Recently, deep learning that is the most popular and effective class of machine learning algorithms is widely applied to various industrial areas. A number of research on various topics about structural engineering was performed by using artificial neural networks, such as structural design optimization, vibration control and system identification etc. When nonlinear semi-active structural control devices are applied to building structure, a lot of computational effort is required to predict dynamic structural responses of finite element method (FEM) model for development of control algorithm. To solve this problem, an artificial neural network model was developed in this study. Among various deep learning algorithms, a recurrent neural network (RNN) was used to make the time history response prediction model. An RNN can retain state from one iteration to the next by using its own output as input for the next step. An eleven-story building structure with semi-active tuned mass damper (TMD) was used as an example structure. The semi-active TMD was composed of magnetorheological damper. Five historical earthquakes and five artificial ground motions were used as ground excitations for training of an RNN model. Another artificial ground motion that was not used for training was used for verification of the developed RNN model. Parametric studies on various hyper-parameters including number of hidden layers, sequence length, number of LSTM cells, etc. After appropriate training iteration of the RNN model with proper hyper-parameters, the RNN model for prediction of seismic responses of the building structure with semi-active TMD was developed. The developed RNN model can effectively provide very accurate seismic responses compared to the FEM model.

가뭄은 일반적으로 장기간에 걸쳐 물 공급이 부족하여 나타나는 환경 재앙 중 하나로 대부분 넓은 지역에 걸쳐 나타난다. 원격탐사 자료는 이러한 넓은 지역에서 나타나는 가뭄 모니터링에 적합한 방법이다. 따라서 이 연구에서는 강원도 소양호 지역의 Landsat 위성 영상 자료를 활용하여 약 30년(1985-2015) 동안의 소양호 면적을 산출하고 이를 가뭄 패턴과 분석하였다. 특히 ISODATA, Maximum likelihood 및 인공신경망을 활용하여 Landsat 영상을 분류하여 소양호 면적을 산출하였다. 또한 가뭄 패턴을 분석하기 위하여 산출된 호수 면적과 소양호 지역의 강수량을 활용한 표준 강수지수(Standardized Precipitation Index: SPI)와의 상관관계를 분석하였다. 영상 분류 연구 결과, ISODATA, Maximum likelihood 및 인공신경망 방법 중에서 호수 면적 산출의 최적의 방법은 인공신경망 방법임을 알 수 있었다. 또한, 인공신경망 방법을 적용하여 산출한 호수 면적과 SPI와의 상관관계 분석 결과 R 2 값이 0.52를 가진다. 즉, SPI 지수가 낮을 때 호수 면적이 감소하는 것을 알 수 있었다. 즉 호수 면적 변화를 통하여 소양호 지역의 가뭄 상태 감지 및 모니터링이 가능하다는 것을 알 수 있었다. 이 연구는 향후 지역 가뭄 모니터링 프로그램 개발 등에 사용이 가능할 것이다.

Vegetation segmentation in a field color image is a process of distinguishing vegetation objects of interests like crops and weeds from a background of soil and/or other residues. The performance of the process is crucial in automatic precision agriculture which includes weed control and crop status monitoring. To facilitate the segmentation, color indices have predominantly been used to transform the color image into its gray-scale image. A thresholding technique like the Otsu method is then applied to distinguish vegetation parts from the background. An obvious demerit of the thresholding based segmentation will be that classification of each pixel into vegetation or background is carried out solely by using the color feature of the pixel itself without taking into account color features of its neighboring pixels. This paper presents a new pixel-based segmentation method which employs a multi-layer perceptron neural network to classify the gray-scale image into vegetation and nonvegetation pixels. The input data of the neural network for each pixel are 2-dimensional gray-level values surrounding the pixel. To generate a gray-scale image from a raw RGB color image, a well-known color index called Excess Green minus Excess Red Index was used. Experimental results using 80 field images of 4 vegetation species demonstrate the superiority of the neural network to existing threshold-based segmentation methods in terms of accuracy, precision, recall, and harmonic mean.

This study was conducted as part of a series of studies to introduce the Convolutional Neural Network(CNN) into the diagnostic field of osteoporosis. The purpose of this study was to compare the results when testing Digital Radiography(DR) and Computed Radiography(CR) panoramic radiographs by CNN that were trained by DR panoramic radiographs. The digital panoramic radiographs of females who visited for the purpose of diagnosis and treatment at Chonnam National University Dental Hospital were taken. Two Oral and Maxillofacial Radiologists were selected for the study to compare the panoramic radiographs with normal and osteoporosis images. Among them, 1068 panoramic radiographs of females{Mean [± standard deviation] age: 49.19 ± 21.91 years} obtained by DR method were used for training of CNN. 200 panoramic radiographs of females{Mean [± standard deviation] age: 63.95 ± 6.45 years} obtained by DR method and 202 panoramic radiographs of females{Mean [± standard deviation] age: 62.00 ± 6.86 years} obtained by CR method were used for testing of CNN. When the DR panoramic radiographs were tested, the Accuracy was 92.5%. When the CR panoramic radiographs were tested, the Accuracy was 76.2%. It can be seen that the CNN trained by DR panoramic radiographs is suitable to be tested with the same DR panoramic radiographs.

