박현우 교수
Hyunwoo Park
서울대학교 데이터사이언스대학원 · 공학
연구실 소개
박현우 교수의 연구실은 선형 탄성체의 시스템 식별을 위한 정규화 기법과 물리 기반 신경망을 활용한 구조물의 시간에 따른 거동 예측을 핵심으로 합니다. 특히 Tikhonov 정규화 기법의 최적화 요인 결정, 1-norm 정규화를 통한 비연속적 구조물 매개변수 식별, 그리고 프리스트레스트 콘크리트 빔의 초기 거동 예측에 응용된 물리 기반 신경망(PINN) 기술을 개발하고 있습니다. 또한, 5G 통신용 저손실 재료로 활용 가능한 YSZ 나노입자 합성 및 유기-무기 프레임워크를 활용한 고성능 촉매 개발 등 재료 및 공정 기반 연구도 병행하고 있습니다.
연구 현황
연구 성과 추이
표시된 성과는 수집된 데이터 기준으로 산출되며, 일부 차이가 있을 수 있습니다.
주요 논문
15Abstract This paper presents a geometric mean scheme (GMS) to determine an optimal regularization factor for Tikhonov regularization technique in the system identification problems of linear elastic continua. The characteristics of non‐linear inverse problems and the role of the regularization are investigated by the singular value decomposition of a sensitivity matrix of responses. It is shown that the regularization results in a solution of a generalized average between the a priori estimates
Abstract This paper presents a new class of regularization functions and the associated regularization scheme for structural system identification. In particular, 1‐norm regularization functions are investigated to overcome the smearing effect of 2‐norm regularization functions for the identification of discontinuous system parameters of structures. The truncated singular value decomposition is employed to filter out noise‐polluted solution components and to impose the 1‐norm regularization func
An economically efficient one-step lactide synthesis process has been developed by using a low-price commercialized catalyst of SiO 2 /Al 2 O 3 in a packed-bed reactor. Under the optimized conditions of preheater temperature 160 °C, reactor temperature 240 °C, and WHSV 3 h –1, 90% lactic acid conversion and 99% lactide selectivity were obtained. Filling of glass beads underneath the catalyst bed proved to be quite effective in improving reaction performance. The catalyst stably maintained its ac
This paper proposes a physics-informed neural network (PINN) for predicting the early-age time-dependent behaviors of prestressed concrete beams. The PINN utilizes deep neural networks to learn the time-dependent coupling among the effective prestress force and the several factors that affect the time-dependent behavior of the beam, such as concrete creep and shrinkage, tendon relaxation, and changes in concrete elastic modulus. Unlike traditional numerical algorithms such as the finite differen
Yttria-stabilized zirconia (YSZ) nanospheres were synthesized by calcination at 900 °C after the adsorption of Y3+ ions into the pores of a zirconium-based metal–organic framework (MOF). The synthesized 3YSZ (zirconia doped with 3 mol% Y2O3), 8YSZ (8 mol% Y2O3), and 30YSZ (30 mol% Y2O3) nanospheres were found to exhibit uniform sizes and shapes. Complex permittivity and complex permeability were carried out in K-band (i.e., 18–26.5 GHz) to determine their suitability for use as low-k materials i
This article analytically investigates the electromechanical admittance of piezoelectric transducers collocated on a finite beam from the perspective of wave propagation. First, the analytic solutions are derived for flexural waves induced by piezoelectric transducers collocated on an infinite beam. Then, the concept of a flexural wave group is used to express the analytic solutions for flexural waves reverberating on a finite beam. Detailed formulation is presented to describe the evolution of
대표 연구 분야
박현우 교수의 연구를 Nubint에서 더 깊이 살펴보세요
이 연구실의 논문을 앱에서 열어 AI와 함께 읽고, 핵심을 요약하고, 내 글에 인용하세요.