Digital Twin-Based Motor Speed Ripple Control System with AI-Driven Virtual Sensing and Reinforcement Learning
- 문서 유형
- 연구출판물 · 학술대회 발표논문
- 등재
- Scopus
- 저널
- Proceedings of the ACM Symposium on Applied Computingpp.301 · 2026
- 피인용
- Scopus 0
- FWCI (Scopus)
- 0
저자
- 김재석제1저자School of Computer Science and Engineering, School of Computer Science and Engineering, Pusan National University, Busan, Republic of Korea
- 서준호School of Computer Science and Engineering, School of Computer Science and Engineering, Pusan National University, Busan, Republic of Korea
- Park, DongminHS R&D Center, LG Electronics, Changwon, Republic of Korea
- Park, ChansooHS R&D Center, LG Electronics, Changwon, Republic of Korea
- 최윤호교신저자School of Computer Science and Engineering, School of Computer Science and Engineering, Pusan National University, Busan, Republic of Korea
초록
Improving the energy efficiency of motor-driven household appliances often requires increasing the compressor operating frequency. However, this inevitably amplifies speed ripple, leading to noise and vibration that degrade the user experience. As a potential solution, methods that inject compensation currents proportional to the load torque have recently gained attention. Nevertheless, sensorless motor control cannot directly measure load torque, and it is difficult to determine the optimal compensation current under varying operating conditions. Furthermore, issues of cost and stability present additional challenges. To address these limitations, this paper proposes a novel motor speed ripple control system. The proposed system leverages a digital twin environment of the motor, combining virtual sensing for load torque pattern prediction with reinforcement learning-based compensation control. Experimental results demonstrate that our system achieves a 27.5% reduction in speed ripple compared to conventional approaches. In addition, the proposed system predicts load torque with a mean squared error (MSE) of 0.2537, achieving superior performance over existing methods.
키워드
adaptive controldeep reinforcement learningdigital twinmotor speed ripplesensorless control
식별자
- Scopus EID
- 2-s2.0-105042936671



로그인
연구물 목록으로