• 한국방사선학회논문지
• /
• 제18권1호
• /
• pp.37-44
• /
• 2024

MR은 우수한 연조직 대비와 기능 정보를 보여줄 수 있지만, 방사선치료에서 정확한 선량 계산을 위해서는 CT영상의 전자밀도 정보가 필요하다. 방사선치료(Radiotherapy) 계획 워크플로우에서 MR영상과 CT영상을 융합하기 위해 환자는 일반적으로 MR과 CT영상 방식 모두에서 스캔된다. 최근에 딥러닝기술 덕분에 MR영상에서 딥러닝 기반의 CT영상 생성이 가능해졌다. 이로 인해 CT 스캔 작업을 할 필요가 없게 된다. 본 연구에서는 MR영상으로부터 CycleGan 딥러닝 기반 CT영상생성을 구현했다. T1가중이나 T2가중 중에 한 가지 또는 그 둘다의 MR영상을 가지고 합습한 3가지의 인공지능 CT생성기를 만들었다. 결과에서 우리는 T1가중 MR 영상 기반으로 학습한 생성기가 T1가중 MR영상이 입력될 때 다른 CT생성기보다 더 나은 결과를 생성할 수 있음을 발견했다. 반면, T2가중 MR영상 기반 CT생성기는 T2가중 MR영상을 입력 받을 때, 다른 시퀀스기반 CT생성기보다 더 나은 결과를 생성할 수 있습니다. MR영상을 기반으로 한 CT생성기는 곧 임상현장에 적용될 수 있는 기술이다. 특정 시퀀스 MR영상으로 학습한 머신러닝 CT생성기는 다른 시퀀스 MR영상으로 학습한 생성기보다 더 그 특정 시퀀스와 같은 MR영상을 입력받을 때 더 나은 CT영상을 생성할 수 있음을 보여주었다.

• Journal of Communications and Networks
• /
• 제18권2호
• /
• pp.227-237
• /
• 2016

Data collection is a key function for wireless sensor networks. There has been numerous data collection scheduling algorithms, but they fail to consider the deep and complex relationship among network lifetime, sink aggregated information and sample cycle for wireless sensor networks. This paper gives the upper bound on the sample period under the given network topology. An optimal schedule algorithm focusing on aggregated information named OSFAI is proposed. In the schedule algorithm, the nodes in hotspots would hold on transmission and accumulate their data before sending them to sink at once. This could realize the dual goals of improving the network lifetime and increasing the amount of information aggregated to sink. We formulate the optimization problem as to achieve trade-off among sample cycle, sink aggregated information and network lifetime by controlling the sample cycle. The results of simulation on the random generated wireless sensor networks show that when choosing the optimized sample cycle, the sink aggregated information quantity can be increased by 30.5%, and the network lifetime can be increased by 27.78%.

• 방사성폐기물학회지
• /
• 제19권4호
• /
• pp.517-532
• /
• 2021

Safe geological disposal of spent nuclear fuel (SNF) requires knowledge of the deep hydrochemical characteristics of the repository site. Here, we conducted a set of deep hydrochemical investigations using a 750-m borehole drilled in a model granite system in Wonju, South Korea. A closed investigation system consisting of a double-packer, Waterra pump, flow cell, and water-quality measurement unit was used for in situ water quality measurements and subsequent groundwater sampling. We managed the drilling water labeled with a fluorescein dye using a recycling system that reuses the water discharged from the borehole. We selected the test depths based on the dye concentrations, outflow water quality parameters, borehole logging, and visual inspection of the rock cores. The groundwater pumped up to the surface flowed into the flow cell, where the in situ water quality parameters were measured, and it was then collected for further laboratory measurements. Atmospheric contact was minimized during the entire process. Before hydrochemical measurements and sample collection, pumping was performed to purge the remnant drilling water. This study on a model borehole can serve as a reference for the future development of deep hydrochemical investigation procedures and techniques for siting processes of SNF repositories.

• 방사성폐기물학회지
• /
• 제19권4호
• /
• pp.533-546
• /
• 2021

Large earthquakes with (M_W > ~ 6) result in ground shaking, surface ruptures, and permanent deformation with displacement. The earthquakes would damage important facilities and infrastructure such as large industrial establishments, nuclear power plants, and waste disposal sites. In particular, earthquake ruptures associated with large earthquakes can affect geological and engineered barriers such as deep geological repositories that are used for storing hazardous radioactive wastes. Earthquake-driven faults and surface ruptures exhibit various fault zone structural characteristics such as direction of earthquake propagation and rupture and asymmetric displacement patterns. Therefore, estimating the respect distances and hazardous areas has been challenging. We propose that considering multiple parameters, such as fault types, distribution, scale, activity, linkage patterns, damage zones, and respect distances, enable accurate identification of the sites for deep geological repositories and important facilities. This information would enable earthquake hazard assessment and lower earthquake-resulted hazards in potential earthquake-prone areas.

