Image recognition is not very effective in the water environment due to multiple factors, such as high scattering and high scattering in the water column. This is why the relevant parameters in the Faster R-CNN network model need to adjust continuously to improve the effectiveness of water detection. The control variable method adjusts the program's learning rate by tuning the network model's parameters. Then, the number of training rounds is adjusted according to the loss function of each round, and finally, we can get the number of matches with the minimum loss function. Based on the experimental results on the dataset, it is shown that the proposed method not only selects the learning rate with the best detection results but also has the strongest robustness and achieves a 96%-99% recognition rate for passenger ships, cargo ships, warships, and bridges compared with other learning rates. Experiments show that the Faster R-CNN network model detects water targets with significant results, and the best network model learning rate parameter is 6×10-3.

This study set up the revenue water ratio that could be achieved within the range of the expenses of the water distribution network maintenance project, developed an analysis methodology that could estimate the additional project quantity to achieve the target revenue water ratio of 85% and applied and verified that to S. City. This methodology allowed the distribution of the leakage quantity for each leakage component by the pipeline through the total revenue water account balance analysis and BABE approach and the redistribution into the calculated leakage quantity more accurately through a step test. In addition, the level of reduction in leakage and the quantity of the project were estimated before and after the application of four strategies for the promotion of the revenue water ratio presented by IWA, according to the leakage components by the pipeline. As a result of the application of this analysis method to S. City, it would be possible to achieve up to the revenue water ratio of 81.0%, which was 74.7% in the beginning, if the water distribution network maintenance project was promoted within the range of the project expenses, and to achieve the revenue water ratio of 85.0%, the goal of the project, it would be necessary to replace the pipeline of 22.2% of the entire pipelines in the target area. As a result of the re-estimation of the revenue water ratio achievable, applying the actual water distribution network maintenance quantity in the scope of the business with the results of this analysis, the revenue water ratio was 81.7% while the actually measured revenue water ratio was 82.3%. Thus, the reliability of this analysis method could be secured to some extent.

이미지 인식에 특화된 CNN (Convolutional Neural Network) 기반의 딥러닝 기법은 영상의 항목별 분류가 필요한 다양한 연구에 적용되고 있다. 본 연구는 건물, 도로, 논, 밭, 산림, 나지의 6가지 항목을 산림복원 대상 후보지로 정의하고 CNN 기반의 산림복원 대상 후보지 추출 및 분류의 최적 방법론을 탐색하였다. 6,640개의 데이터셋을 75:25의 비율로 훈련(4,980개) 및 검증(1,660개)로 구분하여 구축하고 학습에 활용하였다. 모델별 정확도는 픽셀정확도(PA), 평균 교차 겹침 결합(Mean IoU)을 이용하여 평가하였다. 픽셀정확도는 90.6%, 평균 교차 겹침 결합은 80.8%로 산정되어 Inception-Resnet-v2 모델이 세 모델 중 가장 산림복원 대상 후보지 추출에 뛰어난 정확도를 보였다. 이 결과는 기존의 산림복원 대상 후보지 현장조사 혹은 항공사진을 활용한 조사에 비해 시공간적 이점을 가지며, 향후 산림복원 대상지 선정 자료로 적용 가능성이 있다고 판단된다.

We explored the frequency response of krill target strength (TS) to understand the Antarctic krill (Euphausia superba) and ice krill (Euphausia crystallorophias) using the stochastic distorted-wave Born approximation (SDWBA) model. The results showed that the distribution of orientation and the fatness factor could significantly impact on the frequency response of TS. Krill TS is clearly depended on acoustic properties, which could affect to estimate the biomass of two krill species. The results provide insight into the importance of understanding TS variation to estimate the Antarctic krill and ice krill biomass, and their ecology related to the environmental features in the Southern Ocean.

In this study, the target strength for multi-frequency (38 kHz, 120 kHz, 200 kHz) of juvenile silverfish (Pleuragramma antarcticum) was estimated using by the KRM (Kirchhoff-ray mode) model. The body shape of the silverfish was described by a picture and the body length of nine individuals ranged in 1.8 cm to 8.8 cm. The maximum TScm according to the total length for the constant term (b20) was – 92.93 dB at 38 kHz, – 86.63 dB at 120 kHz, and – 85.89 dB at 200 kHz, respectively. The averaged TScm according to total length for the constant term (b20) was – 100.0 dB at 38 kHz, – 93.0 dB at 120 kHz, and – 106.9 dB at 200 kHz, respectively.

부분구조화 기법은 자유도가 많고 복잡한 구조물의 유한요소 해석 모델 단순화에 효율적으로 적용될 수 있는 기법이다. 대표적으 로 선형 문제에 대해서는 Craig-Bampton method 등이 있다. Craig-Bampton method는 경계 요소를 제외한 나머지 요소의 불필요한 자 유도를 제거함으로써 선형 구조물의 축소를 수행한다. 최근에는 부분구조화 기법과 더불어 구조물의 최적설계를 위해 멀티레벨 최적 화 기법이 많이 활용되고 있다. 시스템의 목표를 달성하기 위해 각 부구조에 새로운 목표를 할당하는 기법이다. 본 연구에서는 유전자 알고리즘을 이용하여 시스템 목표 달성을 위한 각 부구조별 내부 자유도 개수를 새로운 목표로 할당하고 최적화를 수행하였다. 최적 화 절차로부터 도출된 부구조별 내부 자유도 개수를 이용하여 시스템의 축소를 수행하였다. 다양한 수치예제들을 통해 축소 모델에 대한 결과를 확인하였으며, 90% 이상의 정확도를 가지는 것을 확인하였다.

