• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2019년도 하계학술대회
• /
• pp.9-12
• /
• 2019

지난 수 년 동안 계속해서 일반 실상 카메라를 이용한 영상분석기술에 대한 연구가 활발히 진행되고 있다. 최근에는 딥러닝 기술을 적용한 지능형 영상분석기술로 발전해 왔으며 국방기지방호, CCTV, 사용자 얼굴인식, 머신비전, 자동차, 드론 산업이 활성화되면서 많은 시너지를 효과를 일으키고 있다. 그러나 어두운 밤과 안개, 날씨, 연기 등 다양한 여건에서 따라서 카메라의 영상분석 정확성 감소와 오류가 수반될 수 있으며 일반적으로 딥러닝 기술을 활용하기 위해서는 고사양의 GPU를 필요로 하기 때문에 다른 추가적인 시스템이 요구된다. 이에 본 연구에서는 열적외선 영상의 객체 검출에 적용하기 위해 SSD(Single Shot MultiBox Detector) 기반의 경량적인 MobilNet 네트워크로 재구성하여, 모바일 기기 등 낮은 사양의 낮은 임베디드 시스템에서도 활용 할 수 있는 방법을 제안한다. 모의 실험결과 제안된 방식의 모델은 적외선 열화상 카메라에서 객체검출과 학습시간이 줄어든 것을 확인 할 수 있었다.

• Nuclear Engineering and Technology
• /
• 제50권8호
• /
• pp.1372-1386
• /
• 2018

Despite recent advances in multi-hazard analysis, the complexity and inherent nature of such problems make quantification of the landslide effect in a probabilistic safety assessment (PSA) of NPPs challenging. Therefore, in this paper, a practical approach was presented for performing an earthquake-induced landslide PSA for NPPs subject to seismic hazard. To demonstrate the effectiveness of the proposed approach, it was applied to Korean typical NPP in Korea as a numerical example. The assessment result revealed the quantitative probabilistic effects of peripheral slope failure and subsequent run-out effect on the risk of core damage frequency (CDF) of a NPP during the earthquake event. Parametric studies were conducted to demonstrate how parameters for slope, and physical relation between the slope and NPP, changed the CDF risk of the NPP. Finally, based on these results, the effective strategies were suggested to mitigate the CDF risk to the NPP resulting from the vulnerabilities inherent in adjacent slopes. The proposed approach can be expected to provide an effective framework for performing the earthquake-induced landslide PSA and decision support to increase NPP safety.

• Nuclear Engineering and Technology
• /
• 제50권7호
• /
• pp.1168-1172
• /
• 2018

This study went beyond making an indicator simply based on theoretical arguments, and explored a wide spectrum of different types of perceptions about energy safety to make a concept of energy safety for the Korean society. The energy safety schemata of people can be divided into three types. Type1 is concern about multi-level risks-responsibility-centric, type2 is concern about security and personal burden-expertise-centric, and type3 is concern about health and personal burden-responsibility-centric. Questions were designed on the basis of the characteristics, differences and commonalities of the three types of perceptions, explored through the Q methodology, and Koreans' perception of nuclear safety was examined. Based on the results of this research the following components of trust in nuclear safety were derived, risk perception, responsibility, honesty, expertise and procedural justification. The items for specifically evaluating them were developed, and factor analysis was conducted, and as a result, the validity of each item was proven. The components of the nuclear safety trust indicator do not exist independently, but influence each other continuously through interactions. For this reason, rather than focusing on any one of them, laws and systems must be improved first so that they can move together in one big frame.

• Nuclear Engineering and Technology
• /
• 제50권8호
• /
• pp.1289-1297
• /
• 2018

Preliminary surface soil Derived Concentration Guideline Levels (DCGLs) were derived conforming to the Multi-Agency Radiation Site Survey and Investigation Manual (MARSSIM) procedure for the site release and reuse of Kori Unit 1 in Korea. Based on the decommissioning experiences of the U.S. nuclear power plants, a suite of residual radionuclides was determined, and uncertainties contributed to the resultant dose by the input parameters were quantified via the sensitivity analysis of parameters. The peak of the mean dose was obtained via the probabilistic analysis of the RESRAD (RESidual RADioactivity)-ONSITE code. Consequently, $DCGL_w$ of Kori Unit 1 in accordance with two scenarios, industrial worker and residential farmer scenario, were derived and the results were compared respectively with other NPPs. It could be used as a basic guideline for establishing regulatory standards for reuse planning, designing the site characterization surveys and implementing final status survey (FSS).

• 한국멀티미디어학회논문지
• /
• 제24권5호
• /
• pp.695-700
• /
• 2021

The Faster RCNN-based shellfish recognition algorithm is introduced for shellfish recognition studies that currently do not have any deep learning-based algorithms in a practical setting. The original feature extraction module is replaced by DenseNet, which fuses multi-level feature data and optimises the NMS algorithm, network depth and merging method; overcoming the omission of shellfish overlap, multiple shellfish and insufficient light, effectively solving the problem of low shellfish classification accuracy. In the complexifier test environment, the test accuracy was improved by nearly 4%. Higher testing accuracy was achieved compared to the original testing algorithm. This provides favourable technical support for future applications of the improved Faster RCNN approach to seafood quality classification.

