http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
서동신 金烏工科大學校 2000 論文集 Vol.21 No.-
An electronic circuit is proposed to drive a plasma display panel(PDP). This circuit utilizes inductors to resonate with the equivalent intrinsic capacitance of the PDP to avoid the abrupt charging and discharging operation. The energy losses due to conventional hard switching driving and the displacement current of the PDP are mainly recovered with a quite simple structure.
서동신 國立金烏工科大學校附設生産技術硏究所 1988 産業技術開發硏究 Vol.4 No.-
Linear Predictive Analysis was turned out to be the most powerful technique estimating speech parameters e.g., pitch, formants and spectrum which can be transmitted and stored as a speech information. The results of this method have shown higher verification and rate of recognition together with the values of poles, frequency and bandwidth.
신경회로망을 이용한 ATM에서 실시간 호 제어에 관한 연구
서동신,김희철,나상동 조선대학교 생산기술연구소 1996 生産技術硏究 Vol.18 No.2
A traffic control using the neuaral network (NN) algorithm is proposed for the ATM communication networks. In this paper, neuaral network algorithm is trained to estimate call admission rate from traffic load and link capacity, and link capacity assignment is optimized by advanced NN algorithm method which used learning rate and moment term. In the real-time call control, link capacity assignment of NN methed is compared to equivalent bandwidth(EB) method. Therefore. simulation results yield efficient ATM traffic control which is used neural networks training between quality of service and traffic parameter in the number of traffic type 1, 2. We Show that call reject rate by neural network algorithm using the NN methed is presented better than EB method in the training pattern.
신경망을 이용한 ATM 망에서 트래픽 제어에 관한 연구
서동신,김희철,임방택,나상동 조선대학교 동력자원연구소 1996 動力資源硏究所誌 Vol.18 No.1
A traffic control using back-propagation neural networks propose for the ATM communications networks. This paper proposes adaptive call admission control using back-propagation algorithm in link capacity control. In this paper, back-propagation algorithm is trained to estimate call admission rate from traffic load and link capacity, and link capacity assignment is optimized by back-propagation algorithm method which used learning rate and moment term. Therefore, simulation results yield efficient ATM traffic control which use neural networks training between quality of service(QOS) and traffic parameter in the number of class 1, 2 and evaluate call loss rate 10exp(-6) using the Erlang-Bequation by trained back-propagation neural networks.
서동신 金烏工科大學校 1991 論文集 Vol.12 No.-
Input signals encountered by DSP are continuous and must first be digitized. Since the hardware complexity of DSP systems is directly related to the digital wordlength, it is important to limit the number of bits in the various processing elements of the system. This thesis is focused on the systems that are linear with respect to the field of real number from 4bits to 2bits which have been used in error-correcting codes, in multiple address coding, counting, fault detection, synchronization, in noise signal processing, in radar or sonar systems and spread-spectrum communications such that two digit length digital filter is studied for its maximum availability to digital application systems.
Least Square FIR System Identification의 실현에 관한 연구
徐東信 金烏工科大學校 1986 論文集 Vol.7 No.-
Minimizing the squared error between the measured system output and an estimate from an FIR system output performs the FIR system Identification, a set of least squares normal equations to be solved for the FIR system coefficients is obtained. If the assumed FIRsystem is of duration M samples, the usual solution for the M least squares normal equations requires a number of computational operations proportional to M³and storageof normal equation coefficients proportional to M². The set of normal equations has an underiying structure, however, that can be exploited to yield a solution with computational operations proportional to M²and storage proportional to M. In this paper several important aspects of the implementation for the problems of spectral estimation and system identification and the computational effects of the various analysis parameters are presented.
有限狀態機械 및 Automation 의 入出力 構造에 관한 연구
徐東信 金烏工科大學校 1985 論文集 Vol.6 No.-
유한상태기계에서 규정되는 制御入力 및 出力은 특별한 의미를 지니고 있기때문에, 이制御入出力의 性質을 연구해 본다는 것은 매우 重要한 일이다. 이 制御入出力의 性質은 比較 說明하기 위해서 다른 모델과의 變換過程을 例로 들었고, 또 몇가지 오토메이톤의 類型도 紹介하였으나, 결과적을 기계가 受理하는 制御入出力은 機械의 內部構造와 密接하게 關係되는 機械從屬的 入力 및 出力構造를 갖고 있음을 밝혔다. The knowledge of the Characteristics of the Control input and Output of the finite state machine has been highly emphasized. Several transform processes of different types of Automatons are processes of different types of Automatons are illustrated. It is proved that Control inputs fed to the machines have machine - dependent structure which are closely associated with the interior structure of the machine.
Adaptive Delta Modulation 에 대한 Hybrid Companding 에 관한 연구
서동신 金烏工科大學校 1982 論文集 Vol.3 No.-
Computer simulation for the comparison with the Hbqrid Companding and other ways when the variance of amplitude as an input is high, shows the results here that for the responses on high amplitude variance, hybrid companding is much superior to the instantaneous companding, the ratio αtends to arise great influence, and improved about 2.7 dB of the response on the input with gaussian distribution increasing the noise at the high step size.