By way of the practical use of the electrical fishing in salt water, all the data I could collect were carefully compared from the earliest kuehne's studies for protoplasm to the precent kreutzer's for electrotaxis of aquatic animals. The consequences are as follows: All the cold blooded animals are forced to move toward the positive pole in the path of an electrical current flowing from pole to pole which has brought about the possibilities of electrical fishing. This method has been found to be practical in fresh water, but there are some obstacles in utilizing the method in salt water especially because of the large amount of electric power consumed. In order to remedy the large power consumption, an intermittent current or pulsive current which stimulate the fish most effectively has been tested, but a few electrical quantative values in detail as to the current intensity and wave forms or any quantative results of these experiments have been made public. According to kreutzer's theory and the above tests, the synthetic observations lead to the conclusion that: (1) The was form of the pulsive current which gives the fish an electric shock most effectively is an almost right angled triangle with a vertical rise from null and a much slower decay. (2) The pulses are let repeatedly after recovering shock on the fish as following figure to minimize the large amount of electric power consumption, the recovering provides the next shock effectively by Weber-Fechner's law on psychophysics. Though wave head rises highly vertical, the decaying wave trails and the inter-mittence of electric current prevent the large consumption of power. (3) The electrotaxis rate among the school of fish is given by Tauti's formula. r=1/√2π σ∫o A log v/voexp{-1/2(x-xo/σ)2dx Where r is an electrotaxis rate, xo a certain sensibility of fish caused at arbitrary voltage stimulation made electrotaxis, x a mean value of the sensibility of fish at the front of fish school, and A log v/vo a mean value of sensibitity of fish at where the voltage is v by Weber-Fechner's law, where vo is an original voltage, A a constant, and σ a standard deviation. The electrotaxis rates are given as above if we regards the distribution iof electrotaxis rate as normal.

딥러닝(Deep Learning) 기술은 이미지 데이터를 비롯하여 텍스트 데이터, 음성 데이터 등을 학습시켜 특성을 추출하고 인식하기 위한 여러 분야에 적용하고 연구되고 있다. 내부에 존재하는 블레이드는 본체와 분리가 불가능하고, 내부의 매우 불리한 환경속에서 검출이 이루어져야 한다. 기존의 영상 검출 방법은 상당한 시간이 요구되며, 기술자들의 개인적 능력과 경험에 의존하고 있다. 본 연구에서는 내부 블레이드의 표면 결함을 효율적으로 검출하고 자동화하기 위하여 Faster R-CNN 알고리즘을 학습시켜 검출 모델을 구축하였다.

The damage detection method of blade systems largely depends on the personal ability of an inspector using a camera. Thus, this paper proposes a deep learning-based detection method that can rapidly and reliably identify and evaluate the damages on the blades.

This study was aimed to investigate the relationship between flood damage and rainfall in the downstream areas of Han River. We used flooding damage and daily rainfall data observed in the 12 downstream areas during 2000-2012. Rainfall variables were analysed which include total rainfall over the basin of downstream areas in the Han River for flooding periods, maximum accumulative rainfall over 1hours, 3 hours, 6 hours and 12 hours, and rainfall in the upper regions of the Han River. Flooding damages of the study areas have signiÿcant positive correlations with rainfall variables. The most significant variable that affect the amount of damage is 3 hour accumulative precipitation. It means that the more three-hour precipitation increases, the more the amount of damage also increases. We found that there is a high possibility of floods when a heavy rainfall day continues for more than 2 days by comparing heavy rainfall days between with and without flooding disasters. The possibility of flood decreases when interval of just before a heavy rainfall day and flooding period increase.