큰볼루션 부호를 위한 비터비 복호 변환 알고리즘

전영희,이용학,김장형 濟州大學校工科大學附屬産業技術硏究所 1994 尖端技術硏究所論文集 Vol.5 No.-

In many applications researchers have been interested in implementation of Viterbi decoding algorithm for convolution codes. But it is problems that long constraint length codes need many storages and hardware implementation is complicated. In this paper, several stages of trellis diagram are integrated and this procedures are formulated into simple matrix operations. Also backtracking decoding procedures is eliminated. This algorithm is called a new Viterbi transform(VT). The number of the trellis search can be reduced using the VT. Because of local memory access, the storages can be taken in. Long constraint length codes can be decoded by combining the unit module processors.

이중면 불연속 구형 도파관의 특성 해석

강부식,장재성,이청민 濟州大學校工科大學附屬産業技術硏究所 1993 尖端技術硏究所論文集 Vol.4 No.-

Double-plane steps discontinuity in rectangular waveguide is investigated with ?? mode analysis. This method accounts for the influence of evanescent fields and power transmission due to higher order modes. For these discontinuites the electromagnetic fields are derived and the scattering matrix is derived by applying the continuous condition in discontinuity. The scattering coefficients of a X-band to Ku-band waveguide transition as well as of resonant iris with finite thickness(t=2.0㎜) are calculated and compared with Patzelt's result. The scattering coefficients of the resonant iris(a=7.112㎜, b=3.556㎜, a₁=4.0㎜, b₁=1.0㎜, t=0.5㎜) are calculated and compared with Bornemann's results.

Design of Multistage Optimal Controller for Nonlinear Systems Based on Multilayer Neural Networks

Kim,Kyung-youn,Bae,Byeong-woo 濟州大學校工科大學附屬産業技術硏究所 1993 尖端技術硏究所論文集 Vol.4 No.-

In this paper we describe a method to solve the optimal control problem of nonlinear systems using neural networks which is called here as a two-level multilayer neural network (TLMNN). the TLMNN has internal networks which consist of an upper-level neural network(UNN) for modeling and a lower-level neural network(LNN) for control. The weights of each network is optimized by introducing an augmented function so that a performance measure is minimized. Some examples were used to illustrate the characteristics of the proposed algorithm.

Shape Invariant Korean Character Recognition Using Optical Associative Memory

Kim, Jeong Woo,Kim, Soo Joong,Doh, Yang Hoi 濟州大學校工科大學附屬産業技術硏究所 1994 尖端技術硏究所論文集 Vol.5 No.-

글자모양의 변형에 무관하게 인쇄체 한글문자를 인식하기 위하여 홀로그램을 이용한 광 연상 기억시스템을 제안하였다. 제안된 인식시스템의 구조는 궤환을 가지는 단층 신경회로망이며, 광기억 루프를 구성하기 위하여 종속접속된 Vander Lugt 상관기를 사용하였다. 첫번째 홀로그램 상관기는 변별력을 높이기 위해 MACE 필터를 이용하여 만들었으며, 인식시스템의 규모를 줄이기 위하여 4개의 MACE 필터를 하나로 다중화 하였다. 이리하여 4개의 MACE 필터와 10개의 모음 MACE 필터를 각각 4개 및 3개의 다중 MACE 필터로 합성할 수 있었다. 한편 두번째 홀로그램 상관기는 원래 영상을 그대로 복원할 수 있도록 광대역 특성을 가지는 CMF를 이용하여 만들었다. 컴퓨터 시뮬레이션을 통하여 제안된 시스템이 글자모양의 변형에 무관하고, 변별력이 우수한 특성을 가짐을 확인하였다. A holographic implementation of an optical associative memory system for shape invariant recognition of the printed Korean character is proposed. The structure of the recognition system is a single-layer neural network employing feedback. To form an optical memory loop, we use cascaded Vander Lugt correlators. The first holographic correlator is made of MACE filter having a high discrimination capability. And to reduce the size of the recognition system, we multiplex four MACE filters into one. Thus, 14 consonant MACE filters and 10 vowel MACE filters can be synthesized into four consonant MMACE filters and three vowel MMACE filters, respectively. The second holographic correlator, on the other hand, is made of a CMF having broadband characteristics so that the reconstructed image has high fidelity. Computer simulation results show that the proposed system has shape invariance and high discrimition capability.

RANDOM 진동하중을 받는 구조물용 재료의 피로파괴에 대한 기초적 연구

김귀식,조성배 濟州大學校工科大學附屬産業技術硏究所 1990 尖端技術硏究所論文集 Vol.1 No.-

This paper was to investigate a fatigue damage of structural material under service loading, random loading. This examined irregular factors(α), wave counting methods and equivalent load amplitudes about three random loading waves generated by a personal computer. Results obtained are follows. 1. It is greatly different of counting results between range and rainflow cycle counting methods applied wide band spectrum, but it is similar those applied narrow bane one. 2. A Equivalent load amplitude, Leq, is obtained from m, the index of Paris' rule on a constant load amplitude test, and the result of cycle counting. Here, m is larger, Leq is greater. 3. In relation between irregular factor(α) and the cycle method, when α is near 1.0, Leq calculated by rainflow method is similarly that by range one, but when α is lower than 0.99, the former is large rather than the latter.

