http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
센서 어레이 시스템과 머신러닝을 이용한 차량 냄새 정량화 연구
김형직,이태희 한국신뢰성학회 2019 신뢰성응용연구 Vol.19 No.3
Purpose: This study presents a method to quantify vehicle odors from interiors of new vehicles and air conditioners of old vehicles using machine learning with a sensor array system we named electronic nose. Methods: We developed a sensor array system (electronic nose) with 17 kinds of sensors that can detect major odors and harmful substances from vehicles. Odors from new vehicles’ interiors and old vehicles’ air conditioners, which are representative odors generated in vehicles, were measured by the sensor array system. K-means clustering algorithm of unsupervised learning and MLP ANN algorithm of supervised learning were applied to analyze odor data. Result: The sensor array based electronic nose and machine learning method were applied to obtain the results of clustering and intensity classification of vehicle odors. Conclusion: Time series data of odors from air conditioner and interiors of new cars were divided into ten species of odor by K-means clustering. Intensity of odor as sensed by human are classified into three levels by applying MLP ANN.
애드혹 네트워크에서 경로 안정성 향상을 위한 라우팅 프로토콜
김형직,최선웅,Kim, Hyungjik,Choi, Sunwoong 한국정보통신학회 2015 한국정보통신학회논문지 Vol.19 No.7
Nodes of Mobile ad-hoc network usually use the energy-limited battery. Balanced energy consumptionis important to maintain path's stability. In this paper, we focus on improving the stability of the routing path in mobile ad-hoc networks. For that purpose, we propose a new routing protocol to find the highest minimum node residual energy path among shortest paths. The largest path of minimum value of the remain energy has a longer life than other paths to improve the reliability to data-transmission. Using ns-3 simulator, we show that the proposed routing protocol can provide more long-life stable routing path than AODV and EA-AODV. 모바일 애드혹 네트워크의 노드는 일반적으로 에너지의 용량이 제한된 배터리를 사용한다. 경로의 안정성을 유지하기 위해 균형 잡힌 에너지 소비가 중요하다. 본 논문에서는 애드혹 네트워크에서 데이터 전송 경로의 안정성을 향상시키는 것을 목표로 한다. 이를 위해 데이터를 전송할 수 있는 최단 전송 경로 중에서 노드 에너지 잔량의 최소값이 가장 큰 경로를 선택하는 새로운 라우팅 프로토콜을 제안한다. 에너지 잔량의 최소값이 가장 큰 경로는 다른 경로보다 상대적으로 긴 수명을 갖게 되어 데이터 전송에 안정성을 향상 시킬 수 있다. NS-3 시뮬레이터를 사용하여 제안하는 라우팅 프로토콜이 AODV와 EA-AODV보다 수명이 긴 안정적인 경로를 제공하는 것을 확인한다.
동태평양 열대해역에서 2009-2010년 침강입자 플럭스의 수직 변화
김형직,조소설,김동선,김경홍,유찬민,Kim, Hyung Jeek,Cho, Sosul,Kim, Dongseon,Kim, Kyeong Hong,Yoo, Chan Min 한국해양과학기술원 2022 Ocean and Polar Research Vol.44 No.3
A sediment trap had been deployed at 1250 m depth in the Eastern Tropical Pacific (ETP) from September 2009 to July 2010, with the aim of understanding the temporal and vertical variability of particle flux. During the monitoring period, total particle flux varied from 12.4 to 101.0 mg m<sup>-2</sup>day<sup>-1</sup>, with the higher fluxes in January-March 2010. Biogenic particle flux varied in phase with the total particle flux. The increase in total particle flux during January-March 2010 was attributed to the enhanced biological production in the surface layer caused by wind-driven mixing in response to the seasonal shifts in the location of the Intertropical convergence zone. The export ratio (e-ratio) was estimated using the particulate organic carbon flux and satellite-derived net primary production data. The estimated e-ratios changed between 0.8% and 2.8% (1.4±0.6% on average). The ratio recorded in the negative phase of Pacific decadal oscillation (PDO) was similar to the previous results obtained from the ETP during the 1992/93 periods in the positive phase of PDO. This suggests that the regime shift of the PDO is not related to the carbon export ratio.
김형직,김종욱,김동선,Michael T. Chandler,손승규 한국해양과학기술원 2018 Ocean science journal Vol.53 No.2
Time-series sediment traps were deployed in the subtropical oligotrophic northwestern Pacific (SONP) at three depths from August to September 2015 to better understand vertical flux of sinking particles. Sinking particles were collected at 5-day intervals over the sediment trap deployment period. The average total mass flux at water depths of 400 m, 690 m, and 1,710 m was 9.1, 4.4, and 4.1 mg m-2day-1, respectively. CaCO3 materials constituted 50 to 70% of sinking particles while in comparison particulate organic carbon (POC) constituted up to 20%. A synchronous variation of total mass flux was observed at the three depths, indicating that calcite-dominated particles sank from 400 to 1,710 m within a 5-day period. POC flux at these water depths was 2.4, 0.38, and 0.31 mg m-2day-1, respectively. Our results indicate low transfer efficiencies of 16% from 400 to 690 m and 13% for the 400 to 1,710 m depth range. The estimated transfer efficiencies were significantly lower than those observed at the K2 station in the northwest Pacific subarctic gyre, presumably because of a prevalence of pico-cyanobacteria in the SONP. Because cyanobacteria have a semi-permeable proteinaceous shell, they are more readily remineralized by bacteria than are siliceous phytoplankton in the northwest Pacific subarctic gyre. Continued surface water warming and expansion of the SONP will likely have a profound impact on ocean acidification in the northwest Pacific, possibly affecting the transfer efficiency of sinking POC to the deep-sea.