• Wind and Structures
• /
• 제32권4호
• /
• pp.321-340
• /
• 2021

In 2019, 5.6% of the total energy produced worldwide came from wind. Offshore wind generation is still a small portion of the total wind generation, yet its growth is exponential. Higher availability of sites, larger producibility and potentially lower environmental impacts make offshore wind generation attractive. On the other hand, as the water depth increases, fixed foundations are no more viable, and the new frontier is that of floating foundations. This paper brings an overview of why and how offshore wind energy should move deep water; it contains material from the Keynote Lecture given by the first author at the ACEM20/Structures20 Conference, held in Seoul in August 2020. The paper is organized into four sections: the first giving general concepts about wind generation especially offshore, the second and the third considering economic and technical aspects, respectively, of offshore deep-water wind generation, in the fourth, some challenges of floating offshore wind generation are presented and some conclusions are drawn.

• Journal of Astronomy and Space Sciences
• /
• 제30권2호
• /
• pp.101-106
• /
• 2013

The length of solar cycle 23 has been prolonged up to about 13 years. Many studies have speculated that the solar cycle 23/24 minimum will indicate the onset of a grand minimum of solar activity, such as the Maunder Minimum. We check the trends of solar (sunspot number, solar magnetic fields, total solar irradiance, solar radio flux, and frequency of solar X-ray flare), interplanetary (interplanetary magnetic field, solar wind and galactic cosmic ray intensity), and geomagnetic (Ap index) parameters (SIG parameters) during solar cycles 21-24. Most SIG parameters during the period of the solar cycle 23/24 minimum have remarkably low values. Since the 1970s, the space environment has been monitored by ground observatories and satellites. Such prevalently low values of SIG parameters have never been seen. We suggest that these unprecedented conditions of SIG parameters originate from the weakened solar magnetic fields. Meanwhile, the deep 23/24 solar cycle minimum might be the portent of a grand minimum in which the global mean temperature of the lower atmosphere is as low as in the period of Dalton or Maunder minimum.

• 한국신재생에너지학회:학술대회논문집
• /
• 한국신재생에너지학회 2010년도 춘계학술대회 초록집
• /
• pp.238.2-238.2
• /
• 2010

Recently, environmental pollution and energy depletion problems have been issued over the world. For this reason, many renewable systems have been developing. Of these, the Ocean Thermal Energy Conservation(OTEC) is drawing attention as the upcoming alternative energy source. In this paper, the efficiency of each of OTEC which harness the effluent from nuclear power plant was analyzed by using computer calculation. The result, shows that Ul-jin Nuclear Power Plant is the best place geographically and the regenerative cycle is most outstanding performance cycle for OTEC. The difference of temperature between surface water and deep water temperature should be greater than $20^{\circ}C$ in order to increase the efficiency.

• 설비공학논문집
• /
• 제16권6호
• /
• pp.566-573
• /
• 2004

Comparison of the cooling performance is provided between the desiccant cool-ing systems incorporating a direct evaporative cooler and a regenerative evaporative cooler, respectively. Cycle simulation is conducted, and the cooling capacity and COP are evaluated at various temperature and humidity conditions. The COP of the system with a regenerative evaporative cooler and the regeneration temperature of 6$0^{\circ}C$ is evaluated 0.65 at the outdoor air condition of 35$^{\circ}C$ and 40% RH. This value is found about 3.4 times larger than that of the system with a direct evaporative cooler. Furthermore, incorporating a regenerative evaporative cooler eliminates the need for deep dehumidification in a desiccant dehumidifier that is necessary to achieve low air temperature in the system with a direct evaporative cooler. Subsequently, the regenerative evaporative cooler enables the use of low temperature heat source to regenerate the dehumidifier permitting the desiccant cooling system more beneficial compared with other thermal driven air conditioners.

• Bulletin of the Korean Chemical Society
• /
• 제23권1호
• /
• pp.75-80
• /
• 2002

Cycle life tests have been carried out to evaluate the influence of safety valve pressure on valve regulated lead/acid batteries under deep cycling applications. Batteries were cycled at 5 hour rates at 100% DOD, and safety valve pressure was set to 1.08 and 2.00 bar, respectively. The batteries lost 248.3 g and 235.3 g of water for each case after about 1,200 cycles, but the cyclic performances of the batteries were comparable. Most of the gas of the battery during discharging was hydrogen, and the oxygen concentration increased to 18% after 3 hours of charging. The micro structure of the positive active materials was completely changed and the corrosion layer of the positive grid was less than $50{\mu}m$, regardless of the pressure of the safety valve after cycle life tests. The cause of discharge capacity decrease was found to be water loss and the shedding of the positive active materials. The pressure of safety valve does not give little effect to the cyclic performances and the failure modes of the gelled electrolyte valve-regulated lead acid batteries.

• 한국군사과학기술학회지
• /
• 제22권1호
• /
• pp.49-59
• /
• 2019

Generative adversarial networks(GANs) have received great attention in the machine learning field for their capacity to model high-dimensional and complex data distribution implicitly and generate new data samples from the model distribution. This paper investigates the model training methodology, architecture, and various applications of generative adversarial networks. Experimental evaluation is also conducted for generating synthetic image dataset for defense using two types of GANs. The first one is for military image generation utilizing the deep convolutional generative adversarial networks(DCGAN). The other is for visible-to-infrared image translation utilizing the cycle-consistent generative adversarial networks(CycleGAN). Each model can yield a great diversity of high-fidelity synthetic images compared to training ones. This result opens up the possibility of using inexpensive synthetic images for training neural networks while avoiding the enormous expense of collecting large amounts of hand-annotated real dataset.