食品工場廢水의 活性汚泥處理中 오존에 依한 酸化分解效果

康京壽,崔永贊,宋大鎭,河璡桓 濟州大學校工科大學附屬産業技術硏究所 1991 尖端技術硏究所論文集 Vol.2 No.-

잔류염소를 함유한 發水를 活性汚泥에 의한 處理效率을 增加시키기 위하여 전압 100Volt, 유속 10ℓ-air/min, 0.8g-O₃/hr의 오존으로 처리했을때 COD의 除去效果및 투과율의 變化를 實驗한 結果를 要約하면 다음과 같다. 1. 잔류염소를 함유한 시료를 0.8g-O₃/hr의 오존으로 處理하였을때 pH의 變化 및 反應時間에 의한 COD의 除去率은 中性溶液(pH=7.2)에서 오존處理時間 50分 후 35%의 除去效果를 나타내어 酸性溶液(pH=3.2)의 20%, 알칼리性溶液(pH=12.0)의 10%의 除去效果보다 좋았다. 2. 잔류염소를 함유한 試料를 오존으로 50分間 處理하여 F/M比 0.1㎏·BOD/㎏·MLSS로 한 후 3. 6. 9 및 15時間 活性汚泥로 反應시켰을 때 反應時間 6時間 후 85.8%의 COD의 除去效果가 있어 티오황산나트륨(Na₂S₂O₃)으로 處理한 試料의 除去率 60.5%보다 좋았으며, 反應時間의 단축효과도 있었다. 3. Polyphenol色素 等 暗褐色을 띄는 試料를 酸性, 中性, 알칼리性 溶液으로 하여 오존으로 50分間 處理하였을 때 투과율이 30~40%에서 70% 以上 상승하였으며 알칼리성溶液에서가 투과율이 더 좋았다. For the increase of treatment efficiency when waste water which included residual chlorine was biologically treated with activated sludge, elimination efficiency of COD and change of transmittance by activated sludge method were examined after the treatment of residual chlorine with ozone of 100 volt and 10ℓ-air/min., 0.8g-O₃/hr. in velocity. The results were as follows. 1) When the sample which included residual chlorine was treated with ozone of 0.8g-O₃/hr., COD removal rate by the change of pH and ozonization time was more efficient than acidity solution of pH 3.2(20%) and alkaline solution of pH 12.0(10%). Since the COD removal rate become 35% after 50 minutes of ozonization time at neutral solution of pH 7.2. 2) After the residual chlorine included sample was treated with ozone for 50 minutes to make the F/M rate of 0.1㎏·BOD/㎏·MLSS, the sample was reacted with activated sludge for 3, 6, 9 and 15 hours. The 6 hours reaction gave 85.8% COD elimination which showed higher efficiency than 60.5% by Na₂S₂O₂ treatment togather with shorting efficiency of reaction time. 3) When such sample as polyphenol which gives dark brown color was treated with ozone for 50 minutes after being made into acidic, neutral and alkaline solution, the transmittance increased from 30-40% to 70% and gave better result in alkaline solution.

제주지역에서 무선호출 주파수의 전파 특성

이청민,장재성,강부식 濟州大學校工科大學附屬産業技術硏究所 1993 尖端技術硏究所論文集 Vol.4 No.-

Received field strength in land mobile-radio communication must consider characteristics besides. This paper analysis propagation chracterics to multipath and then compute theoretical value of field strength which vectorial compose and compared with measured value. Seorm base station is to unity south area of Cheju by single propagation area. But Gunwelak base station cause a weak and impossible received field area to increase obstruction. In order to replace it, we necessary technique study of new mobile communication propagation environment. For take a accuracy received field strength, the first thing field strength of a point must measure many times and calculate mean value of it. The second we must suggest and analytical modeling to compose to these measured values.

