• 한국작물학회지
• /
• 제55권4호
• /
• pp.312-318
• /
• 2010

이 연구는 칼라 디지털 카메라 영상 분석에 의한 유수분화기 벼 군락의 건물중 및 질소흡수량 추정 모델, 유수분화기 질소흡수량과 질소 수비량에 따른 유수분화기 이후 질소 흡수량 추정 모델, 유수분화기 질소 흡수량과 유수분화기 이후 질소 흡수량에 따른 수량 및 쌀 단백질함량 추정 모델을 구축하고 이를 종합하여 목표 수량 또는 단백질함량을 달성하기 위한 질소 수비량을 추천하는 원시 프로그램을 개발하여 현장 적용성 예비 검토를 실시하고자 한 것이다. 1. 군락피복도(CC)와 군락체적(CV)은 초장, 건물중 및 질소흡수량과 고도로 유의한정의 상관을 나타내었으며, R, G, B, NDI 및 명도 값은 이들과 유의한 부의 상관을 나타내었다. 한편 표준화된 색지표인 r은 잎 및 지상부 질소함량과 유의한 부의 상관을 나타내었으나 표준화 색지표 b와 g는 이들과 유의한 상관을 보이지 않았다. 2. 벼 군락의 디지털 카메라 영상분석을 이용한 벼 지상부 건물중 및 질소 흡수량을 추정하기 위한 비선형회귀 모델을 작성하였다. 지상부 건물중 추정 모텔에는 CC와 정규화된 R값(r, NorR)이 변수로 채택이 되었고, 질소흡수량 추정에는 CC와 정규화된 G값(g, NorG)이 채택되었으며, 이들 모델의 결정계수는 각각 0.81과 0.68이었다. 영성분석 색지표 이외에 초장을 모델에 도입하는 경우 모델의 결정계수는 더 높아졌다. 3. 유수분화기에 적정 질소 추비량을 처방하기 위해서는 유수분화기 식물체의 질소 영양 상태(질소 흡수량, PNup) 및 질소 시비량이 수량과 단백질함량에 미치는 영향을 정량화하여야 하는데, 이를 위하여 Npi와 PNup이 유수 분화기부터 성숙기까지의 지상부 질소흡수량에 미치는 영향 및 PNup과 PHNup이 벼 수량 및 쌀 단백질 함량에 미치는 영향을 검토하여 중회귀 모델을 작성하였으며 이 모델들은 결정계수가 모두 0.8이상으로 높았다. 4. 상기의 모델들을 종합하여 유수분화기 벼 군락 영상분석을 이용한 수비처방 원시 프로그램을 작성하여 예비 검증 실험을 하였다. 벼 수비 처방 프로그램에 의해 쌀단백질 함량 6.0%를 기준으로 처방된 수비질소 분시율이 19%~21%로 표준재배 분시율 30%에 비해 낮은 수준으로 처방되었으나 완전미 수량은 대등하였고, 단백질함량은 수비처방 목표단백질 함량 6%보다는 다소 낮은 5.7~5.8%였으며, 수량과 단백질함량의 변이 계수는 관행 수비 처방구에 비하여 프로그램 처방구에서 크게 낮아져서 프로그램 처방에 의하여 수량과 품질이 균질화되는 결과였다.

