• 한국ITS학회 논문지
• /
• 제22권1호
• /
• pp.103-114
• /
• 2023

본 연구에서는 도시부 도로의 다양한 자료를 수집하여 통행속도 변화에 대한 영향을 분석하였고, 이와 같은 빅데이터를 활용하여 GRU 기반의 단기 통행속도 예측 모형을 개발하였다. 그리고 Baseline 모형과 이중지수평활 모형을 비교 모형으로 선정하여 RMSE 지표로 예측 오차를 평가하였다. 모형 평가 결과, Baseline 모형과 이중지수평활 모형의 RMSE는 평균 7.46, 5.94값으로 각각 산출되었다. 그리고 GRU 모형으로 예측한 평균 RMSE는 5.08 값이 산출되었다. 15개 링크별로 편차가 있지만, 대부분의 경우 GRU 모형의 오차가 최소의 값을 나타내었고, 추가적인 산점도 분석 결과도 동일한 결과를 제시하였다. 이러한 결과로부터 도시부 도로의 통행속도 정보 생성 과정에서 GRU 기반의 예측 모형 적용 시 예측 오차를 감소시키고 모형 적용 속도의 개선을 기대할 수 있을 것으로 판단된다.

• 한국물환경학회지
• /
• 제36권3호
• /
• pp.229-244
• /
• 2020

Hydrologic models, as a useful tool for understanding the hydrologic phenomena in the watershed, have become more complex with the increase of computer performance. The hydrologic model, with complex configurations and powerful performance, facilitates a broader understanding of the effects of climate and soil in hydrologic partitioning. However, the more complex the model is, the more effort and time is required to drive the model, and the more parameters it uses, the less accessible to the user and less applicable to the ungauged watershed. Rather, a parsimonious hydrologic model may be effective in hydrologic modeling of the ungauged watershed. Thus, a semi-distributed hydrologic partitioning model was developed with minimal composition and number of parameters to improve applicability. In this study, the validity and performance of the proposed model were confirmed by applying it to the Namgang Dam, Andong Dam, Hapcheon Dam, and Milyang Dam watersheds among the Nakdong River watersheds. From the results of the application, it was confirmed that despite the simple model structure, the hydrologic partitioning process of the watershed can be modeled relatively well through three vertical layers comprising the surface layer, the soil layer, and the aquifer. Additionally, discussions were conducted on antecedent soil moisture conditions widely applied to stormwater estimation using the soil moisture data simulated by the proposed model.

• 대한안전경영과학회지
• /
• 제10권3호
• /
• pp.89-98
• /
• 2008

A preventive maintenance model, caller FNBM($\alpha$, $\delta$, $\gamma$)model, is proposed to decide an optimal repair number under achieved availability requirements(r) along with taking two types of failures (repairable or irrepairable) into account. In this model, the current system is replaced by a new one in case when it doesn't meet the achieved availability requirement, even though it is repairable failure; Otherwise it is replaced in time of the first irrepairable failure. Assumed that the j-th failure is repairable with probability ${\alpha}_j$ minimal repairs are allowed for repairable failure between replacements. Expected cost rate for preventive maintenance model is developed using NHPP(Non-Homogeneous Poisson Process) in order to determine the optimal number $n^*$, also numerical examples are shown in order to explain the proposed model. Since the proposed FNBM($\alpha$, $\delta$, $\gamma$)model includes Park FNBM model(1979) and Nakagawa FNBM(p)model(1983) this proposed model is thought to be better than previous model, especially for weapon system which requires availability as primary parameter.

• The Plant Pathology Journal
• /
• 제30권2호
• /
• pp.125-135
• /
• 2014

A population model of bacterial spot caused by Xanthomonas campestris pv. vesicatoria on hot pepper was developed to predict the primary disease infection date. The model estimated the pathogen population on the surface and within the leaf of the host based on the wetness period and temperature. For successful infection, at least 5,000 cells/ml of the bacterial population were required. Also, wind and rain were necessary according to regression analyses of the monitored data. Bacterial spot on the model is initiated when the pathogen population exceeds $10^{15}cells/g$ within the leaf. The developed model was validated using 94 assessed samples from 2000 to 2007 obtained from monitored fields. Based on the validation study, the predicted initial infection dates varied based on the year rather than the location. Differences in initial infection dates between the model predictions and the monitored data in the field were minimal. For example, predicted infection dates for 7 locations were within the same month as the actual infection dates, 11 locations were within 1 month of the actual infection, and only 3 locations were more than 2 months apart from the actual infection. The predicted infection dates were mapped from 2009 to 2012; 2011 was the most severe year. Although the model was not sensitive enough to predict disease severity of less than 0.1% in the field, our model predicted bacterial spot severity of 1% or more. Therefore, this model can be applied in the field to determine when bacterial spot control is required.

• 대한음성학회지:말소리
• /
• 제50호
• /
• pp.51-69
• /
• 2004

In this paper, we present a one-pass semi-dynamic network decoding framework that inherits both advantages of fast decoding speed from static network decoders and memory efficiency from dynamic network decoders. Our method is based on the novel language model network representation that is essentially of finite state machine (FSM). The static network derived from the language model network [1][2] is partitioned into smaller subnetworks which are static by nature or self-structured. The whole network is dynamically managed so that those subnetworks required for decoding are cached in memory. The network is near-minimized by applying the tail-sharing algorithm. Our decoder is evaluated on the 25k-word Korean broadcast news transcription task. In case of the search network itself, the network is reduced by 73.4% from the tail-sharing algorithm. Compared with the equivalent static network decoder, the semi-dynamic network decoder has increased at most 6% in decoding time while it can be flexibly adapted to the various memory configurations, giving the minimal usage of 37.6% of the complete network size.

