http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Smoke Particle-Source Prediction Model Based on Multiple Optical Wavelengths Using Deep Learning
Yusun Ahn(Yusun Ahn),Kyuwon Han(Kyuwon Han),Hoesung Yang(Hoesung Yang),Soocheol Kim(Soocheol Kim),Jin Hwa Ryu(Jin Hwa Ryu),Kangbok Lee(Kangbok Lee) 한국화재소방학회 2023 International Journal of Fire Science and Engineer Vol.37 No.2
Recently, installing smoke detectors has become crucial owing to the risk of fatal human damage that may be caused by inhaling smoke during a fire. Smoke detectors have been reported as highly efficient in detecting smoke particles from fire; however, they may generate false alarms because of their limitation in distinguishing the fire smoke from the smoke generated by daily activities. Despite the frequent occurrence of these false alarms, research on predicting the types of sources through smoke particles remains insufficient. This study involved the development process of an intelligent smoke detector for false alarm reduction that aims to predict the occurrence and type of fire and the evaluation of its performance using the light-scattering characteristics for fire/non-fire sources. First, a previous experimental dataset of fire-related conditions was collected from three fire sources and three non-fire sources to train the model with the light-scattering characteristics of the smoke generated from each source. In addition, to reduce the computing power, data preprocessing was performed on the collected dataset using the median and RobustScaler. Finally, we evaluated the prediction performance of the three deep learning models using three networks: RNN, LSTM, and CNN-LSTM. As a result, we confirmed that the scattering intensity of smoke particles has unique characteristics for each source. When the data preprocessing and prediction models were applied, all three models achieved an accuracy of 0.90 or higher. However, some errors occurred that appeared at similar scattering intensities. The proposed method differs from existing methods in that it presents the possibility of predicting fire and non-fire sources and can be used as an alternative for improving false alarms in the future.
초단기 기상 예보 데이터 기반 일사 계산을 통한 단기간 미래 건물 전력 소비량 예측
안유선(Yusun Ahn),이용준(Yong-jun Lee),오은주(Eun Joo Oh),김병선(Byungseon sean Kim) 대한설비공학회 2021 설비공학 논문집 Vol.33 No.3
With recent development of new and renewable systems and data processing technology, attention is being focused on the development of microgrid technology. One of key technologies of energy sharing is the prediction of power consumption of buildings for promoting economic and efficient energy use by providing decision making when sharing energy. Currently, studies that predict actual building power consumption by applying changed ultra-short weather forecast data from the Korea Meteorological Administration are insufficient. Studies that calculate and predict solar radiation without additional equipment are also insufficient. This study aims to predict power consumption of various buildings based on only meteorological data provided by the Korea Meteorological Administration. It is meaningful in that no additional devices are installed except for the power consumption data. Hourly solar radiation was derived using calculation formula. Accuracy was derived by applying meteorological data and calculated insolation as input data for predicting building power consumption. Solar radiation was derived with CVRMSE of 27.61%, MBE of -10.62%, and RMSE of 53.41 W/㎡ regarding the accuracy of calculated and measured data. Results of predicting building power consumption by applying input data to actual four buildings are presented by THE ASHRAE Guideline. They were found to fit the accuracy criterion.
전이 학습 기반 고정된 레이어 수에 따른 건물 전력 소비량 예측 정확도 검증
안유선(Ahn Yusun),진경일(Chin Kyung-Il),이용준(Lee Yong-Jun) 한국태양에너지학회 2020 한국태양에너지학회 논문집 Vol.40 No.6
Owing to recent advances in data-driven technologies, prediction models can now be utilized in various fields such as energy consumption prediction. Among them, transfer learning, which enables prediction using data collected for a short term, is actively gaining ground. In the field of building energy, attention is focused on predicting the power consumption of buildings with the development of smart grids and micro-grids. The purpose of this study is to implement a machine learning (ML) method using the transfer learning algorithm that enables the prediction of energy consumption of the buildings using data collected for a short-term period. This study also validates the accuracy of the predicted energy consumption by adjusting the number of frozen layers based on the input parameters used in the ML method. A total of 5 layers were set, and the frozen layers were increased from 0 to 5. In addition, to validate the prediction rate of the transfer learning algorithm, only the existing ML model was used. As a result, when 0 through 3 layers were fixed, the accuracy criterion tended to be met. In contrast, when the 4-5 layers were reflected, the accuracy criterion was not met. It should also be noted that the model without using transfer learning showed poor accuracy.