• 한국수자원학회논문집
• /
• 제55권10호
• /
• pp.723-736
• /
• 2022

본 연구에서는 금강권역을 대상으로 최대 12개월까지 선행예측이 가능한 월 강수량 예측모형을 구축하였으며, 예측모형 구축에는 다중회귀분석과 인공신경망의 두 가지 통계적 기법을 적용하였다. 예측인자 후보로 NOAA에서 제공하는 글로벌 기후패턴 39종과 금강권역에 대한 기상인자 8종 등 총 47종의 기후지수를 활용하였다. 예측대상월을 기준으로 과거 40년간의 월 강수량과 기후지수와의 지연상관성 분석을 통해 상관도가 높은 기후지수를 예측인자로 활용하여 다중회귀모형 및 인공신경망 모형을 구축하였다. 1991~2021년에 대해 매월 예측결과의 평균값과 관측값과의 적합도를 분석한 결과, 다중회귀모형은 PBIAS -3.3~-0.1%, NSE 0.45~0.50, r 0.69~0.70으로 분석되었으며, 인공신경망모형은 PBIAS -5.0~+0.5%, NSE 0.35~0.47, r 0.64~0.70로, 다중회귀모형에 의해 도출된 예측치의 평균값이 인공신경망모형보다 관측치에 좀 더 근접한 것으로 나타났다. 각 월의 예측범위 안에 관측치가 포함될 확률을 분석한 결과에서는 다중회귀모형이 57.5~83.6%(평균 72.9%), 인공신경망모형의 경우에는 71.5~88.7%(평균 81.1%)로 인공신경망모형 결과가 우수한 것으로 나타났다. 3분위 예측확률을 비교한 결과는 다중회귀모형의 경우에는 25.9~41.9%(평균 34.6%), 인공신경망모형은 30.3~39.1%(평균 34.7%)로 비슷하며, 두 모형 모두 평균 33.3% 이상으로 월 강수량에 대한 장기예측성을 확인 할 수 있었다. 이상과 같이 두 모형의 예측성 차이는 비교적 크지 않은 것으로 나타났으나, 예측범위에 대한 적중률이나 3분위 예측확률로부터 판단할 때 예측성에 대한 월별 편차는 인공신경망모형의 결과가 상대적으로 작게 나타났다.

• 한국농업기계학회:학술대회논문집
• /
• 한국농업기계학회 2017년도 춘계공동학술대회
• /
• pp.45-45
• /
• 2017

On-site monitoring of vegetable growth parameters, such as leaf length, leaf area, and fresh weight, in an agricultural field can provide useful information for farmers to establish farm management strategies suitable for optimum production of vegetables. Unmanned Aerial Vehicles (UAVs) are currently gaining a growing interest for agricultural applications. This study reports on validation testing of previously developed vegetable growth estimation models based on UAV-based RGB images for white radish and Chinese cabbage. Specific objective was to investigate the potential of the UAV-based RGB camera system for effectively quantifying temporal and spatial variability in the growth status of white radish and Chinese cabbage in a field. RGB images were acquired based on an automated flight mission with a multi-rotor UAV equipped with a low-cost RGB camera while automatically tracking on a predefined path. The acquired images were initially geo-located based on the log data of flight information saved into the UAV, and then mosaicked using a commerical image processing software. Otsu threshold-based crop coverage and DSM-based crop height were used as two predictor variables of the previously developed multiple linear regression models to estimate growth parameters of vegetables. The predictive capabilities of the UAV sensing system for estimating the growth parameters of the two vegetables were evaluated quantitatively by comparing to ground truth data. There were highly linear relationships between the actual and estimated leaf lengths, widths, and fresh weights, showing coefficients of determination up to 0.7. However, there were differences in slope between the ground truth and estimated values lower than 0.5, thereby requiring the use of a site-specific normalization method.

• 한국작물학회지
• /
• 제36권2호
• /
• pp.97-106
• /
• 1991

온도와 일장을 예측변수로 하는 2차원 non-par-ametric model을 개발하여, 건답직파재배에서 파종기 이동 및 단일처리 (26개품종, 4월 10일부터 2주 간격으로 8회 파종, 해지기 직전 1시간 차광)를 하여 얻은 자료로부터 출아에서 출수까지의 일평균발육속도(DVR)를 추정하였다. 또한 여기서 추정한 DVR을 이용 독립자료에 대하여 모델을 검증하였다. 1. 발육 예측정도는 온도와 일장에 대한 smoothing parameter λ$_{T}$ 와 λ$_{L}$에 따라서 단조적으로 변하였으며 예측정도를 가장 높게하는 λ$_{T}$ 와 λ$_{L}$이 존재하였다. 2. 최적 λ$_{T}$와 λ$_{L}$은 품종에 따라서 달랐으며 5～100,000의 범위내에 있었다 3. 최적 λ$_{T}$와 λ$_{L}$에서 구한 DVR을 이용하여 발육을 예측하는 경우 C.V는 품종에 따라 0.5-2.6% 였으며 기존의 함수모델들 보다 예측 정도가 높았다 4. DVR을 계산하는데 이용되지 않은 독립자료를 이용하여 11개 품종을 대상으로 출수기를 예측한 결과 예측오차는 0-3일로 추정 정도가 높았다.

