신호인자와 특성치가 연속형 변수인 경우를 고려한 동적파라미터 설계절차에 관한 연구

배홍석,이만웅,송서일 한국산업경영시스템학회 1996 한국산업경영시스템학회지 Vol.19 No.39

都市化와 公害 : 淸州市를 中心으로

裵弘碩 청주대학교 행정학회 1979 行政論叢 Vol.3 No.-

정적 파라미터 설계에 있어서 성능척도의 대안들에 대한 비교.분석

배홍석,이만웅,송서일 한국산업경영시스템학회 1995 한국산업경영시스템학회지 Vol.18 No.36

In parameter design, Taguchi's stated objective is to find the setting of product or process design parameters that minimize average quadratic loss-that is, the average squared deviation of the response from its target value. Yet in practice to choose the settings of design parameters he maximizeds a set of measures called signal-to-noise(SN) ratios. In general, Taguchi gave no justification for the use the SN ratios and no explanation of why the two-step procedure that he recommened will minimize average loss. The purpose of this study is comparing and analyzing of performance statistics by Leon et al(PerMIA), Box(Transformation theory), Vining & Myers(Dual Response Systems) and Taguchi(Signal-to-Noise ratios).

망목특성에 있어서 최적변수설계 절차의 비교·분석에 관한 연구

배홍석,이만웅,송서일 동아대학교 공과대학 부설 한국자원개발연구소 1995 硏究報告 Vol.19 No.2

The purpose of this study is to verify the usefulness and effectiveness of Taguchi method, by comparing and analyzing the results of parameter design by transformation theory of Box and Taguchi method which uses two stage optimization procedure for Nominal Target Best(NTB). Signal-to-noise ratios as performance statistic in Taguchi method and σ_(Y) and Y￣ in Box method are used to select the optimal condition of design variable for the cases of Ouinlan report. Different control factors, adjustment factors and levels are obtained. The use of Taguchi method gave getter result than that of Box to the results of estimated value and optimal condition of design variable.

비정규분포하에서 신경회로망을 이용한 불안정 공정상태의 탐지

배홍석(Hong Seok Bae),송서일(Suh Ill Song) 대한설비관리학회 2000 대한설비관리학회지 Vol.5 No.2

N/A Numerous authors suggests that, under some conditions, neural networks offer an advantageous alternative to traditional statistical methods. Among these advantages is the employment of a pattern recognition approach rather than a complex statistical appro

신경회로망을 이용한 공정평균 변화의 조사

배홍석(Hong Seok Bae),송서일(Suh Ill Song) 대한설비관리학회 2000 대한설비관리학회지 Vol.5 No.2

N/A One of the most widely used statistical methods to improve quality and productivity is control charts. A fundamental assumption in the development of control charts is that the underlying distribution of monitored quality characteristic is normal. However

생산 공정 변화를 모니터링하기 위한 관리도의 수행 능력 비교

배홍석 ( Hong Seok Bae ) 한국경영공학회 2013 한국경영공학회지 Vol.18 No.3

This paper develops a new VSI X-CRL synthetic control chart that considers convenience of use in the field, and perception of change of process applying VSI techniques to synthetic control char,. simultaneously. We found the optimal sampling interval and various control limit factor of the suggested chart using markov chain. The results in this study are summarized as follows : First, in case that the process follows normal distribution, the proposed VSI X-CRL synthetic control chart in detecting process mean shift was performed the most excellently in aspect of statistical performance, regardless of control limit L of CRL/S control chart. Second, Detecting process shift is faster, L value is greater in small shift of process, in case that applies individual control chart, L value may be proposed more than 5 in large shift. Also, Lvlaue is proper at 4~6 that detects rapidly the small shift as well as large, in case of VSI X-CRL synthetic control chart in sample size 4 or 5 that uses the most in traditional control chart.

