• ETRI Journal
• /
• 제25권4호
• /
• pp.238-246
• /
• 2003

This paper introduces the modeling and analysis of a discrete-time, two-phase queueing system for both exhaustive batch service and gated batch service. Packets arrive at the system according to a Bernoulli process and receive batch service in the first phase and individual services in the second phase. We derive the probability generating function (PGF) of the system size and show that it is decomposed into two PGFs, one of which is the PGF of the system size in the standard discrete-time Geo/G/1 queue without vacations. We also present the PGF of the sojourn time. Based on these PGFs, we present useful performance measures, such as the mean number of packets in the system and the mean sojourn time of a packet.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1995년도 춘계공동학술대회논문집; 전남대학교; 28-29 Apr. 1995
• /
• pp.729-738
• /
• 1995

We consider a priority-based packet-switching system with three phases of the packet transmission time. Each packet belongs to one of several priority classes, and the packets of each class arrive at a switch in a Poison process. The switch transmits queued packets on a priority basis with three phases of preemption mechanism. Namely, the transmission time of each packet consists of a preemptive-repeat part for the header, a preemptive-resume part for the information field, and a nonpreemptive part for the trailer. By an exact analysis of the associated queueing model, we obtain the Laplace-Stieltjes transform of the distribution function for the delay, i.e., the time from arrival to transmission completion, of a packet for each class. We derive a set of equations that calculates the mean response time for each class recursively. Based on this result, we plot the numerical values of the mean response times for several parameter settings. The probability generating function and the mean for the number of packets of each class present in the system at an arbitrary time are also given.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2012년도 제42회 동계 정기 학술대회 초록집
• /
• pp.239-240
• /
• 2012

With advancements in semiconductor device technologies, manufacturing processes are getting more complex and it became more difficult to maintain tighter process control. As the number of processing step increased for fabricating complex chip structure, potential fault inducing factors are prevail and their allowable margins are continuously reduced. Therefore, one of the key to success in semiconductor manufacturing is highly accurate and fast fault detection and classification at each stage to reduce any undesired variation and identify the cause of the fault. Sensors in the equipment are used to monitor the state of the process. The idea is that whenever there is a fault in the process, it appears as some variation in the output from any of the sensors monitoring the process. These sensors may refer to information about pressure, RF power or gas flow and etc. in the equipment. By relating the data from these sensors to the process condition, any abnormality in the process can be identified, but it still holds some degree of certainty. Our hypothesis in this research is to capture the features of equipment condition data from healthy process library. We can use the health data as a reference for upcoming processes and this is made possible by mathematically modeling of the acquired data. In this work we demonstrate the use of recurrent neural network (RNN) has been used. RNN is a dynamic neural network that makes the output as a function of previous inputs. In our case we have etch equipment tool set data, consisting of 22 parameters and 9 runs. This data was first synchronized using the Dynamic Time Warping (DTW) algorithm. The synchronized data from the sensors in the form of time series is then provided to RNN which trains and restructures itself according to the input and then predicts a value, one step ahead in time, which depends on the past values of data. Eight runs of process data were used to train the network, while in order to check the performance of the network, one run was used as a test input. Next, a mean squared error based probability generating function was used to assign probability of fault in each parameter by comparing the predicted and actual values of the data. In the future we will make use of the Bayesian Networks to classify the detected faults. Bayesian Networks use directed acyclic graphs that relate different parameters through their conditional dependencies in order to find inference among them. The relationships between parameters from the data will be used to generate the structure of Bayesian Network and then posterior probability of different faults will be calculated using inference algorithms.