• Geomechanics and Engineering
• /
• 제2권4호
• /
• pp.253-280
• /
• 2010

A new prediction model is derived for the uplift capacity of suction caissons using a hybrid method coupling genetic programming (GP) and simulated annealing (SA), called GP/SA. The predictor variables included in the analysis are the aspect ratio of caisson, shear strength of clayey soil, load point of application, load inclination angle, soil permeability, and loading rate. The proposed model is developed based on well established and widely dispersed experimental results gathered from the literature. To verify the applicability of the proposed model, it is employed to estimate the uplift capacity of parts of the test results that are not included in the modeling process. Traditional GP and multiple regression analyses are performed to benchmark the derived model. The external validation of the GP/SA and GP models was further verified using several statistical criteria recommended by researchers. Contributions of the parameters affecting the uplift capacity are evaluated through a sensitivity analysis. A subsequent parametric analysis is carried out and the obtained trends are confirmed with some previous studies. Based on the results, the GP/SA-based solution is effectively capable of estimating the horizontal, vertical and inclined uplift capacity of suction caissons. Furthermore, the GP/SA model provides a better prediction performance than the GP, regression and different models found in the literature. The proposed simplified formulation can reliably be employed for the pre-design of suction caissons. It may be also used as a quick check on solutions developed by more time consuming and in-depth deterministic analyses.

• 산업경영시스템학회지
• /
• 제38권3호
• /
• pp.56-63
• /
• 2015

Several fields of science have demanded large-scale workflow support, which requires thousands of CPU cores or more. In order to support such large-scale scientific workflows, high capacity parallel systems such as supercomputers are widely used. In order to increase the utilization of these systems, most schedulers use backfilling policy: Small jobs are moved ahead to fill in holes in the schedule when large jobs do not delay. Since an estimate of the runtime is necessary for backfilling, most parallel systems use user's estimated runtime. However, it is found to be extremely inaccurate because users overestimate their jobs. Therefore, in this paper, we propose a novel system for the runtime prediction based on workload-aware clustering with the goal of improving prediction performance. The proposed method for runtime prediction of parallel applications consists of three main phases. First, a feature selection based on factor analysis is performed to identify important input features. Then, it performs a clustering analysis of history data based on self-organizing map which is followed by hierarchical clustering for finding the clustering boundaries from the weight vectors. Finally, prediction models are constructed using support vector regression with the clustered workload data. Multiple prediction models for each clustered data pattern can reduce the error rate compared with a single model for the whole data pattern. In the experiments, we use workload logs on parallel systems (i.e., iPSC, LANL-CM5, SDSC-Par95, SDSC-Par96, and CTC-SP2) to evaluate the effectiveness of our approach. Comparing with other techniques, experimental results show that the proposed method improves the accuracy up to 69.08%.

• Earthquakes and Structures
• /
• 제2권4호
• /
• pp.357-376
• /
• 2011

Ground motion models predict the mean and standard deviation of the logarithm of spectral acceleration, as a function of predictor variables such as earthquake magnitude, distance and site condition. Such models have been developed for a variety of seismic environments throughout the world. Some calculations, such as the Conditional Mean Spectrum calculation, use this information but additionally require knowledge of correlation coefficients between logarithmic spectral acceleration values at multiple periods. Such correlation predictions have, to date, been developed primarily from data recorded in the Western United States from active shallow crustal earthquakes. This paper describes results from a study of spectral acceleration correlations from Japanese earthquake ground motion data that includes both crustal and subduction zone earthquakes. Comparisons are made between estimated correlations for Japanese response spectral ordinates and correlation estimates developed from Western United States ground motion data. The effect of ground motion model, earthquake source mechanism, seismic zone, site conditions, and source to site distance on estimated correlations is evaluated and discussed. Confidence intervals on these correlation estimates are introduced, to aid in identifying statistically significant differences in correlations among the factors considered. Observed general trends in correlation are similar to previous studies, with the exception of correlation of spectral accelerations between orthogonal components, which is seen to be higher here than previously observed. Some differences in correlations between earthquake source zones and earthquake mechanisms are observed, and so tables of correlations coefficients for each specific case are provided.