• 대한의용생체공학회:의공학회지
• /
• 제25권3호
• /
• pp.207-215
• /
• 2004

본 연구에서는 유전 프로그램을 이용하여 6가지 형태의 금속 와이어 배치를 가진 전극을 생성하여 각 전극이 3차원 형태의 달팽이관에 미치는 기계적인 영향과 그 거동을 유한요소법을 이용하여 분석하였다. 그리고 전극이 삽입되는 달팽이관은 helico-spiral 근사방법에 의해 결정된 식을 이용하였으며, 고실계단(scala tympani)의 형태는 실제 고실계단 단면의 영상정보를 추출하여 얻어내었다. 해석결과 금속 와이어가 평행하게 배치될 경우 가장 큰 힘을 달팽이관 외벽에 전달하며, 삽입에도 가장 큰 힘이 필요한 것을 알 수 있었다. 그리고 달팽이관은 3차원 구조이므로 전극이 삽입되면서 삽입 전에 자극 사이트가 바라보도록 설정한 방향보다 조금 돌아감을 확인하였다. 이 연구의 결과로부터 전극의 강성이 외상(trauma)을 줄이는데 중요하다는 것을 알 수 있다.

• 한국정보시스템학회지:정보시스템연구
• /
• 제15권3호
• /
• pp.131-157
• /
• 2006

A lack of trust in the technical and institutional environments surrounding the web can hinder e-commerce adoption, because Internet social cues are minimal and trust is difficult th establish. Web vendors must act purposefully to overcome consumer perceptions of uncertainty and risk by building trust-both in their own web sites and in the broader Internet environments. Trust makes consumers comfortable by sharing personal information making purchases, and acting on web vendor advice-behaviors essential to wide-spread adoption of e-commerce. Understanding the nature and antecedents of trust is, therefore, a major issue for both Internet researchers and practitioners. Prior research on e-commerce trust has used diverse, incomplete, and inconsistent definitions of trust therefore, making it difficult to compare results across studies. This study tried to and the empirical relationships among the trust-related factors on the usage of Internet shooing mall. The model includes five high-level constructs-disposition to trust institution-based trust trusting beliefs, trusting intention and environmental factors. The results showed that the disposition to trust and the environmental factors had a significant effect on the web site trust however, their effect was not consistent Moreover, the model suggested in this study need to be extended including with the more sociological factors and results in this study required to be verified with those extended model.

• 한국군사과학기술학회지
• /
• 제7권4호
• /
• pp.5-12
• /
• 2004

The tracking algorithm based on the interacting multiple model(IMM) requires a considerable number of sub-models for the various maneuvering targets in order to have a good performance. But it is not feasible to use the nm algorithm in the real system because of the computational burden. Therefore, we need an algorithm which requires less computing resources while maintaining a good performance. In this paper, we propose a fuzzy interacting multiple model algorithm(FIMMA) for the tracking of maneuvering targets, which uses a minimal number of sub-models by considering the maneuvering properties and adjusts the mode transition probabilities by using the mode probability as a fuzzy input. In order to verify the performance of FIMMA, the developed algorithm is applied to the tracking of i borne targets. Simulation results show that the FIMMA is very effective in the tracking of maneuvering targets.

• 인터넷정보학회논문지
• /
• 제14권4호
• /
• pp.53-62
• /
• 2013

클라우드 로봇에서, 효율적인 정보공유 모델은 연구과제이다. 본 논문에서는 클라우드 기반 로봇들이 서로 연동하고 지식을 공유하기 위한 지식공유 모델을 제시한다. 효율적인 메시지 전송을 위해 멀티 에이전트 기반 적응형 그룹 통신을 제시하며, 연결(link) 노드는 주요 링크를 결정하기 위한 비중함수를 사용한다. 성능평가 결과, 제시한 알고리즘이 최소 메세지 오버헤드를 나타내었으며, 전통적인 그룹통신 방법에 비해 주요 링크를 사용함으로 인해 질의에 대한 응답이 빠름을 보였다.

• 한국산학기술학회논문지
• /
• 제5권2호
• /
• pp.191-196
• /
• 2004

경영자가 가격이나 원가와 관련된 정확한 정보를 얻을 수 있다면 DEA모형을 이용하여 기술적 효율성 뿐 만 아니라 배분적 효율성과 관련된 분석이 가능해진다. 배분적 효율성을 분석하므로서 총체적 효율성을 기술적 효율성과 배분적 효율성으로 분해할 수 있고 이를 통해 보다 다양한 분석수단을 얻을 수 있다. 본 연구에서는 기본적 DEA 모형을 변형하여 배분효율성을 분석할 수 있는 DEA모형을 설계한다. 우선 예산선에 바탕을 둔 비용최소화 모형을 설계하고 이것을 다시 이익 극대화 모형으로 변환한 후 가산형의 DEA 모형으로 재설계한다. 가격정보가 주어진 경우 기술적 효율성의 범주를 벗어난 보다 다양한 형태의 효율성 분석이 가능해져 성과평가 및 그 결과의 해석 그리고 보다 현실성 있는 경영개선방안의 도출이 가능해진다.