• 전자공학회논문지B
• /
• 제31B권7호
• /
• pp.211-218
• /
• 1994

Stochastic computation employs random pulse streams to represent numbers. In this paper, we study a new method to implement the number system which uses the ratio of the numbers of ones and zeros in the pulse streams. In this number system. if P is the probability that a pulse is one in a pulse stream then the number X represented by the pulse stream is defined as P/(1-P). We propose circuits to implement the basic operations such as addition multiplication and sigmoid function with this number system and examine the error characteristics of such operations in stochastic computation. We also propose a neuron model and derive a learning algorithm based on backpropagation for the 3-layered feedforward neural networks. We apply this learning algorithm to a digit recognition problem. To analyze the results, we discuss the errors due to the variance of the random pulse streams and the quantization noise of finite length register.

• 전자공학회논문지 IE
• /
• 제47권2호
• /
• pp.54-59
• /
• 2010

본 논문은 무선 데이터 네트워크에 적용 가능한 적응 프레임 길이 제어 방식을 가진 데이터 링크 프로토콜을 제안한다. 무선 데이터망은 수시로 변화는 에러 환경에 놓여 있다. 이런 에러 환경에서도 잘 적용할 수 있는 가변 프레임 제어 방식을 제안하고 그 성능을 분석하였다. 2상태 마코프 모델을 이용하여 채널 환경을 모델링하였다. 수치해석방식을 통하여 새롭게 제안한 방식이 다양한 종류의 에러환경에서도 좋은 성능을 발휘할 수 있음을 보여 주었다.

• Journal of the Korean Statistical Society
• /
• 제35권2호
• /
• pp.115-142
• /
• 2006

This paper studies the problem of option hedging when the underlying asset price process is a compound Poisson process. By adopting an asymptotic approach to let the security price converge to a continuous process, we find a closed-form hedging strategy that improves the classical Black-Scholes hedging strategy in a quadratic sense. We first show that the scaled Black-scholes hedging error has a limit in law, and that limit is decomposed into a part that can be traded away and a part that is purely unreplicable. The Black-Scholes hedging strategy is then modified by adding the replicable part of its hedging error and by adding the mean-variance hedging strategy to the nonreplicable part. Some results of simulation experiment s are also provided.

• 한국근적외분광분석학회:학술대회논문집
• /
• 한국근적외분광분석학회 2001년도 NIR-2001
• /
• pp.1521-1521
• /
• 2001

The objective of the research being presented was to construct and calibrate a spectrometer for the analysis of single kernels of corn. In light of the difficulties associated with capturing the spatial variability in composition of corn kernels by single-beam spectrometry, a hyperspectral imaging spectrometer was constructed with the intention that it would be used to analyze single kernels of corn for the prediction of moisture and oil content. The spectrometer operated in the range of 750- 1090 nanometers. After evaluating four methods of standardizing the output from the spectrometer, calibrations were made to predict whole-kernel moisture and oil content from the hyperspectral image data. A genetic algorithm was employed to reduce the number of wavelengths imaged and to optimize the calibrations. The final standard errors of prediction during cross-validation (SEPCV) were 1.22% and 1.25% for moisture and oil content, respectively. It was determined, by analysis of variance, that the accuracy and precision of single-kernel corn analysis by hyperspectral imaging is superior to the single kernel reference chemistry method (as tested).

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1995년도 추계학술대회 논문집
• /
• pp.907-910
• /
• 1995

Though SLA(Stereolithography Apparatus) is being recognized as an innovative technology, it still can not be used to fully practical applications since it lacks of dimensional accuracy compared to conventional processes. In SLA, the accuracy of cured part depends upon the set of process parmeters. In order to improve the accuracy of SLA, this paper quantitatively evaluates how largely each process parameter of SLA contributes to the part accuracy and estimates the optimal set of process parameter which minimizes the dimensional errors of the test part, "letter-H" part. For this purpose, we use ANOVA(analysis of variance) and S/N(signal-to-noise)ratio of Taguchi method.hi method.

