• 에너지공학
• /
• 제17권3호
• /
• pp.139-144
• /
• 2008

최근 유가의 고공행진 때문에 한국은 해외 에너지 자원개발뿐만 아니라 에너지 소비를 줄이기 위한 국가적 정책으로 기존 설비의 에너지 효율을 증가시키는 방안을 모색하는 것이 필요한 시점이다. 따라서 본 연구는 이런 국가적 사안에 부합하고자 인천생산기지 고압 LNG 배관망에 대하여 수정유량방정식을 사용한 Newton Method로 접근하였고, 유창조절밸브(FCV)에 의해 지배적인 영향을 받는 것을 확인할 수 있었다. 또한, 고압펌프는 유량조절밸브 50%의 개도율에서 최고효율을 보여 주었고, 고압배관망 내에서 배관저항곡선은 LNG 헤드가 1,500m 이상이 되어야만 토출이 가능한 것을 보였다. 고압펌프의 운전점으로부터 운전비용을 산출하였고, 최고 효율시 운전비용과 비교하여 운전비용을 절감할 수 있는 금액을 산출하였다. 특히 일간 시간대별 운전비용 절감액뿐만 아니라 연간 일별 운전비용 절감액을 산출하였으며, 그 결과 고압배관망은 연간 138백만원을 절감할 수 있다. 이것은 연간 고압펌프 1기당 9,823천원을 절감할 수 있다는 것을 의미한다. 결론적으로 본 연구는 복잡한 고압 LNG 배관망에서 고압펌프의 운전특성과 운전비용 절감효과를 확인할 수 있었다. 또한 이것은 미시적으로 생산기지의 효율적 미래운영에 대한 기여와 더불어 거시적으로 국가 에너지 경쟁력 제고에 기여할 수 있을 것이다.

• 한국수자원학회논문집
• /
• 제55권spc1호
• /
• pp.1177-1185
• /
• 2022

상·하수도 시스템은 사람들에게 안전하고 깨끗한 물을 공급해주는 사회기반시설이며, 특히 상·하수도 관로는 지중에 매설되어 있기 때문에 시스템의 결함검출이 매우 어렵다. 이러한 이유로 상·하수도 관로의 진단은 관로 내부에 카메라 및 드론을 통한 촬영을 하여 사후에 촬영된 영상을 바탕으로 시스템 진단하는 등의 사후 결함검출로 제한되기 때문에, 작업자의 업무 효율 증대와 진단의 신속성을 위해서는 관로의 실시간 탐지기술이 필요하다. 최근 첨단장비 및 인공지능 기법을 활용한 시설물 진단 기술이 개발되고 있지만, 인공지능기반 결함검출 기술은 결함 데이터의 종류 및 형태, 수가 검출 성능에 영향을 주기 때문에 다양한 학습데이터가 필요하다. 따라서, 본 연구에서는 상·하수도 관로의 결함검출 시 탐지 성능 향상을 위해 다양한 결함 시나리오를 3D 프린트를 이용하여 구현하고 이를 수집된 결함 데이터와 함께 학습데이터로 사용한다. 이후 수집된 이미지는 위험도에 따른 분류 및 객체의 라벨링 등의 전처리 작업이 수행되고 실시간 결함탐지를 수행한다. 제안된 기법은 상·하수도시스템 결함검출 시 실시간 피드백을 제공함으로써, 작업자의 진단 누락 가능성을 최소화하며 기존의 상·하수도관 진단업무 처리능력을 향상할 수 있다.