공정 개선을 위한 가변 샘플링 간격을 이용한 계수치 관리도의 최적 설계

배홍석(Hong-Seok Bae) 한국산업경제학회 2013 산업경제연구 Vol.26 No.5

본 연구에서는 Shewhart 계수치 관리도의 관리한계 설정 방법에 대한 문제점과 비편향 ARL 설계에 대한 개념을 고찰하였고, 기존에 개발된 VSI 기법을 이용한 계수치 관리도의 문제점들을 관리모수에 따른 ATS 곡선으로 분석한다. 그리고 이러한 문제점을 해결하기 위하여 근사 비편향 ATS 설계를 통한 관리모수들을 설정한다. 이때 설정할 관리모수는 정해진 샘플링 간격에 대한 관리상한, 관리하한, 경고한계선이 된다. 본 연구에서는 Acosta-Mejia(1999)의 연구를 확장하여 최적의 계수치 관리도에 대한 근사 비편향 ARL 설계를 제시하였고, 보다 빠른 공정변화의 이상 상태를 감지하기 위하여 VSI 기법을 이용하였다. 그리고 계수치 관리도로써 공정 개선에 대한 식별신호와 이상상태에 대한 신호를 주는 기능을 함께 가지는 관리도를 설계하는 것으로 이러한 설계 방법을 위하여 근사 비편향 ATS 특성을 가지는 계수치 관리도의 설계방법을 제시하였다. 끝으로 계수치 데이터의 특성인 연속적인 데이터가 아니므로 완벽하게 비편향성을 가지는 ATS 곡선을 설계하는 것은 불가능하기 때문에 비편향 ATS 설계에 “근사(nearly)"를 언급하였다. The proposed design method minimizes bias, i.e., an average time to signal for the shift of process from the target value (ATS) curve, as well as it applies a VSI method to an attribute control chart for detecting a process shift efficiently. First, the results of the factorial experiment obtained by various parameter circumstances show that the VSI control chart using nearly unbiased ATS design provides the smallest decreasing rate in ATS among other charts for all experimental cases. Second, we represent that a process change is rapidly detected when short sampling interval d1 approximate to dmin . In addition, the change of the sampling interval may not influence in a set of control chart parameters. Third, we demonstrate ATS curve has large bias establishing the parameter of VSI control chart using the Shewhart method. However, the proposed design method for the development of a control chart robust of sample size and control parameter size. Finally, We present an optimization model that may efficiently provide optimal ATS values for process parameters in a steady state, and that may efficiently establish parameters of correct control chart by supplying an optimization algorithm.

고품질과 저비용을 위한 통계적-경제적 관리도의 최적 설계

배홍석 ( Hong Seok Bae ) 한국생산성학회 2011 生産性論集 Vol.25 No.1

The results in this study are summarized as follows: First, in case that the process follows normal distribution, the proposed VSI X-CRL synthetic control chart in detecting process mean shift was performed the most excellently in aspect of statistical performance, regardless of control limit L of CRL/S control chart. Second, Detecting process shift is faster, L value is greater in small shift of process, in case that applies individual control chart, L value may be proposed more than 5 in large shift. Also, L vlaue is proper at 4~6 that detects rapidly the small shift as well as large, in case of VSI X-CRL synthetic control chart in sample size 4 or 5 that uses the most in traditional control chart. Third, in case of process mean shift σ=1.0, when economic optimal parameter of VSI X-CRL synthetic control chart is at L=5, n=4, the expense amount is the least. The control parameters of CRL/S control chart are determined as k=2.239, k=0.974, d1=0.100 and d2=1.397. Proposed control chart brings cost-cutting effect of 3.04% control expense less than FSI control chart. Fourth, control cost shows a tendency to increase gradually as increasing λ, parameter of exponential distribution. There is no effect in control limit effect of each control chart by increasing λ, but sampling interval, d1 and d2, goes shortly to reduce time to find special causes. Proposed VSI X-CRL synthetic control chart in this research can use more easily than traditional control chart along with applying run rule and the performance is superior to traditional control chart. It may not be difficult to establish the optimal economic control parameters to apply the practical cost parameters in field.

온라인 공정관리를 위한 VSI 계수치 관리도의 수행도 평가

배홍석 ( Hong Seok Bae ) 한국생산성학회 2013 生産性論集 Vol.27 No.4

The proposed design method minimizes bias, i.e., an average time to signal for the shift of process from the target value (ATS) curve, as well as it applies a VSI method to an attribute control chart for detecting a process shift efficiently. First, the results of the factorial experiment obtained by various parameter circumstances show that the VSI control chart using nearly unbiased ATS design provides the smallest decreasing rate in ATS among other charts for all experimental cases. Second, we illustrate that the control limits on the VSI control vhart using Shewhart method cannot satisfy false alarms when control parameters are small. This problem causes process over-control, unnecessary costs increase, and changeful factor. While, about all parameter values, a nearly unbiased ATS design satisfies restrictions for given false alarm. Third, we represent that a process change is rapidly detected when short sampling interval ? approximate to ?min . In addition, the change of the sampling interval may not influence in a set of control chart parameters. Fourth, we demonstrate ATS curve has large bias establishing the parameter of VSI control chart using the Shewhart method. However, the proposed design method for the development of a control chart robust of sample size and control parameter size. Finally, We present an optimization model that may efficiently provide optimal ATS values for process parameters in a steady state, and that may efficiently establish parameters of correct control chart by supplying an optimization algorithm.