http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Dynamic Mode Decomposition으로 유도된 차량 동역학 특성 분석
박채훈(Chaehun Park),정철민(Cheolmin Jeong),손영섭(Young Seop Son),안상각(An Sang Gak),강창묵(Chang Mook Kang) 한국자동차공학회 2022 한국자동차공학회 부문종합 학술대회 Vol.2022 No.6
Designing a control is the process of deriving the system’s dynamic equation and finding a way to stabilize the derive system model. However, the derivation of system model is complicated, and depending on the characteristics of the system, it may be difficult to design the controller. In this paper, we propose a data-driven model derivation method DMD(Dynamic Mode Decomposition) that simplifies model derivation process and has similarity to the system. DMD can be used for complicated dynamic system and has advantages in designing controllers because the derived model is linear. However, in order to take these advantages in the control design, it is first show the similarity between the DMD derived model and the system. Therefore, we used the vehicle dynamics simulation CarSim to derive various models and analyze how to derive validate models.
박채훈(Chaehun Park),정철민(Cheolmin Jeong),이화수(Hwa Su Lee),안대용(Dae-Young An),배문규(Mun Kyu Bae),강창묵(Chang Mook Kang) 대한전기학회 2021 대한전기학회 학술대회 논문집 Vol.2021 No.10
이 논문은 드론 고장 중 프로펠러 파손을 연구했으며, 프로펠러 파손 시 발생하는 관성 모멘트 변화를 DC 모터 동역학과 연관 지어 고장을 모사하였다. 추가적으로 다중 필터 모델기반 상태 추정 알고리즘인 Interacting Multiple Model(IMM)로 시스템의 정상 모드와 고장 모드 분류를 수행하였다.
정철민(Cheolmin Jeong),박채훈(Chaehun Park),안대용(Dae-Young An),강창묵(Chang Mook Kang) 한국자동차공학회 2021 한국자동차공학회 부문종합 학술대회 Vol.2021 No.6
Electric power steering (EPS) system is used to reduce driver burden and increase convenience. In the case of an EPS system of autonomous vehicles, more efficient driving is possible by controlling the system to obtain the desired steering wheel angle. However, for reliable control, there must be a fail-safety function for the failure. To ensure control reliability and stability, the Interacting Multiple Model (IMM) is applied as an algorithm that can diagnose mechanical and electrical faults. IMM consists of several Kalman filter models consisting of parallel structures, each of which updates the probability values according to the Markov chain rule via interaction. The Kalman filter of each model not only allows the state to be estimated, but also finds the best model for the current state, and the failure condition can be diagnosed quickly because it is estimated according to mode probability. Therefore, stable failure factor blocking and control performance improvement can also be expected when diagnosing failures with IMM.