• 한국정보통신학회논문지
• /
• 제10권5호
• /
• pp.865-870
• /
• 2006

셀룰러 신경회로망(Cellular Neural Networks: CNN)은 그 구조가 간단함에도 불구하고 강력한 연산능력을 가지고 있어 영상처리에 이용되어 왔다. 그러나 실제의 대규모 영상에 포함된 화소의 양과 같은 막대한 셀들을 필요로 하는 CNN하드웨어를 구현하는 것은 불가능하다. 본 논문에서는 시 다중화 처리 기법으로 대규모 실영상을 처리할 수 있는 $5\times5$ CNN 하드웨어와 전 후 처리기를 구현하였다. 구현된 $5\times5$ CNN 하드웨어와 전 후 처리기의 성능을 평가하기 위해 $ 레나영상에 대해 윤곽선 검출을 수행하였으며, 약 4,000번의 시다중화 블록처리와 각 블록 마다 10번의 제어 펄스에 의한 파이프라인 동작에 의해 영상처리가 수행되었다. 따라서 본 논문에서 구현된 $5\times5$ CNN 하드웨어와 전 후 처리기를 실영상 처리에 이용할 수 있다.

• BMB Reports
• /
• 제41권3호
• /
• pp.184-193
• /
• 2008

Living organisms are comprised of various systems at different levels, i.e., organs, tissues, and cells. Each system carries out its diverse functions in response to environmental and genetic perturbations, by utilizing biological networks, in which nodal components, such as, DNA, mRNAs, proteins, and metabolites, closely interact with each other. Systems biology investigates such systems by producing comprehensive global data that represent different levels of biological information, i.e., at the DNA, mRNA, protein, or metabolite levels, and by integrating this data into network models that generate coherent hypotheses for given biological situations. This review presents a systems biology framework, called the 'Integrative Proteomics Data Analysis Pipeline' (IPDAP), which generates mechanistic hypotheses from network models reconstructed by integrating diverse types of proteomic data generated by mass spectrometry-based proteomic analyses. The devised framework includes a serial set of computational and network analysis tools. Here, we demonstrate its functionalities by applying these tools to several conceptual examples.

• 산업공학
• /
• 제19권2호
• /
• pp.106-116
• /
• 2006

In this study, the priority of risk factors in supplying water through water supply pipeline network was evaluated by PROMETHEE and ANP multi-criteria decision analysis. We chose 'corrosion', 'burst' and 'water pollution' in pipe as major reference criteria and selected eight risk factors to evaluate the priority, and then we compared the results of PROMETHEE with those of ANP. We also analyzed the results of the sensitivity analysis by changing the weights and parameters of preference functions in PROMETHEE. We investigated the possibility of integrating two methods by using the results of ANP as the weights of preference function in PROMETHEE. The priority of risk factors for supplying municipal water which is evaluated by this study may provide basic data to establish a contingency plan for accidents, or to establish the specific emergency response procedures.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2016년도 춘계학술발표대회
• /
• pp.633-635
• /
• 2016

Convolutional Networks (ConvNets) are powerful models that learn hierarchies of visual features, which could also be used to obtain image representations for transfer learning. The basic pipeline for transfer learning is to first train a ConvNet on a large dataset (source task) and then use feed-forward units activation of the trained ConvNet as image representation for smaller datasets (target task). Our key contribution is to demonstrate superior performance of multiple ConvNet layer features over single ConvNet layer features. Combining multiple ConvNet layer features will result in more complex feature space with some features being repetitive. This requires some form of feature selection. We use AdaBoost with single stumps to implicitly select only distinct features that are useful towards classification from concatenated ConvNet features. Experimental results show that using multiple ConvNet layer activation features instead of single ConvNet layer features consistently will produce superior performance. Improvements becomes significant as we increase the distance between source task and the target task.

• ETRI Journal
• /
• 제37권4호
• /
• pp.743-751
• /
• 2015

Cloud gaming services are heavily dependent on the efficiency of real-time video streaming technology owing to the limited bandwidths of wire or wireless networks through which consecutive frame images are delivered to gamers. Video compression algorithms typically take advantage of similarities among video frame images or in a single video frame image. This paper presents a method for computing and extracting both graphics information and an object's boundary from consecutive frame images of a game application. The method will allow video compression algorithms to determine the positions and sizes of similar image blocks, which in turn, will help achieve better video compression ratios. The proposed method can be easily implemented using function call interception, a programmable graphics pipeline, and off-screen rendering. It is implemented using the most widely used Direct3D API and applied to a well-known sample application to verify its feasibility and analyze its performance. The proposed method computes various kinds of graphics information with minimal overhead.