건물 내 화재 발생 시 사물 인터넷과 강화 학습을 활용한 실시간 안전 대피 경로 방안 개발
안유선,최하늘,Ahn, Yusun,Choi, Haneul 한국안전학회 2022 한국안전학회지 Vol.37 No.2
Human casualties from fires are increasing worldwide. The majority of human deaths occur during the evacuation process, as occupants panic and are unaware of the location of the fire and evacuation routes. Using an Internet of Things (IoT) sensor and reinforcement learning, we propose a method to find the safest evacuation route by considering the fire location, flame speed, occupant position, and walking conditions. The first step is detecting the fire with IoT-based devices. The second step is identifying the occupant's position via a beacon connected to the occupant's mobile phone. In the third step, the collected information, flame speed, and walking conditions are input into the reinforcement learning model to derive the optimal evacuation route. This study makes it possible to provide the safest evacuation route for individual occupants in real time. This study is expected to reduce human casualties caused by fires.
광전식 연기 감지기의 반응 특성 분석을 위한 다중 산란 기반 광학 시뮬레이션 연구
김수철(Soocheol Kim),양회성(Hoesung Yang),한규원(Kyuwon Han),안유선(Yusun Ahn),홍상기(Sang Gi Hong),이강복(Kangbok Lee) 한국화재소방학회 2023 한국화재소방학회논문지 Vol.37 No.6
광전식 연기 감지기의 응답 특성을 관찰하기 위해서는 화재 및 비화재 시험을 필요로 하지만 감지기의 구조와 구성 변화에 따른 모든 실험을 반복 수행하는 것은 막대한 시간과 비용으로 인해 한계가 있다. 화재의 특성 및 확산, 연기 입자의 광산란 특성을 관찰하는 시뮬레이션 연구는 많이 수행되고 있지만 연기 감지기의 응답 특성을 관찰하는 시뮬레이션 연구는 소개된 것이 없다. 본 연구는 광전식 연기 감지기의 반응 특성 분석을 위한 광학 시뮬레이션 연구이며, 다중 산란을 기반으로 매질 내 광자가 전파하면서 산란 및 흡수되는 현상을 살펴보았다. 감광 방식과 광산란 방식의 연기 감지기와 유사한 환경을 구현하여 흰 연기, 검은 연기, 수증기, 먼지에 대해 다양한 농도에서 광신호를 관찰하였다. 화재나 비화재 시험 없이 광학 시뮬레이션을 활용하여 다양한 조건에서 광전식 연기 감지기의 응답특성을 분석해볼 수 있을 것이며, 광전식 연기 감지기의 구성이나 구조도 반영하여 새로운 연기 감지기 개발에 적극적으로 활용될 수 있을 것으로 기대한다. Fire and nonfire experiments are required to observe the response characteristics of photoelectric smoke detectors; however, repeating all experiments by changing the structure and configuration of photoelectric smoke detectors is time-consuming and expensive. No simulations have been performed to evaluate the response characteristics of photoelectric smoke detectors. In this study, multiple-scattering-based optical simulations were performed to investigate the response characteristics of a photoelectric smoke detector. The radiant energy of light propagating in various media (white smoke, black smoke, water vapor, and dust) with different particle sizes and refractive indices was measured using a photodetector as a function of concentration. Multiple-scattering-based optical simulations that can analyze the response characteristics of photoelectric smoke detectors without fire or nonfire experiments are expected to be actively used for developing new smoke detectors. In addition, the results for the nonspherical soot particles can be analyzed via multiple-scattering-based optical simulations using the discrete dipole approximation method.