• Industrial Engineering and Management Systems
• /
• 제10권2호
• /
• pp.134-139
• /
• 2011

The problem of scheduling in permutation flowshops has been extensively investigated by many researchers. Recently, attempts are being made to consider more than one objective simultaneously and develop algorithms to obtain a set of Pareto-optimal solutions. Varadharajan et al. (2005) presented a multi-objective simulated-annealing algorithm (MOSA) for the problem of permutation-flowshop scheduling with the objectives of minimizing the makespan and the total flowtime of jobs. The MOSA uses two initial sequences obtained using heuristics, and seeks to obtain non-dominated solutions through the implementation of a probability function, which probabilistically selects the objective of minimizing either the makespan or the total flowtime of jobs. In this paper, the same problem of heuristically developing non-dominated sequences is considered. We propose an effective heuristics based on simulated annealing (SA), in which the weighted sum of the makespan and the total flowtime is used. The essences of the heuristics are in selecting the initial sequence, setting the weight and generating a solution in the search process. Using a benchmark problem provided by Taillard (1993), which was used in the MOSA, these conditions are extracted in a large-scale experiment. The non-dominated sets obtained from the existing algorithms and the proposed heuristics are compared. It was found that the proposed heuristics drastically improved the performance of finding the non-dominated frontier.

• 한국콘텐츠학회논문지
• /
• 제15권1호
• /
• pp.475-481
• /
• 2015

종속변수와 설명변수 사이의 관계가 선형이 아닌 경우에는 비선형 관계를 반영할 수 있는 다항회귀분석을 이용하여 회귀분석을 수행한다. 한편, 다항회귀분석에는 설명변수의 거듭제곱항들이 설명변수에 추가되므로 설명변수들 사이에 상관관계가 발생하여 다항회귀모형의 성능 저하 문제가 발생할 수 있다. 본 논문에서는 PGF 수치역변환 문제를 사례로 하여 주성분회귀분석을 통해 다항회귀분석의 성능을 극적으로 향상시킬 수 있음을 보인다. 본 논문에서는 PGF의 정의를 이용하여 PGF를 다항회귀분석으로 모형화한다. 다항회귀분석을 이용하여 PGF 전개식의 회귀계수를 추정하면 회귀계수의 추정 자체가 불가능하거나 계수 추정의 정확성이 저하되는 문제가 발생한다. 이 경우 다항회귀분석에 주성분회귀분석을 적용하면 계수 추정의 정확도가 극적으로 향상되어 다항회귀분석의 계수 추정 시 발생하는 문제를 해결할 수 있음을 밝힌다.

• 한국통신학회논문지
• /
• 제18권3호
• /
• pp.301-309
• /
• 1993

본 논문은 m-분포 페이딩 채널하에서 trellis 부호화된 8-PSK 시스팀의 성능분석에 관한 연구이다. 이동 통신 채널에서 페이딩에 의한 시스팀의 성능 저하를 보상하기 위하여 대역폭의 확장없이 부호 이득을 얻을 수 있는 trellis 부호를 이용하였다. 성능 분석은 4-상태와 8-상태를 갖는 trellis 부호화된 8-PSK 시스팀에 대하여 Chernoff bound와 생성함수 기법을 이용하여 비트 오류 확률의 상한식을 구하였다. 그 결과 4-상태 trellis 부호보다는 8-상태 trellis 부호가 오류 정정 능력이 우수함을 알 수 있었으며, 페이딩이 심할 수록 부호화에 의한 성능 개선이 효과적임을 보였다.

• 산업경영시스템학회지
• /
• 제20권43호
• /
• pp.153-162
• /
• 1997

M/G/1 queueing system is one of the most widely used one to model the real system. When operating a real systems, since it often takes cost, some control policies that change the operation scheme are adopted. In particular, the N-policy is the most popular among many control policies. Almost all researches on queueing system are based on the assumption that the arrival rates of customers into the queueing system is constant, In this paper, we consider the M/G/1 queueing system whose arrival rate varies according to the servers status : idle, set-up and busy states. For this study, we construct the steady state equations of queue lengths by means of the supplementary variable method, and derive the PGF(probability generating function) of them. The L-S-T(Laplace Stieltjes transform) of waiting time and average waiting time are also presented. We also develop an algorithm to find the optimal N-value from which the server starts his set-up. An analysis on the performance measures to minimize total operation cost of queueing system is included. We finally investigate the behavior of system operation cost as the optimal N and arrival rate change by a numerical study.