• International Journal of Computer Science & Network Security
• /
• 제21권6호
• /
• pp.1-6
• /
• 2021

Recently, pre-trained convolutional neural network CNNs have been widely used and applied for medical image classification. These models can utilised in three different ways, for feature extraction, to use the architecture of the pre-trained model and to train some layers while freezing others. In this study, the ResNet18 pre-trained CNNs model is used for feature extraction, followed by the support vector machine for multiple classes to classify medical images from multi-classes, which is used as the main classifier. Our proposed classification method was implemented on Kvasir and PH2 medical image datasets. The overall accuracy was 93.38% and 91.67% for Kvasir and PH2 datasets, respectively. The classification results and performance of our proposed method outperformed some of the related similar methods in this area of study.

• Nuclear Engineering and Technology
• /
• 제54권4호
• /
• pp.1167-1174
• /
• 2022

Cast austenitic stainless steels (CASSs) are widely used as structural materials in the nuclear industry. The main drawback of CASSs is the reduction in fracture toughness due to long-term exposure to operating environment. Even though ultrasonic non-destructive testing has been conducted in major nuclear components and pipes, the detection of cracks is difficult due to the scattering and attenuation of ultrasonic waves by the coarse grains and the inhomogeneity of CASS materials. In this study, the ultrasonic signals measured in thermally aged CASS were discriminated for the first time with the simple ultrasonic technique (UT) and machine learning (ML) models. Several different ML models, specifically the K-nearest neighbors (KNN), Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP) models, were used to classify the ultrasonic signals as thermal aging condition of CASS specimens. We identified that the ML models can predict the category of ultrasonic signals effectively according to the aging condition.

• Nuclear Engineering and Technology
• /
• 제54권6호
• /
• pp.2262-2275
• /
• 2022

The Korea Atomic Energy Research Institute (KAERI) studied the sodium-water reaction (SWR) minimized steam generator for the safety of the sodium-cooled fast reactor (SFR), and selected the copper bonded steam generator (CBSG) as the optimal concept. This paper introduces the conceptual design of the CBSG and the development of the CBSG sizing analyzer (CBSGSA). The CBSG consists of multiple heat transfer modules with a crossflow heat transfer configuration where sodium flows horizontally and water flows vertically. The heat transfer modules are stacked along a vertical direction to achieve the targeted large heat transfer capacity. The CBSGSA code was developed for the thermal-hydraulic analysis of the CBSG in a multi-pass crossflow heat transfer configuration. Finally, we conducted a preliminary sizing and rating analysis of the CBSG for the trans-uranium (TRU) core system using the CBSGSA code proposed by KAERI.

• Nuclear Engineering and Technology
• /
• 제54권4호
• /
• pp.1185-1194
• /
• 2022

The maintenance of condenser main pipe is the key to achieve film-forming amine maintenance effectiveness. In this work, oxygen content, pH and temperature of the solution were controlled to simulate the condition of condenser main pipe, and magnetite coated carbon steel sample was prepared by pre-oxidization. CAM was used to characterize the hydrophobicity of film formed samples. Hydrophobic film was formed on pre-oxidized carbon steel samples when octadecylamine concentration reaches 20 mg/kg. SEM, EDS, EIS, and PD were used to characterize the influence of octadecylamine concentration on maintenance effectiveness. It was found that the maintenance effectiveness was enhanced and the corrosion rate was suppressed with the increase of octadecylamine concentration. FIB and TEM were used to detect the adsorbed octadecylamine film thickness founding that octadecylamine adsorbed onto the surface of pre-oxidized carbon steel by multi-layer adsorption mechanism.

• Nuclear Engineering and Technology
• /
• 제53권1호
• /
• pp.30-43
• /
• 2021

In the 2D/1D method, a global adjoint CMFD based on the generalized equivalence theory is built to synthesize the 2D radial MOC adjoint and 1D axial NEM adjoint calculation and also to accelerate the iteration convergence of 3D whole-core adjoint transport calculation. Even more important, an advanced yet accurate two-level (TL) CMFD acceleration technique is proposed, in which an equivalent one-group adjoint CMFD is established to accelerate the multi-group adjoint CMFD and then to accelerate the 3D whole-core adjoint transport calculation efficiently. Based on these method, a new code is developed to perform 3D adjoint neutron flux calculation. Then a set of VERA and C5G7 benchmark problems are chosen to verify the capability of the 3D adjoint calculations and the effectiveness of TL CMFD acceleration. The numerical results demonstrate that acceptable accuracy of 2D/1D adjoint calculations and superior acceleration of TL CMFD are achievable.