• 대한화학회지
• /
• 제60권4호
• /
• pp.225-234
• /
• 2016

The activity of 34 sulfonamide derivatives has been estimated by means of multiple linear regression (MLR), artificial neural network (ANN), simulated annealing (SA) and genetic algorithm (GA) techniques. These models were also utilized to select the most efficient subsets of descriptors in a cross-validation procedure for non-linear -log (IC50) prediction. The results obtained using GA-ANN were compared with MLR-MLR, MLR-ANN, SA-ANN and GA-ANN approaches. A high predictive ability was observed for the MLR-MLR, MLR-ANN, SA-ANN and MLR-GA models, with root mean sum square errors (RMSE) of 0.3958, 0.1006, 0.0359, 0.0326 and 0.0282 in gas phase and 0.2871, 0.0475, 0.0268, 0.0376 and 0.0097 in solvent, respectively (N=34). The results obtained using the GA-ANN method indicated that the activity of derivatives of sulfonamides depends on different parameters including DP03, BID, AAC, RDF035v, JGI9, TIE, R7e+, BELM6 descriptors in gas phase and Mor 32u, ESpm03d, RDF070v, ATS8m, MATS2e and R4p, L1u and R3m in solvent. In conclusion, the comparison of the quality of the ANN with different MLR models showed that ANN has a better predictive ability.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2021년도 춘계학술대회
• /
• pp.538-540
• /
• 2021

본 논문에서는 드론의 표적탐지 임무 수행 시 상대운동 정보 제공을 위하여 지정된 표적을 탐지하고 인식하는 스마트 표적탐지 시스템을 제안하였다. 제안된 시스템은 적절한 정확도(i.e. mAP, IoU) 및 높은 실시간성을 동시에 확보할 수 있는 알고리즘을 개발하는데 중점을 두었다. 제안된 시스템은 Google Inception V2 딥러닝 모델의 100k 학습 후 test 결과가 1.0에 가까운 정확성을 보였고 실시간성도 Nvidia GTX 2070 Max-Q를 기반으로 한 고성능 노트북 활용 시에 추론 속도가 약 60-80[Hz]를 기록하였다. 제안된 스마트 표적탐지 시스템은 드론과 같이 운용되어 컴퓨터 영상처리를 활용하여 표적을 자동으로 인식하고 표적을 따라가면서 감시정찰 임무를 성공적으로 수행하는데 도움이 될 것이다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2004년도 ICCAS
• /
• pp.869-874
• /
• 2004

The inverse kinematics of five-axis milling machines produce large errors near stationary points of the required surface. When the tool travels cross or around the point the rotation angles may jump considerably leading to unexpected deviations from the prescribed trajectories. We propose three new algorithms to repair the trajectories by adjusting the rotation angles in such a way that the kinematics error is minimized. Given the tool orientations and the inverse kinematics of the machine, we first eliminate the jumping angles exceeding ${\pi}$ by using the angle adjustment algorithm, leaving the jumps less than ${\pi}$ to be further optimized. Next, we propose to apply an angle switching algorithm to compute the rotations and identify an optimized sequence of rotations by the shortest path scheme. Further error reduction is accomplished by the angle insertion algorithm based an o special interpolation to obtain the required rotations near the singularity. We have verified the algorithms by five-axis milling machines, namely, MAHO600E at the CIM Lab of Asian Institute of Technology and HERMLE UWF902H at the CIM Lab of Kasetsart University.

• 한국해양학회지:바다
• /
• 제19권3호
• /
• pp.169-179
• /
• 2014

황해 남동부 해역에 설치한 해양부이(YSROB)에서 약 27개월간 관측된 장파, 단파 복사량을 포함한 대기, 해양 변수와 COARE 3.0 알고리즘을 이용하여 월평균 해양-대기간 열속을 산출하고 기존 연구결과와 비교하였다. YSROB 위치에서 열속은 순 단파복사(Qi)에 의해 해양은 대기로부터 열을 얻고 순 장파복사($Q_b$), 현열($Q_h$), 잠열($Q_e$)에 의해서 열손실이 일어난다. 전체 열손실 중 $Q_e$에 의한 손실이 51%로 가장 크게 나타났으며 $Q_b$와 $Q_h$에 의한 손실은 각각 34%, 15% 이다. 순열속($Q_n$)은 $Q_i$가 최대인 5월에 최대($191.4W/m^2$)이며 모든 열속 성분이 최소인 12월에 최소($-264.9W/m^2$)이다. 연평균 $Q_n$은 $1.9W/m^2$ 이지만 관측기기의 정확도에 의한 오차산정 결과(최대 ${\pm}19.7W/m^2$)를 고려하면 무시할 정도로 작다. YSROB과 동일한 위치에서의 기존 월별 열속 산출 결과는 YSROB에서 실측값에 기반한 열속에 비해 여름철 $Q_i$가 약 $10{\sim}40W/m^2$ 과소 평가된 반면에 겨울철에는 $Q_e$와 $Q_h$에 의한 열 손실이 각각 약 $50W/m^2$, $30{\sim}70W/m^2$ 과다하게 산출되었다. 이로 인하여 해양이 열을 얻는 4월~8월에는 기존 연구에서의 열 획득량이 본 연구 결과보다 적게 나타나며, 해양이 열을 잃는 겨울철에는 기존 연구에서의 해양으로부터의 열 손실이 본 연구 결과에 비해 크게 나타난다. 특히, 12월과 1월의 $Q_n$ 차이는 약 $70{\sim}130W/m^2$에 달한다. 장기적인 재분석장(MERRA) 분석 결과에 의하면 이와 같은 월평균 열속의 차이는 연변동 등 시간 변동에 의한 것이 아니라 열속 산출 시 사용된 자료의 부정확성에 기인하는 것으로 판단된다. 본 연구 결과로부터 기존의 기후적인 열속을 연구에 활용하거나 수치모델에 사용함에 있어 주의가 요망된다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1990년도 하계학술대회 논문집
• /
• pp.154-157
• /
• 1990