바다방석고둥의 凍結貯藏中의 品質의 變化

金昌龍,宋大鎭 濟州大學校工科大學附屬産業技術硏究所 1992 尖端技術硏究所論文集 Vol.3 No.-

바다방석고둥(일명 보말)을 冷凍食品化하기 위하여 脫殼, 處理 後 -18˚C, -25˚C, -30˚C로 3개월 동안 凍結 貯藏하면서 品質變化에 대하여 物理, 化學, 組織的인 면에서 實驗한 結果를 要約하면 다음과 같다. 1. 凍結貯藏中에 pH는 漸次 增加하였는데 -30˚C 凍結貯藏은 變化가 가장 적었으며, 遊離 드립量은 凍結貯藏溫度가 높고 貯藏期間이 길어질수록 서서히 增加하였다. 2. 保水性은 生試料의 경우 70%를 나타내었으며 -30˚C 凍結 貯藏은 凍結直後에 비해서 거의 變化가 없었다. 그러나 凍結貯藏 3개월째에 -18˚C에서는 49%, -25˚C에서는 57%로서 貯藏期間이 길어질수록 保水性의 低下가 顯著하였다. 3. 筋肉蛋白質의 溶出性에서 筋形質蛋白質과 基質蛋白質은 凍結溫度와 貯藏期間에 상관없이 비교적 變化가 적었다. 筋原纖維蛋白質의 溶出性은 凍結貯藏溫度가 높고 貯藏期間이 長期化할수록 減少하였다. 4. 筋肉組織構造의 變化로서는 1)生試料의 筋肉組織構造에서 筋纖維는 平滑筋(2-3μ)으로 橫斷面과 縱斷面의 구분이 없이 가느다란 筋纖維가 筋肉組織內에서 緻密하게 分布하였는데, 中間部로 갈수록 약간 굵은 筋纖維가 zig zag로 얼켜져 있는 狀態였다. 2)凍結直後의 筋肉組織은 凍結溫度와는 상관없이 筋細胞外로 작은 氷結晶이 多數 形成되었으며, 解凍後에는 양호한 復元狀態를 보였다. 3)解凍狀態의 組織構造는 凍結貯藏溫度가 높을수록 永結晶의 成長을 볼 수 있었으며, -18˚C에서 3개월 貯藏한 組織에서는 筋細胞의 凝集에 의해 空間 面積이 넓어지는 경향이었다. 4)解凍後의 筋肉組織構造는 1개월 貯藏까지는 貯藏溫度에 관계없이 잘 復元되었으며, 3개월 貯藏에서 -30˚C 貯藏의 경우는 잘 復元되었으나 -18˚C에서는 復元不充分에 의한 空間 面積이 많이 남아 있었다. To investigate the physical, chemical, and histological quality changes during frozen storage, top shell, Omphalius pfeifferi capenteri, was stored under the temperatures of -18˚C, -25˚C, and -30˚C immediately after shelling. The results are as follows : 1. Generally, pH was increased but a little change was observed at -30˚C. Amount of free drip was increased with higher frozen temperaure and longer frozen storage period. 2. Water holding capacity was almost as constant at -30˚C as immediately after frozen, but the longer storage period goes the lower water holding capacity was observed. 3. The extractibility and composition of muscle protein, sarcoplasmic protein and stroma protein were rather stable regardless of frozen temperature and frozen storage period however, extractibility of myofibrillar protein was decreased with higher frozen temperature and longer frozen storage period. 4. On the changes of muscle tissue structure, following points were observed. 1) In the muscle tissue structure of fresh sample, fine muscle fiber was closely distributed all over the tissue regardless of cross and longitudinal section. 2) Immediately after freezing, tiny ice crystals were in extracellular part regardless of different frozen temperature, these were restored satisfactorily into the muscle tissue after thawed. 3) In tissue structure under frozen state, it was observed that ice crystals apparently grew with the higher storage temperature. Empty spaces between muscle bundles which were formed by aggregation of muscle fiber were observed after 3 months storage at -18˚C. 4) Tissue structure in thawed state was restored satisfactorily after 1 month storage regardless of storage temperature. After 3 months storage at -30˚C, muscle tissue was well restored, but at -18˚C, empty spaces were appeared due to imperfected restoration.

반파정류브러쉬없는 동기전동기의 디지탈 위치제어

吳性寶 濟州大學校工科大學附屬産業技術硏究所 1994 尖端技術硏究所論文集 Vol.5 No.-

Digital position control of brushless synchronous motor is one of the attractive and interesting research subjects in the field of industrial robot. A direct - drive(D-D) system is being developed to control the forces and position of high speed and high accuracy. The P I control without any additional control process has very poor response and inaccuracy of position. In order to improve those problem, We proposed a repetitive control with the position memory type, learning process. It provides us the good response of system and the improval of position accuracy.

格子型 웨이브 디지탈 필터의 性能에 關한 硏究

고주홍,이용학 濟州大學校工科大學附屬産業技術硏究所 1994 尖端技術硏究所論文集 Vol.5 No.-

For one input signal the wave digital filter(WDF) has a merit producing two output signals which are complemented each other. The WDF are described in terms of analog LC filters and WDF are classified into lattice WDF(LWDF) and Ladder WDF(LWDF) depending on their references. In this paper, firstly theory of the LWDF is summarized and the design procedures are outlined. By introducing examples LWDF is compared with the conventional IIR filters and WDLF. Also the coefficients quantization error are discussed quantitatively. It is verified that the characteristics both transition and stopband attenuation of LWDF are more improved than that of two types.