http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
강치운,강석규 동의공업대학 1999 論文集 Vol.25 No.1
For the most part, the type of element used in the array system is known as microstrip patch antennas(MPAs). In simplest form, a MPAs is composed of a radiating patch seperated from a ground plane by a dielectric substrate. Their light weight and conformability have made MPAs very attractive for aircraft, missile, rocket and satellite applications. They also have other general advantages of MIC technology including low cost, ease of integration with other active components, and variety in design. Unfortunately, they suffer in general from narrowband performance, pattern distortion due to surface wave modes excited around the patch, and relatively large conductor losses which increase with frequency. In this paper, it is designed a microstrip patch antenna with double resonances to have a broadband impedance performance. In the design of the antenna, there are considered the input impedance, the width of patch, the total length of the slot, the height of dielectric substrate. the position of the probe and the radius of fed pin. The radiation pattern. impedance locus, and VSWR of the antenna are calculated and compared with the experimental results.
광대역 마이크로스트립 슬롯 결합 유전체 공진기 안테나 설계
강치운 동의공업대학 2001 論文集 Vol.27 No.1
In this paper, microstrip slot-coupled dielectric resonator antenna operating in TE^(x)_(III) mode is designed to enhance the bandwidth of antenna. The DR is fabricated from MgTiO_(3) ceramic with a dielectric constant of 21 and low-loss temperature stable material, and has dimension of 12(2a)×4(d)×6(b) mm. CAD tools for 3D finite element analysis structure simulation, HFSS is used to optimize characteristics of antenna. The microstrip slot-coupled dielectric resonator antenna within less than -l0dB return loss has been measured about 350 MHz (6.0%) bandwidth. Numerical results for the return loss, input impedance, and radiation pattern of the antennas are presented and compared with measurements, showing reasonable agreement. The results presented should be useful for the design of this type of antenna.
수분측정장치용 RF 센서의 Probe 모델 개발에 관한 연구
강석규,강치운 동의공업대학 1998 論文集 Vol.24 No.1
In this paper, three types of sleeves for the sensing antenna probe were modeled and the characteristics of near field radiation distribution above the ceramic face plate was analyzed for the optimum antenna-mounting place by a computer simulation method using SI Eminence. At the proper mounting place for sensing antenna in each of the sleeve models, the near field distribution characteristics showed a concentric radiation pattern. When the near field radiation pattern showed a concentric and constant distribution pattern on the ceramic face plate, the measurement accuracy of moisture percentage was high. The half-cylinder type of sleeve model with plate ceramic face, as the most conformable sleeve model, satisfied such a field distribution characteristics.
다중회귀신경회로망을 이용한 시계열패턴의 예측에 관한 연구
김광희,강치운 동의공업대학 1998 論文集 Vol.24 No.1
In this paper, the Multiple Recurrent Neural Network(MRNN). which is a dynamic neural network, is proposed and is applied to predict a time series. The prediction performance of the MRNN measured by the normalized mean square error(NMSE) is superior to that of the adaptive time-delay neural network(ATNN) which adapts its time-delays as well as its synaptic weights. The MRNN is based on the multi layer perceptrons(MLP) with two hidden layers and several feedback routes with different delays added to the MLP. In this configuration several previous activation values in the upper hidden layer are fed back into the lower hidden layer. The conventional error back-propagation is used to learn the MRNN. In the experiments we know that the network performance is more affected by the number of feedback routes than the number of nodes or layers. Since the prediction error of the MRNN with 14 feedback routes is smaller than that of the ATNN, we conclude that the added feedback routes improve the context learning ability of the network. However, there are many problems, which must be solved in the future concerning the dynamic behaviors of the MRNN. The architecture of the MRNN is simpler than that of the feedforward time-delay networks but the context learning ability of the MRNN can be better. The MRNN. therefore can be used to recognize time-varying patterns like speech signals in time domain without any other preprocessing.