기억용이성은 자극이 얼마나 잘 기억될 수 있는가를 나타내는 척도로 자극의 지각적, 의미적 특성으로는 설명되지 않는 자극 고유의 특성이다. 기존의 기억용이성 연구는 주로 단독으로 제시된 자극을 얼마나 잘 기억할 수 있는가를 측정했다. 그러나 단일 자극만 제시하는 연구로는 다수의 자극이 동시에 나타나는 현실 세계의 시각 정보 처리 과정 을 설명하는 데 한계가 있다. 이에 본 연구에서는 다수의 자극이 동시에 제시되었을 때 자극의 기억용이성 수준에 따라 기억 과제 수행이 어떻게 달라지는가를 알아봤다. 참가자들은 기억용이성이 높거나 낮은 방해자극과 동시에 제시된 목표자극을 기억하는 과제를 수행했다. 실험 결과, 목표자극의 기억용이성이 높았을 때 기억 수행이 향상되 었으나, 목표자극의 기억용이성 효과는 동시에 제시된 방해자극의 기억용이성 수준의 영향을 받지 않았다. 이러한 결과는 높은 기억용이성을 가진 자극이 자동적으로 상향적 주의를 유도하지 않는다는 선행 연구 결과를 지지한다.

In this study, a drop analysis of metallic disposal containers for radioactive wastes is performed according to accident scenarios at the disposal site. The weight of the disposal container is about 8 tons, and the ingot-type wastes are loaded in the disposal container. To simulate the floor of the disposal site as the impact target, the reinforced concrete pad is modeled. High impact energy of the disposal container due to their heavy weight and high drop height causes excessive deformation and failure of the concrete target having relatively weak strength. Dynamic growth of cracks due to such failures causes penetration and delamination of concrete. Since the impact force delivered to the container strongly depends on the failure of the concrete pad, it is important to properly simulate the failure of the concrete in the drop analysis. A material erosion method can be used to simulate the concrete failure. In the case of applying erosion based on the finite element method (FEM), the element is deleted when the element exceeds a certain criterion, which causes material and energy loss problem. To solve this problem, mesh-free methods such as smoothed particle hydrodynamics (SPH) can be commonly used, but the mesh-free method has the disadvantage of incurring high numerical cost. Therefore, an adaptive method combining SPH and FEM-based SOLID elements is used for concrete target modeling to simulate excessive deformation and failure of the concrete target. In the adaptive coupling method of SPH and SOLID, the concrete target is first modeled as a solid element. When the damage of concrete exceeds the failure criterion, the solid element is eroded and the SPH element replacing the solid element is activated. Since the activated SPH element continues to participate in the impact, the problem of loss of materials and energy can be effectively solved. In this way, analysis results consistent with actual physical phenomena can be obtained.

In accordance with the Enforcement Decree of the Act on Physical Protection and Radiological Emergency, operators of Nuclear Power Plants (NPP)s must conduct full cyber security exercise once a year and partial exercise at least once every half year. Nuclear operators need to conduct exercise on systems with high attack attractiveness in order to respond to the unauthorized removal of nuclear or other radioactive material and sabotage of nuclear facilities. Nuclear facilities identify digital assets that perform SSEP (Safety, Security, and Emergency Preparedness) functions as CDA (Critical Digital Assets), and nuclear operators select exercise target systems from the CDA list and perform the exercise. However, digital assets that have an indirect impact (providing access, support, and protection) from cyber attacks are also identified as CDAs, and these CDAs are relatively less attractive to attack. Therefore, guidelines are needed to select the exercise target system in the case of unauthorized removal of nuclear or other radioactive material and sabotage response exercise. In the case of unauthorized removal of nuclear or other radioactive material, these situations cannot occur with cyber attacks and external factors such as terrorists must be taken into consideration. Therefore, it is necessary to identify the list of CDAs that terrorists can use for cyber attacks among CDAs located in the path of stealing and transporting nuclear material and conduct intensive exercise on these CDAs. A typical example is a security system that can delay detection when terrorists attack facilities. In the case of sabotage exercise, a safety-related system that causes an initiating event by a cyber attack or failure to mitigate an accident in a DBA (Design Basis Accident) situation should be selected as an exercise target. It is difficult for sabotage to occur through a single cyber attack because a nuclear facility has several safety concepts such as redundancy, diversity. Therefore, it can be considered to select an exercise target system under the premise of not only a cyber attack but also a physical attack. In the case of NPPs, it is assumed that LOOP (Loss of Offsite Power) has occurred, and CDA relationships to accident mitigation can be selected as an exercise target. Through exercise on the CDA, which is more associated with unauthorized removal of nuclear or other radioactive material and sabotage of nuclear facilities, it is expected to review the continuity plan and check systematic response capabilities in emergencies caused by cyber attacks.