This paper describes a new method of calculating expected energy generation and loss of load probability (L.O.L.P) for electric power system operation and expansion planning. The method represents an equivalent load duration curve (E.L.D.C) as a mixture of cumulants approximation (M.O.C.A), which is the general case of mixture of normals approximation (M.O.N.A). By regarding a load distribution as many normal distributions-rather than one normal distribution-and representing each of them in terms of Gram-Charller expansion, we could improve the accuracy of results. We developed an algorithm which automatically determines the number of distribution and demarcation points. In modelling of a supply system, we made subsets of generators according to the number of generator outage: since the calculation of each subset's moment needs to be processed rapidly, we futher developed specific recursive formulae. The method is applied to the test systems and the results are compared with those of cumulant, M.O.N.A and Booth-Baleriaux method. It is verified that the M.O.C.A method is faster and more accurate than any other methods.

• 한국정보통신학회논문지
• /
• 제18권4호
• /
• pp.876-884
• /
• 2014

본 논문에서는 개선된 정규화 최소합(improved normalized min-sum) 복호 알고리듬을 적용한 LDPC 복호기를 설계하였다. 설계된 LDPC 복호기는 IEEE 802.16e 모바일 WiMAX 표준의 19가지 블록길이(576~2304)에 따른 6가지 부호율(1/2, 2/3A, 2/3B, 3/4A, 3/4B, 5/6)과 IEEE 802.11n 무선 랜 표준의 3가지 블록길이(648, 1296, 1944)에 따른 4가지 부호율(1/2, 2/3, 3/4, 5/6)을 지원한다. INMS 복호 알고리듬과 SM(sign-magnitude) 수체계 연산을 기반으로 하는 DFU(decoding function unit)을 구현하여 하드웨어 복잡도와 복호 성능을 최적화시켰다. 설계된 LDPC 복호기는 0.18-${\mu}m$ CMOS 셀 라이브러리를 이용하여 100 MHz 동작 주파수로 합성한 결과, 284,409 게이트와 62,976 비트의 메모리로 구현되었으며, FPGA 구현을 통해 하드웨어 동작을 검증하였다. 1.8V 전원전압에서 100 MHz로 동작 가능할 것으로 평가되며, 부호율과 블록길이에 따라 약 82~218 Mbps의 성능을 가질 것으로 예상된다.

• The Journal of the Acoustical Society of Korea
• /
• 제20권1E호
• /
• pp.14-19
• /
• 2001

We have proposed BIST method and circuit for embedded 16M SDRAM with logic. It can test the AC parameter of embedded 16M SDRAM using the BIST circuit capable of detecting the address of a fail cell installed in an Merged Memory with Logic(MML). It generates the information of repair for redundancy circuit. The function and AC parameter of the embedded memory can also be tested using the proposed BIST method. It is possible to test the embedded SDRAM without external test pin. The total gate of the BIST circuit is approximately 4,500 in the case of synthesizing by 0.25μm cell library and is verified by Verilog simulation. The test time of each one AC parameter is about 200ms using 2Y-March 14n algorithm.

• The Journal of the Acoustical Society of Korea
• /
• 제15권2E호
• /
• pp.24-32
• /
• 1996

A rigorous theory for modeling, analysis, optimum nonseparable filter bank in multidimensional M-band quantized subband codec are developed in this paper. Each pdf-optimized quantizer is modeled by a nonlinear gain-plus-additive uncorrelated noise and embedded into the subband structure. We then decompose the analysis/synthesis filter banks into their polyphase components and shift the down-and up-samplers to the right and left of the analysis/synthesis polyphase matrices respectively. Focusing on the slow clock rate signal between the samplers, we derive the exact expression for the output mean square quantization error by using spatial-invariant analysis. We show that this error can be represented by two uncorrelated components : a distortion component due to the quantizer gain, and a random noise component due to fictitious uncorrelated noise at the uantizer. This mean square error is then minimized subject to perfect reconstruction (PR) constraints and the total bit allocation for the entire filter bank. The algorithm gives filter coefficients and subband bit allocations. Numerical design example for the optimum nonseparable orthonormal filter bank is given with a quincunx subsampling lattice.

• 전자통신동향분석
• /
• 제33권3호
• /
• pp.89-99
• /
• 2018

Human activity recognition (HAR) is a technology that aims to offer an automatic recognition of what a person is doing with respect to their body motion and gestures. HAR is essential in many applications such as human-computer interaction, health care, rehabilitation engineering, video surveillance, and artificial intelligence. Smartphones are becoming the most popular platform for activity recognition owing to their convenience, portability, and ease of use. The noticeable change in smartphone-based activity recognition is the adoption of a deep learning algorithm leading to successful learning outcomes. In this article, we analyze the technology trend of activity recognition using smartphone sensors, challenging issues for future development, and a strategy change in terms of the generation of a activity recognition dataset.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1985년도 춘계학술대회
• /
• pp.15-19
• /
• 1985

This study deals with a kind of signal processing, Kalman filtering techniques, which are applied to ultrasound signal to improve resolution capability. The main advantage of Kalman filter algorithm for the analysis of reflected ultrasound signal is its recursive structure which can be easily adapted to time-varing system. In state-space description of the system, the 6th order system produces the best spectral approximation to the source pulse. As a result of spectrum analysis, 6th order estimator for two closely spaced (0.5mm) reflectors enhance resolution by 4dB - 10dB. By using this result, the possibility to detect even minute tumor is found.