화재의 초기 검출은 인명과 재화의 손실을 최소화하기 위한 중요한 요소이다. 불꽃과 연기를 신속하면서 동시에 검출해야 하며 이를 위해 영상 기반의 화재 검출에 관한 연구가 다양하게 진행되고 있다. 기존의 화재 검출은 불꽃과 연기의 특징을 추출하기 위해 여러 알고리즘을 거쳐서 화재의 검출 유무를 판단하므로 연산량이 많이 소모되었으나, 딥러닝 알고리즘인 합성곱 신경망을 이용 하면 별도의 과정이 생략되므로 신속하게 검출할 수 있다. 본 논문에서는 선박 기관실에서 화재 영상을 녹화한 데이터로 실험을 수행 하였다. 불꽃과 연기의 특징을 외각 상자로 추출한 후 합성곱 신경망 중 하나인 욜로(YOLO)를 이용하여 학습하고 결과를 테스트하였 다. 실험 결과를 검출률, 오검출률, 정확도로 평가하였으며 불꽃은 0.994, 0.011, 0.998, 연기는 0.978, 0.021, 0.978을 나타내었고, 연산시간 은 0.009s를 소모됨을 확인하였다.

This study aimed to test a convolutional neural network (CNN) in two different settings of training and testing data. Panoramic radiographs were selected from 1170 female dental patients (mean age 49.19 ± 21.91 yr). The cortical bone of the mandible inferior border was evaluated for osteoporosis or normal condition on the panoramic radiographs. Among them, 586 patients (mean age 27.46 ± 6.73 yr) had normal condition, and osteoporosis was interpreted on 584 patients (mean age 71.00 ± 7.64 yr). Among them, one data set of 569 normal patients (mean age 26.61 ± 4.60 yr) and 502 osteoporosis patients (mean age 72.37 ± 7.10 yr) was used for training CNN, and the other data set of 17 normal patients (mean age 55.94 ± 4.0 yr) and 82 osteoporosis patients (mean age 62.60 ± 5.00 yr) for testing CNN in the first experiment, while the latter was used for training CNN and the former for testing CNN in the second experiment. The error rate was 15.15% in the first experiment and 5.14% in the second experiment. This study suggests that age-matched training data make more accurate testing results.

Self Organizing Map (SOM) is a neural network that is effective in classifying patterns that form the feature map by extracting characteristics of the input data. In this study, we propose an algorithm to determine the cell formation and the machine layout within the cell for the cell formation problem with operation sequence using the SOM. In the proposed algorithm, the output layer of the SOM is a one-dimensional structure, and the SOM is applied to the parts and the machine in two steps. The initial cell is formed when the formed clusters is grouped largely by the utilization of the machine within the cell. At this stage, machine cell are formed. The next step is to create a flow matrix of the all machine that calculates the frequency of consecutive forward movement for the machine. The machine layout order in each machine cell is determined based on this flow matrix so that the machine operation sequence is most reflected. The final step is to optimize the overall machine and parts to increase machine layout efficiency. As a result, the final cell is formed and the machine layout within the cell is determined. The proposed algorithm was tested on well-known cell formation problems with operation sequence shown in previous papers. The proposed algorithm has better performance than the other algorithms.

Objectives of this study were to identify the hotspot for displacement of the on-line water quality sensors, in order to detect illicit discharge of untreated wastewater. A total of twenty-six water quality parameters were measured in sewer networks of the industrial complex located in Daejeon city as a test-bed site of this study. For the water qualities measured on a daily basis by 2-hour interval, the self-organizing maps(SOMs), one of the artificial neural networks(ANNs), were applied to classify the catchments to the clusters in accordance with patterns of water qualities discharged, and to determine the hotspot for priority sensor allocation in the study. The results revealed that the catchments were classified into four clusters in terms of extent of water qualities, in which the grouping were validated by the Euclidean distance and Davies-Bouldin index. Of the on-line sensors, total organic carbon(TOC) sensor, selected to be suitable for organic pollutants monitoring, would be effective to be allocated in D and a part of E catchments. Pb sensor, of heavy metals, would be suitable to be displaced in A and a part of B catchments.

