25
작성자
기계공학부
게시글 내용

기계공학부 구성원들의 많은 관심과 참여 부탁드립니다.


▣ 주   제: Application of Physics Informed Machine Learning Approach for Modeling, Optimization and Monitoring of Machining Processes

연   사: Erhan Budak 교수

소   속: Sabanci University

일   시: 2025. 7. 2.(수) 11:00

장   소제1공학관 A205호

초   청: 민병권 교수

▣ 초   록

Traditional approaches to machining process R&D heavily rely on experimental data which are time-consuming to obtain for large-scale industrial applications. To overcome these limitations, this study presents a hybrid modeling, monitoring and optimization framework that leverages physics-informed machine learning (PIML).

In the first stage of the framework, analytical models are combined with machine learning algorithms to improve the accuracy of simulations. These enhanced simulations are then used to generate comprehensive datasets reducing the dependency on test data. The tool wear estimation process begins with analytical modeling and is refined using a limited number of experimental results minimizing measurement efforts while ensuring reliable predictions.

The proposed system employs readily available CNC controller data, such as spindle current, torque, velocity, and acceleration. To overcome the challenge of low sampling rates in CNC data, a two-layered machine learning system is introduced. This enables accurate real-time monitoring without the need for external sensors. The proposed intelligent monitoring framework eliminates the limitations of conventional hardware-based systems and offers a scalable, cost-effective solution for real-world machining applications. Demonstrations of the system will highlight its ability to enhance process reliability and optimization in industrial settings.

첨부
20250702_BK21_Erhan Budak(민병권 교수님).jpg