• 전자통신동향분석
• /
• 제34권4호
• /
• pp.55-64
• /
• 2019

In this study, we introduce trends in neural-network-based deep learning research applied to dialogue systems. Recently, end-to-end trainable goal-oriented dialogue systems using long short-term memory, sequence-to-sequence models, among others, have been studied to overcome the difficulties of domain adaptation and error recognition and recovery in traditional pipeline goal-oriented dialogue systems. In addition, some research has been conducted on applying reinforcement learning to end-to-end trainable goal-oriented dialogue systems to learn dialogue strategies that do not appear in training corpora. Recent neural network models for end-to-end trainable chit-chat systems have been improved using dialogue context as well as personal and topic information to produce a more natural human conversation. Unlike previous studies that have applied different approaches to goal-oriented dialogue systems and chit-chat systems respectively, recent studies have attempted to apply end-to-end trainable approaches based on deep neural networks in common to them. Acquiring dialogue corpora for training is now necessary. Therefore, future research will focus on easily and cheaply acquiring dialogue corpora and training with small annotated dialogue corpora and/or large raw dialogues.

• Journal of Veterinary Science
• /
• 제19권6호
• /
• pp.798-807
• /
• 2018

Several studies have reported the effect of absorption of procyanidins and their contribution to the small intestine. However, differences between dietary interventions of procyanidins and interventions via antibiotic feeding in pigs are rarely reported. Following 16S rRNA gene Illumina MiSeq sequencing, we observed that both procyanidin administration for 2 months (procyanidin-1 group) and continuous antibiotic feeding for 1 month followed by procyanidin for 1 month (procyanidin-2 group) increased the number of operational taxonomic units, as well as the Chao 1 and ACE indices, compared to those in pigs undergoing antibiotic administration for 2 months (antibiotic group). The genera Fibrobacter and Spirochaete were more abundant in the antibiotic group than in the procyanidin-1 and procyanidin-2 groups. Principal component analysis revealed clear separations among the three groups. Additionally, using the online Molecular Ecological Network Analyses pipeline, three co-occurrence networks were constructed; Lactobacillus was in a co-occurrence relationship with Trichococcus and Desulfovibrio and a co-exclusion relationship with Bacillus and Spharerochaeta. Furthermore, metabolic function analysis by phylogenetic investigation of communities by reconstruction of unobserved states demonstrated modulation of pathways involved in the metabolism of carbohydrates, amino acids, energy, and nucleotides. These data suggest that procyanidin influences the gut microbiota and the intestinal metabolic function to produce beneficial effects on metabolic homeostasis.

• 한국산업융합학회 논문집
• /
• 제27권1호
• /
• pp.87-95
• /
• 2024

This study is on modulo scheduling algorithms for multicore processor in machine learning applications. Machine learning algorithms are designed to perform a large amount of operations such as vectors and matrices in order to quickly process large amounts of data stream. To support such large amounts of computations, processor architectures to support applications such as artificial intelligence, neural networks, and machine learning are designed in the form of parallel processing such as multicore. To effectively utilize these multi-core hardware resources, various compiler techniques are being used and studied. In this study, among these compiler techniques, we analyzed the modular scheduler, which is especially important in one core's computation pipeline. This paper looked at and compared the iterative modular scheduler and the swing modular scheduler, which are the most widely used and studied. As a result, both schedulers provided similar performance results, and when measuring register pressure as an indicator, it was confirmed that the swing modulo scheduler provided slightly better performance. In this study, a technique that divides recurrence edge is proposed to improve the minimum initiation interval of the modulo schedulers.