• 한국자동차공학회논문집
• /
• 제4권2호
• /
• pp.115-126
• /
• 1996

In order to obtain the principal design data for developing the Autonomous Robot Vehicle(ARV), Sensitivity analysis on the trajectory error and friction force with respect to the dynamic parameters is performed. In the straight motion, the trajectory error has been proved to be much affected by the mass variance of the ARV while the lateral friction force is much affected by the location of the mass center. In the curved motion, the effect of mass and moment of inertia is considered importantly. In addition, the lateral offset gives more effect than the geometric dimension of the ARV on the trajectory errors and friction force.

• 대한산업공학회지
• /
• 제22권2호
• /
• pp.231-245
• /
• 1996

Semiconductor manufacturing line includes several batch processes which are to be controlled effectively to enhance the productivity of the line. The key problem in batch processes is a dynamic batch sizing problem which determines number of lots processed simultaneously in a single botch. The batch sizing problem in semiconductor manufacturing has to consider delay of lots, setup cost of the process, machine utilization and so on. However, an optimal solution cannot be attainable due to dynamic arrival pattern of lots, and difficulties in forecasting future arrival times of lots of the process. This paper proposes an efficient batch sizing heuristic, which considers delay cost, setup cost, and effect of the forecast errors in determining the botch size dynamically. Extensive numerical experiments through simulation are carried out to investigate the effectiveness of the proposed heuristic in four key performance criteria: average delay, variance of delay, overage lot size and total cost. The results show that the proposed heuristic works effectively and efficiently.

• Architectural research
• /
• 제16권2호
• /
• pp.75-80
• /
• 2014

Risks, uncertainties, and associated cost overruns are critical problems for construction projects. Cost contingency is an important funding source for these unforeseen events and is included in the base estimate to help perform financially successful projects. In order to predict more accurate contingency, many empirical models using regression analysis and artificial neural network method have been proposed and showed its viability to minimize prediction errors. However, categorical factors on contingency cannot have been treated and thus considered in these empirical models since those models are able to treat only numerical factors. This paper identified potential factors on contingency in transportation construction projects and evaluated categorical factors using the one-way ANOVA statistical method. Among factors including project work type, delivery method type, contract agreement type, bid award type, letting type, and geographical location, two factors of project work type and contract agreement type were found to be statistically important on allocating cost contingency.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.527-530
• /
• 1996

The Kalman filter has been used as a self-localization method for the mobile robot. To satisfy the assumptions inherent in the Kalman filter, we should calibrate the sensors of the robot before use of them. However, it is generally hard to find exact sensor parameters, and the parameters may change during the robot task as the environment varies. Thus we need to perform on-line sensor calibration, by which we can obtain more credible location of the mobile robot. In this paper, we present an on-line sensor calibration scheme which estimates the unknown sensor bias and the current position of the robot. To this end, first we find out the calibration errors of the sensor from redundant sensory data using the parity vector and recursive minimum variance estimation. Then we calculate the current position of the robot by weighted least square estimation without internal encoder data. The performance of the proposed method is evaluated through computer simulation.

• 전기의세계
• /
• 제22권4호
• /
• pp.17-24
• /
• 1973

In the view point of practical engineering the identification problem may be considered as a problem to determine the optimal model in the sense of minimizing a given criterion function using the input-output records of the plant. In the system identification the statistical approach has been known to be very effective when the topological structure of the system and the statistical characteristics of the observation noises are known a priori. But in the practical situation there are many cases when the inforhation about the observation noises or the system noises are not available a priori. Here, the authors propose a new identification method which can be used effectively even in the cases when the variances of observation noises are unknown a priori. In the method, the identification of unknown parameters of a linear diserete system is achieved by minimizing the improved quadratic criterion function which is composed of the term of square equation errors and the term to eliminate the affection of observation noises. The method also gives the estimate of noise variance. Numerical computations for several examples show that the proposed procedure gives satisfactory results even when the short time observation data are provided.