http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
열화상 이미지와 환경변수를 이용한 콘크리트 균열 깊이 예측 머신 러닝 분석
김지형,장아름,박민재,주영규,Kim, Jihyung,Jang, Arum,Park, Min Jae,Ju, Young K. 한국공간구조학회 2021 한국공간구조학회지 Vol.21 No.2
This study presents the estimation of crack depth by analyzing temperatures extracted from thermal images and environmental parameters such as air temperature, air humidity, illumination. The statistics of all acquired features and the correlation coefficient among thermal images and environmental parameters are presented. The concrete crack depths were predicted by four different machine learning models: Multi-Layer Perceptron (MLP), Random Forest (RF), Gradient Boosting (GB), and AdaBoost (AB). The machine learning algorithms are validated by the coefficient of determination, accuracy, and Mean Absolute Percentage Error (MAPE). The AB model had a great performance among the four models due to the non-linearity of features and weak learner aggregation with weights on misclassified data. The maximum depth 11 of the base estimator in the AB model is efficient with high performance with 97.6% of accuracy and 0.07% of MAPE. Feature importances, permutation importance, and partial dependence are analyzed in the AB model. The results show that the marginal effect of air humidity, crack depth, and crack temperature in order is higher than that of the others.
다단 임팩터 Nanosampler를 이용한 진주시 대기에어로졸입자의 입경별 질량농도 특성
박정호 ( Jeong Ho Park ),장민재 ( Min Jae Jang ),김형갑 ( Hyoung Kab Kim ) 한국환경과학회 2015 한국환경과학회지 Vol.24 No.5
Atmospheric aerosol particles were investigated at GNTECH university in Jinju city. Samples were collected using the Nanosampler period from January to December 2014. The Nanosampler is a 6 stage cascade impactor(1 stage : > 10 μm, 2 stage : 2.5~10 μm, 3 stage : 1.0~2.5 μm, 4 stage : 0.5~1.0 μm, 5 stage : 0.1~0.5 μm, back-up : < 0.1 μm) with the stages having 50% cut-off ranging from 0.1 to 10 μm in aerodynamic diameter. The mass size distribution of Atmospheric aerosol particles was unimodal with peak at 1.0~2.5 μm or 0.5~1.0 μm. The annual average concentrations of TSP, PM10, PM2.5, PM1, PM0.5 and PM0.1 were 44.0 μg/m3, 40.3 μg/m3, 31.4 μg/m3, 18.0 μ g/m3, 8.2 μg/m3, 3.0 μg/m3, respectively. On average PM10, PM2.5, PM1, PM0.5 and PM0.1 make up 0.91, 0.70, 0.41, 0.19 and 0.07 of TSP, respectively. The annual average of PM2.5/PM10 ratio was 0.77.
박정호 ( Jeong Ho Park ),장민재 ( Min Jae Jang ),김형갑 ( Hyoung Kab Kim ) 한국산업보건학회 2016 한국산업보건학회지 Vol.26 No.1
Objectives: This study was performed to measure and evaluate the concentration, size distribution and fugitive emission of particulate matter from process operations at foundries. Methods: Particle matter was collected from three foundries, and samples were also collected from a background site for calculating the fugitive emission concentration of the foundries. For the collection of the samples, a Nanosampler cascade impactor was used. Results: The concentration of TSP in the samples collected from the three foundries was 0.675~1.222 mg/m(3), PM10 was 0.525~1.018 mg/m3 and PM(2.5) was 0.192~0.615 mg/m(3). The mass size distribution was bimodal or monomodal with maximum peak at two stage(size 2.5~10 μm). The mass median aerodynamic diameter(MMAD) was 1.80~3.98 μm. The fugitive emission concentration of TSP varies in the range of 0.65 to 1.21 mg/m(3), which exceeds the emission standard of fugitive dust(0.5 mg/m(3)). Conclusions: Particle concentration and size is an important industrial hygiene factor to protect foundry workers. Furthermore, the presence of high emission of particulate pollutants has a significant negative impact on the ambient air of the study area. Therefore, it is important to improve both the process and prevention facility in oder to reduce particulate pollutants in foundries.