이 연구의 목적은 인공신경망 기법을 이용하여 사면의 내진 성능을 비교적 정확하면서도 효율적으로 예측하는 모델을 도 출하는데 있다. 사면의 내진 성능은 지진입력 및 사면모델의 무작위성 및 불확실성으로 인하여 정량화하기 쉽지 않다. 이러한 배경 아래 사면에 대한 확률론적 지진 취약도 분석이 몇몇 연구자에 의해 수행되었고, 이를 기반으로 다중 선형회귀분석 을 통하여 사면 내진성능에 대한 닫힌식이 제안된 바 있다. 그러나 전통적인 통계학적 선형회귀분석은 다양한 조건의 사면과 이에 따른 내진 성능 사이의 비선형적 관계를 정확하게 표현하지 못하는 한계를 보였다. 이에 따라 본 연구에서는 이러한 문제점을 극복하고자 인공신경망 기법을 사면 내진성능 예측 모델을 생성하는데 적용하였다. 도출된 모델의 유효성은 기존 의 다중 선형 및 다중 비선형 회귀분석을 통한 모델과 비교하여 검증하였다. 결과적으로 이전 연구의 전통적인 통계학적 회귀 분석을 통한 모델과 비교 결과, 기본적으로 인공신경망 기법을 통하여 도출된 모델이 사면의 내진성능을 예측하는데 있어 우수한 성능을 보여주었다. 이러한 정확도 높은 모델은 향후 확률에 기반한 사면의 지진취약도 지도를 개발하고, 주요 구 조물의 인근 사면으로 인한 리스크를 효과적으로 평가하는데 활용될 수 있을 것이라 기대된다.

This research is about a study on the flow stress of Inconel 601 under hot deformation. For Inconel 601, hot compression tests on gleeble 3500 system under 925℃, 1050℃ and 1150℃ and 0.001/s, and 5/s of strain rates were done. The flow behavior of the Inconel 601 was studied and modeled. In this study, the flow stress was modeled using deep neural network and support vector regression algorithm. The flow stress of Inconel 601 was dependent on strain rate and temperature. It was found that both the deep neural network and support vector regression adequately described the flow stress variation of Inconel 601. However, the model by the support vector regression was found to be superior to the model by the deep neural network. The construction of the model by SVR was more efficient than the construction by DNN. Also the prediction accuracy of the model by SVR was better than the accuracy of the model by DNN. It is found that the MAPE(Mean absolute percentage error) of the DNN based model was 4.89% while the MAPE of the SVR based model was 1.98%.

픽셀 아트는 낮은 해상도와 제한된 색 팔레트를 가지고 영상을 표현한다. 픽셀 아트는 낮은 연 산 성능과 적은 저장 공간을 가지는 초기 컴퓨터 게임에서 주로 사용되었다. 현대에 이르러, 픽셀 아트는 예술이나 퍼즐, 게임과 같은 다양한 분야에서 찾아볼 수 있게 되었다. 본 논문에서는 게임 캐릭터 영상을 입력으로 받는 픽셀 아트 생성 모델을 제안한다. 기존 방법 과는 달리, 합성곱 신경망(CNN:Convolutional-Neural Network)를 픽셀 아트 생성 목적에 맞게 변형하여 이를 이용하는 방법을 제시한다. 기존의 합성곱 연산 후에 upsampling 과정을 추가하여 픽셀 아트가 생성될 수 있도록 하였다. 네트워크는 ground truth와 생성된 픽셀 아트와의 평균 오차 제곱(MSE:Mean Squared Error)을 최소화해나가며 학습을 수행한다. Ground truth는 실제 아티스트가 생성하도록 하였고, 이미지 회전과 반전 기법을 이용하여 augumentation을 수행하였다. 생성된 데이터 집합은 학습, 검증, 시험 데이터로 나누었다. 이러한 데이터 집합을 기반으로 감독 학습을 실시하여 픽셀 아트 생성 네트워크를 학습하였다. 학습 모델의 학습 과정과 학습 정확도를 제시하고, 시험 데이터 뿐만 아니라 다양한 영상에 대한 픽셀 아트 결과도 함께 제시한다.