• The Plant Pathology Journal
• /
• 제39권6호
• /
• pp.548-565
• /
• 2023

Armillaria root disease affects forests around the world. It occurs in many habitats and causes losses in the infested stands. Weather conditions are important factors for growth and development of Armillaria species. Yet, the relation between occurrence of damage caused by Armillaria disease and weather variables are still poorly understood. Thus, we used generalized linear mixed models to determine the relationship between weather conditions of current and previous year (temperature, precipitation and their deviation from long-term averages, air humidity and soil temperature) and the incidence of Armillaria-induced damage in young (up to 20 years old) and older (over 20 years old) coniferous stands in selected forest districts across Poland. We used unique data, gathered over the course of 23 years (1987-2009) on tree damage incidence from Armillaria root disease and meteorological parameters from the 24-year period (1986-2009) to reflect the dynamics of damage occurrence and weather conditions. Weather parameters were better predictors of damage caused by Armillaria disease in younger stands than in older ones. The strongest predictor was soil temperature, especially that of the previous year growing season and the current year spring. We found that temperature and precipitation of different seasons in previous year had more pronounced effect on the young stand area affected by Armillaria. Each stand's age class was characterized by a different set of meteorological parameters that explained the area of disease occurrence. Moreover, forest district was included in all models and thus, was an important variable in explaining the stand area affected by Armillaria.

• Journal of Chest Surgery
• /
• 제57권2호
• /
• pp.184-194
• /
• 2024

Background: Left ventricular assist devices (LVADs) are widely employed as a therapeutic option for end-stage heart failure. We evaluated the outcomes associated with centrifugal-flow LVAD implantation, comparing 2 device models: the Heartmate 3 (HM3) and the Heartware Ventricular Assist Device (HVAD). Methods: Data were collected from patients who underwent LVAD implantation between June 1, 2015 and December 31, 2022. We analyzed overall survival, first rehospitalization, and early, late, and LVAD-related complications. Results: In total, 74 patients underwent LVAD implantation, with 42 receiving the HM3 and 32 the HVAD. A mild Interagency Registry for Mechanically Assisted Circulatory Support score was more common among HM3 than HVAD recipients (p=0.006), and patients receiving the HM3 exhibited lower rates of preoperative ventilator use (p=0.010) and extracorporeal membrane oxygenation (p=0.039). The overall early mortality rate was 5.4% (4 of 74 patients), with no significant difference between groups. Regarding early right ventricular (RV) failure, HM3 implantation was associated with a lower rate (13 of 42 [31.0%]) than HVAD implantation (18 of 32 [56.2%], p=0.051). The median rehospitalization-free period was longer for HM3 recipients (16.9 months) than HVAD recipients (5.3 months, p=0.013). Furthermore, HM3 recipients displayed a lower incidence of late hemorrhagic stroke (p=0.016). In the multivariable analysis, preoperative use of continuous renal replacement therapy (odds ratio, 22.31; p=0.002) was the only significant predictor of postoperative RV failure. Conclusion: The LVAD models (HM3 and HVAD) demonstrated comparable overall survival rates. However, the HM3 was associated with a lower risk of late hemorrhagic stroke.

• 대한통합의학회지
• /
• 제12권2호
• /
• pp.155-166
• /
• 2024

Purpose : This study aims to identify key factors for predicting dropout risk at the university level and to provide a foundation for policy development aimed at dropout prevention. This study explores the optimal machine learning algorithm by comparing the performance of various algorithms using data on college students' dropout risks. Methods : We collected data on factors influencing dropout risk and propensity were collected from N University. The collected data were applied to several machine learning algorithms, including random forest, decision tree, artificial neural network, logistic regression, support vector machine (SVM), k-nearest neighbor (k-NN) classification, and Naive Bayes. The performance of these models was compared and evaluated, with a focus on predictive validity and the identification of significant dropout factors through the information gain index of machine learning. Results : The binary logistic regression analysis showed that the year of the program, department, grades, and year of entry had a statistically significant effect on the dropout risk. The performance of each machine learning algorithm showed that random forest performed the best. The results showed that the relative importance of the predictor variables was highest for department, age, grade, and residence, in the order of whether or not they matched the school location. Conclusion : Machine learning-based prediction of dropout risk focuses on the early identification of students at risk. The types and causes of dropout crises vary significantly among students. It is important to identify the types and causes of dropout crises so that appropriate actions and support can be taken to remove risk factors and increase protective factors. The relative importance of the factors affecting dropout risk found in this study will help guide educational prescriptions for preventing college student dropout.