• 한국산학기술학회논문지
• /
• 제16권4호
• /
• pp.2757-2763
• /
• 2015

p-Snake는 기존의 동적윤곽모델(Active Contour Model)에 원형에너지를 추가로 적용한 에너지 최소화 알고리즘으로 에지 정보가 명확하지 않은 영역에서의 윤곽선 추출을 위해 사용된 방법이다. 본 논문에서는 원과 직선 프리미티브(primitive)의 조합으로 표현되는 가변 원형(prototype)과 퍼지 함수를 적용한 원형에너지장의 생성 기법을 제안하여 p-Snake의 윤곽선 추출 성능을 개선하였다. 제안 방법은 입력된 부품 코드를 기반으로 원형을 정의하고 전처리 과정을 통해 구해진 각 프리미티브 구간에서 대략적인 초기 윤곽을 검출한 후, 프리미티브들이 가변적으로 적응하여 원형을 생성하고 여기에 원형과의 거리에 따른 윤곽 확률을 퍼지 함수를 통해 계산하여 원형에너지 장을 생성하였다. 이를 p-Snake에 적용하여 다양한 소형부품들을 대상으로 준비한 200장의 영상에서 윤곽선을 검출하고, 원형과의 유사도를 비교한 결과 적응 원형을 사용한 p-Snake가 기존의 Snake에 비해 약 4.6% 가량 우수함을 보였다.

• 한국진공학회:학술대회논문집
• /
• 한국진공학회 2012년도 제42회 동계 정기 학술대회 초록집
• /
• pp.241-241
• /
• 2012

Semiconductor industry has been taking the advantage of improvements in process technology in order to maintain reduced device geometries and stringent performance specifications. This results in semiconductor manufacturing processes became hundreds in sequence, it is continuously expected to be increased. This may in turn reduce the yield. With a large amount of investment at stake, this motivates tighter process control and fault diagnosis. The continuous improvement in semiconductor industry demands advancements in process control and monitoring to the same degree. Any fault in the process must be detected and classified with a high degree of precision, and it is desired to be diagnosed if possible. The detected abnormality in the system is then classified to locate the source of the variation. The performance of a fault detection system is directly reflected in the yield. Therefore a highly capable fault detection system is always desirable. In this research, time series modeling of the data from an etch equipment has been investigated for the ultimate purpose of fault diagnosis. The tool data consisted of number of different parameters each being recorded at fixed time points. As the data had been collected for a number of runs, it was not synchronized due to variable delays and offsets in data acquisition system and networks. The data was then synchronized using a variant of Dynamic Time Warping (DTW) algorithm. The AutoRegressive Integrated Moving Average (ARIMA) model was then applied on the synchronized data. The ARIMA model combines both the Autoregressive model and the Moving Average model to relate the present value of the time series to its past values. As the new values of parameters are received from the equipment, the model uses them and the previous ones to provide predictions of one step ahead for each parameter. The statistical comparison of these predictions with the actual values, gives us the each parameter's probability of fault, at each time point and (once a run gets finished) for each run. This work will be extended by applying a suitable probability generating function and combining the probabilities of different parameters using Dempster-Shafer Theory (DST). DST provides a way to combine evidence that is available from different sources and gives a joint degree of belief in a hypothesis. This will give us a combined belief of fault in the process with a high precision.

• Journal of Ecology and Environment
• /
• 제35권4호
• /
• pp.359-365
• /
• 2012

We investigated leaf disease intensity of Kalopanax septemlobus (prickly castor oil tree) caused by the parasitic fungus Mycosphaerella acanthopanacis, in thirty natural host populations in the Japanese Archipelago. The disease intensity observed for individual trees were analyzed using a generalized additive model as a function of tree size, tree density, climatic terms and spatial trend surface. Individual tree size and conspecific tree density were shown to have significant negative and positive effects on disease intensity, respectively. The findings suggest that the probability of disease infection is partly determined by dispersal of infection agents (ascospores) from the fallen leaves on the ground, which can be enhanced by aggregation of host trees in a forest stand. Regional-scale spatial bias was also present in disease intensity; the populations in northern Japan and southern Kyushu were more severely infected by the fungus than those in southwestern Honshu and Shikoku. Regional variation of disease intensity was explained by both climatic factors and a trend surface term, with a latitudinal cline detected, which increases towards the north. Further research should be conducted in order to understand all of the factors generating the latitudinal cline detected in this study.