https://scholar.dgist.ac.kr/handle/20.500.11750/10150 Sat, 04 Apr 2026 09:03:25 GMT 2026-04-04T09:03:25Z https://scholar.dgist.ac.kr/handle/20.500.11750/60180 Title: Long Short-Term Memory Network-Based H∞ Synchronization Control and Anomaly Detection for Cyber-Physical Systems Author(s): Kwon, Hyoeun; Lee, Suwoong; Kwon, Wookyong; Lim, Yongseob; Jin, Yongsik Abstract: In the synchronization of cyber-physical systems (CPSs), modeling the nonlinear dynamics of physical plants is a challenging task. To address this challenge, we propose a novel H∞ controller design method that leverages a data-driven approach to robustly synchronize CPSs and ensure their stability. In the proposed approach, the input-output relationship of the physical system is learned using long short-term memory (LSTM) networks to approximate the unknown dynamics of CPSs. Furthermore, we exploit an effective control scheme for trained LSTM networks to effectively handle the nonlinearity of activation functions. To ensure stability and performance in the convergence of synchronization error, a controller design criterion is derived for the trained LSTM network in terms of linear matrix inequalities, and the controller gain is computed using convex optimization techniques. In addition, we present an anomaly detection algorithm using the proposed method, which can synchronize CPSs and detect abnormal signals without requiring any prior physical model information. Consequently, the stability of the synchronization control system can be ensured, enabling its application to anomaly detection. Finally, the effectiveness of the proposed method is validated through an experiment on a motor control system even in abnormal operating conditions. Mon, 06 Oct 2025 15:00:00 GMT https://scholar.dgist.ac.kr/handle/20.500.11750/60180 2025-10-06T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/60081 Title: 뿜칠 로봇의 강화 학습 기반 작업 초기 위치 인식 방법 Author(s): 이승열 Abstract: Recently, there has been a lot of interest concerning automation and robotics in hazardous environments including construction sites. Spray robots paint walls or ceilings on construction sites instead of workers. The main functions of the spray robot include initial spray position recognition, robot path generation, robot path following, and spray quality inspection. This paper describes a method to automatically recognize the initial spray position using AI. In particular, this paper discusses a reinforcement learning method for a spray robot to recognize the corner where the ceiling and adjacent wall meet as the initial spray position when spraying a wall. Thu, 11 Dec 2025 15:00:00 GMT https://scholar.dgist.ac.kr/handle/20.500.11750/60081 2025-12-11T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/60080 Title: 뿜칠 로봇 프레임 경량화 및 강성 확보를 위한 구조해석 Author(s): 이동훈; 이용석; 이승열 Abstract: In the development of a spray robot, structural deformation occurred when the robot was mounted on an aerial lift platform. Despite applying only moderate load during attachment, the original frame showed permanent deformation, leading to sensor misalignments such as errors in the zero-point calibration of cameras and laser pointers. To address this issue, finite element method(FEM) simulations were conducted with the objective of improving stiffness while reducing the weight of the robot frame. The analysis was carried out using beam elements in ANSYS SpaceClaim, where sectional properties such as cross-sectional area and moment of inertia were automatically calculated to ensure accurate stiffness evaluation. This approach significantly reduced meshing time while maintaining high accuracy. Through iterative simulations and design optimization, the improved frame design demonstrated enhanced stiffness and weight reduction. Experimental validation confirmed that the optimized frame prevented permanent deformation under operating conditions and eliminated sensor alignment errors during robot operation. This study shows that FEM-based structural optimization, especially with beam element modeling, is an effective and efficient method for ensuring both lightweight and stiffness in robotic systems for construction applications. Thu, 11 Dec 2025 15:00:00 GMT https://scholar.dgist.ac.kr/handle/20.500.11750/60080 2025-12-11T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/60078 Title: PAttn과 LightTS를 결합해 전류 신호를 이용한모터 이상 탐지 딥러닝 기반 모델 Author(s): 김해영; 구본근; 반재필; Koo, Gyogwon Abstract: 모터 이상 탐지는 산업안전과 밀접하게 연관 되어 있어 매우 중요하다. 본 연구에서는 모터의 고장을 효과적으로 감지하기 위한 딥러닝 기반 이상 탐지 모델을 제안하였다. 모델은 전역적 의존성을 학습하는 attention 기반 PAttn 모듈과 지역적 및 장기적 패턴의 상호작용을 학습하는 LightTS 모듈을 결합하여 구성하였다. 또한 두 모듈의 출력을 각각 최적화하기 위해 가중 합 기반 평균 제곱 오차 손실 함수를 적용하여, 두 모듈의 손실을 동시에 최소화하고 균형적으로 학습될 수 있도록 설계하였다. 제안된 모델은 area under the receiver operating characteristic curve (AUC) 0.8873의 성능을 보여 비교 모델들보다 우수한 결과를 보여준다. 이러한 결과를 통해 제안된 모델의 구조가 전류 신호 기반 모터 이상 탐지에서 시계열 패턴을 효과적으로 학습하고, 모델의 안정성을 향상시킴을 확인하였다. Wed, 26 Nov 2025 15:00:00 GMT https://scholar.dgist.ac.kr/handle/20.500.11750/60078 2025-11-26T15:00:00Z