이 논문에서는 강화학습 기반 제어기와 전통적인 제어기를 동일한 조건에서 비교함으로써 구조 진동 제어 문제에서 강화학습 제 어기의 성능 특성과 한계를 규명하는 것을 목적으로 한다. 가장 단순한 비선형 제어로서 단자유도 가변 강성 시스템을 대상으로 심층 결정적 정책 경사(DDPG) 기반의 강화 학습 제어기를 설계하고, bang-bang 제어 및 제한 최적 제어와의 성능 비교를 수행하였다. 자유 진동 및 El Centro 지진 가속도에 의한 강제 진동 조건에서 공칭 성능과 센서 잡음이 존재하는 경우의 강인 성능을 분석하였다. 그 결 과, 강화학습 제어기는 자유 진동 조건에서 우수한 강인 성능을 보였으나, 강제 진동 제어에서는 기존 제어기를 일관되게 상회하지는 못하였다. 이 연구는 동일한 보상 함수와 시스템 조건 하에서 강화학습 기반 진동 제어의 실질적 기여와 적용상의 한계를 기초적으로 제시하였다.

Purpose: Since the COVID-19 pandemic, virtual simulation practice has been increasingly activated as an alternative to clinical practice in nursing colleges. This study aimed to provide basic data by confirming changes in self-efficacy and nursing knowledge in the virtual simulations of nursing students, and identifying virtual presence, virtual patient learning system evaluation (VPLSE), and practical satisfaction. Methods: This was a single-group pre-post quasi-experimental study. The subjects were 28 third-grade nursing students. Results: Self-efficacy and nursing knowledge increased significantly (p<.001). Virtual presence had a significant positive correlation with VPLSE) (p=.002) and practice satisfaction (p=.011). There was also a significant positive correlation between virtual simulation learning evaluation and practice satisfaction (p<.001). Conclusion: Based on these results, virtual simulation practice can be used with clinical practice as an educational method to improve nursing students' self-efficacy and nursing knowledge in nursing education. Virtual presence was confirmed as a significant variable to improve practice satisfaction and VPLSE. It is necessary to develop a virtual simulation program that can improve virtual presence through collaboration with virtual reality technology experts.

Three CNN (Convolutional Neural Network) models of GoogLeNet, VGGNet, and Alexnet were evaluated to select the best deep learning based image analysis mothod that can detect pavement distresses of pothole, spalling, and punchout on expressway. Education data was obtained using pavement surface images of 11,056km length taken by Gopro camera equipped with an expressway patrol car. Also, deep learning framework of Caffe developed by Berkeley Vision and Learning Center was evaluated to use the three CNN models with other frameworks of Tensorflow developed by Google, and CNTK developed by Microsoft. After determing the optimal CNN model applicable for the distress detection, the analyzed images and corresponding GPS locations, distress sizes (greater than distress length of 150mm), required repair material quantities are trasmitted to local maintenance office using LTE wireless communication system through ICT center in Korea Expressway Corporation. It was found out that the GoogLeNet, AlexNet, and VGG-16 models coupled with the Caffe framework can detect pavement distresses by accuracy of 93%, 86%, and 72%, respectively. In addition to four distress image groups of cracking, spalling, pothole, and punchout, 22 different image groups of lane marking, grooving, patching area, joint, and so on were finally classified to improve the distress detection rate.

  This study aims to design and implement a learning evaluation system using .NET which is developed by Microsoft. .NET technology supports higher processing speed than ASP technology. The learning evaluation system is based on the web, consists of admini