In this paper, we propose an Elman recurrent neural network to predict and analyze a time series of gas energy consumption in an air handling unit. To this end, we consider the volatility of the time series and demonstrate that there exists a correlation in the time series of the volatilities, which suggests that the gas consumption time series contain a non-negligible amount of the non-linear correlation. Based on this finding, we adopt the Elman recurrent neural network as the model for the prediction of the gas consumption. As the simplest form of the recurrent network, the Elman network is designed to learn sequential or time-varying pattern and could predict learned series of values. The Elman network has a layer of “context units” in addition to a standard feedforward network. By adjusting two parameters in the model and performing the cross validation, we demonstrated that the proposed model predicts the gas consumption with the relative errors and the average errors in the range of 2%~5% and 3kWh~8kWh, respectively. The results of this study can be used to the energy management system in terms of the effective control of the cross usage of the electric and the gas energies.

The contemporary high-tech structures have become enlarged and their functions more diversified. Steel concrete structure and composite material structures are not exceptions. Therefore, there have been on-going studies on fiber reinforcement materials to improve the characteristics of brittleness, bending and tension stress and others, the short-comings of existing concrete. In this study, the purpose is to develop the estimated model with dynamic characteristics following the steel fiber mixture rate and formation ration by using the nerve network in mixed steel fiber reinforced concrete (SFRC). This study took a look at the tendency of studies by collecting and analyzing the data of the advanced studies on SFRC, and facilitated it on the learning data required in the model development. In addition, by applying the diverse nerve network model and various algorithms to develop the optimal nerve network model appropriate to the dynamic characteristics. The accuracy of the developed nerve network model was compared with the experiment data value of other researchers not utilized as the learning data, the experiment data value undertaken in this study, and comparison made with the formulas proposed by the researchers. And, by analyzing the influence of learning data of nerve network model on the estimation result, the sensitivity of the forecasting system on the learning data of the nerve network is analyzed.

감성은 복잡하고 다양한 요인들에 의해 영향을 받기 때문에 다각적인 측면에서 고려되어야 한다. 본 연구에서는 심리 평가 척도의 하나인 각성(arousal) 지표와 다중 생체신호에서 추출된 생체지표 반응을 이용하여 중립 및 부정감성(슬픔, 공포, 놀람)의 분류하였다. 이를 위하여 감성에 따른 생체지표 반응의 차이를 확인하였고, 다중 신경망 알고리즘 기반의 감성 인식기를 적용하여 이들 감성이 얼마나 정확하게 분류되는가를 확인하였다. 총 146명의 실험 참가자(평균 연령 20.1±4.0, 남성 41%)를 대상으로 감성 유발 자극을 제시하고 동시에 생체신호(심전도, 혈류맥파, 피부전기활동)를 측정하였다. 또한 감성 유발 자극에 대한 심리 반응을 감성 평가 척도로 평가하였다. 측정된 생체신호에서 심박률(HR), NN 간격의 표준편차(SDNN), 혈류량(BVP), 맥파전달시간(PTT), 피부전도수준(SCL), 피부전도반응 (SCR)을 추출하였다. 결과 분석을 위하여 감성 자극에 대한 각성도와 안정 상태와 감성 상태의 생체지표 반응을 활용하였다. 또한 감성 분류를 위하여 다중 신경망 기반의 감성 인식기를 활용하였다. 그 결과, 감성에 따른 생체지표 반응의 차이를 확인하였고, 이들 감성의 분류 성능은 각성도와 모든 생체지표 특징들을 조합하였을 때 정확도가 가장 높음(86.9%)을 확인하였다. 본 연구는 심리 및 생체지표 추출과 기계학습 기술의 적용을 통하여 부정 감성을 분류할 수 있음을 제안하며, 이는 인간의 감성을 탐지하는 감성 인식 기술을 확립하는데 기여할 것으로 예상한다.

In this paper, we propose an Elman recurrent neural network to predict and analyze a time series of power energy consumption. To this end, we consider the volatility of the time series and apply the sample variance and the detrended fluctuation analyses to the volatilities. We demonstrate that there exists a correlation in the time series of the volatilities, which suggests that the power consumption time series contain a non-negligible amount of the non-linear correlation. Based on this finding, we adopt the Elman recurrent neural network as the model for the prediction of the power consumption. As the simplest form of the recurrent network, the Elman network is designed to learn sequential or time-varying pattern and could predict learned series of values. The Elman network has a layer of “context units” in addition to a standard feedforward network. By adjusting two parameters in the model and performing the cross validation, we demonstrated that the proposed model predicts the power consumption with the relative errors and the average errors in the range of 2%~5% and 3kWh~8kWh, respectively. To further confirm the experimental results, we performed two types of the cross validations designed for the time series data. We also support the validity of the model by analyzing the multi-step forecasting. We found that the prediction errors tend to be saturated although they increase as the prediction time step increases. The results of this study can be used to the energy management system in terms of the effective control of the cross usage of the electric and the gas energies.