• 품질경영학회지
• /
• 제29권4호
• /
• pp.38-53
• /
• 2001

In resolving industrial quality control problems, a vector of multiple quality characteristic variables is involved rather than a single variable. However, it is not guaranteed that a multivariate control chart based on statistical methods can monitor abnormal signal in case that small changes of relationship between each variables causes abnormal production process. Hence a quality control system for real-time monitoring of the multi-dimensional quality characteristic vector under a multivariate normal process is needed to enhance tile production system quality performance. A pattern analysis approach based on self-organizing map (SOM), an unsupervised learning technique of neural network, is applied to the design of such a quality control system. In this study we present a new material quality control system based on pattern analysis approach and illustrate the effectiveness of proposed system using actual electronic company material data.

• Journal of the Korean Data and Information Science Society
• /
• 제26권2호
• /
• pp.525-533
• /
• 2015

Modern production process is a very complex structure combined observations which are correlated with several factors. When the error signal occurs in the process, it is very difficult to know the root causes of an out-of-control signal because of insufficient information. However, if we know the time of the change, the system can be controlled more easily. To know it, we derive a maximum likelihood estimator (MLE) of the change point in a process when observations are from a multivariate IMA(1,1) process by monitoring residual vectors of the model. In this paper, numerical results show that the MLE of change point is effective in detecting changes in a process.

• Smart Structures and Systems
• /
• 제14권3호
• /
• pp.377-395
• /
• 2014

Vibration-based fault detection and condition monitoring of rotating machinery, using statistical process control (SPC) combined with statistical pattern recognition methodology, has been widely investigated by many researchers. In particular, the discrete wavelet transform (DWT) is considered as a powerful tool for feature extraction in detecting fault on rotating machinery. Although DWT significantly reduces the dimensionality of the data, the number of retained wavelet features can still be significantly large. Then, the use of standard multivariate SPC techniques is not advised, because the sample covariance matrix is likely to be singular, so that the common multivariate statistics cannot be calculated. Even though many feature-based SPC methods have been introduced to tackle this deficiency, most methods require a parametric distributional assumption that restricts their feasibility to specific problems of process control, and thus limit their application. This study proposes a nonparametric multivariate control chart method, based on multiscale wavelet scalogram (MWS) features, that overcomes the limitation posed by the parametric assumption in existing SPC methods. The presented approach takes advantage of multi-resolution analysis using DWT, and obtains MWS features with significantly low dimensionality. We calculate Hotelling's $T^2$-type monitoring statistic using MWS, which has enough damage-discrimination ability. A bootstrap approach is used to determine the upper control limit of the monitoring statistic, without any distributional assumption. Numerical simulations demonstrate the performance of the proposed control charting method, under various damage-level scenarios for a bearing system.

• 산업공학
• /
• 제24권1호
• /
• pp.8-14
• /
• 2011

Principal Component Analysis (PCA) reduces the dimensionality of the process by creating a new set of variables, Principal components (PCs), which attempt to reflect the true underlying process dimension. However, for highly nonlinear processes, this form of monitoring may not be efficient since the process dimensionality can't be represented by a small number of PCs. Examples include the process of semiconductors, pharmaceuticals and chemicals. Nonlinear correlated process variables can be reduced to a set of nonlinear principal components, through the application of Kernel Principal Component Analysis (KPCA). Support Vector Data Description (SVDD) which has roots in a supervised learning theory is a training algorithm based on structural risk minimization. Its control limit does not depend on the distribution, but adapts to the real data. So, in this paper proposes a non-linear process monitoring technique based on supervised learning methods and KPCA. Through simulated examples, it has been shown that the proposed monitoring chart is more effective than $T^2$ chart for nonlinear processes.

• Journal of the Korean Data and Information Science Society
• /
• 제26권6호
• /
• pp.1353-1366
• /
• 2015

공정의 안정성을 평가하기 위해 사용되는 Shewhart 관리도 기법은 최근 다양한 분야에서 널리 응용되고 있지만, 품질 특성치에 대한 엄격한 확률분포를 가정한다. 하지만 현업에서 수집되고 있는 데이터의 확률분포는 알려진 경우가 많지 않으며, 다변량 데이터로 확장될수록 확률분포를 결정하는데 더 큰 어려움이 따른다. 이러한 문제점을 해결하기 위해 다양한 비모수 관리도 기법이 연구되었는데, 최근 연구되고 있는 비모수 관리도 기법 중 하나인 RBF (Radial Basis Function) 커널 기반의 SVDD (Support Vector Data Description) 관리도는 관리상태 하의 데이터 영역에 대한 경계를 결정함으로써 공정의 이상상태를 탐지하는 기법으로 K 관리도로 불리우며, 다양한 분야에서 적용되고 있다. 그런데 K 관리도를 적용하기 위해서는 관리도의 성능을 결정짓는 커널모수 등의 선택이 중요하며, 관리도를 작성하기 전에 미리 결정되어야 한다. 이를 위해 기존의 연구들은 격자 탐색법 등을 활용하여 모수를 결정하고 있지만, 선택 가능한 범위에 대한 반복적인 계산으로 최적값을 선택하고 있어 계산 비용이 커지고 또 시간 등의 문제로 실제 문제에 적용하기 어려운 점이 있다. 따라서 본 연구에서는 데이터의 구조에 따라 모의실험을 통해 선택 가능한 영역에서의 효율성을 비교 검토하고, 이를 바탕으로 쉽게 적용할 수 있는 새로운 모수 선택 방법을 제안하고자 한다. 이를 통해 데이터 구조에 대해 강건함을 보이는 모수의 선택과 K 관리도의 구성을 논의하고 실제 자료에 적용해 보았다.

• 응용통계연구
• /
• 제35권3호
• /
• pp.335-346
• /
• 2022

통계적 공정관리에서 프로파일 관리도란 다수의 품질 특성치 간 함수관계의 변화를 탐지하는 것을 말한다. 두 변수 간 선형의 관계가 있는 경우, 선형 프로파일을 가정하고 절편과 기울기가 일정한지 모니터링한다. 이때 선형 프로파일에 관한 대부분의 기존 연구에서는 모든 프로파일에서 설명변수의 관측치가 동일하다고 가정한다. 그러나 프로파일마다 설명변수의 값이 랜덤하게 관측되는 경우도 존재한다. 본 논문에서는 단순 선형 프로파일 모니터링에서 설명변수가 프로파일마다 랜덤하게 관측된다는 가정하에 기존의 방법을 확장 적용하고자 한다. 모의실험을 통해 제안한 방법의 탐지 성능을 확인하고 네트워크 침입 탐지 알고리즘 성능을 비교하기 위한 NSL-KDD 데이터를 이용하여 제안된 침입 탐지 결과를 비교해 보았다.

• 대한안전경영과학회지
• /
• 제9권1호
• /
• pp.145-156
• /
• 2007

This paper is to propose model classification and evaluation of measurement uncertainty. In order to obtain type A and B uncertainty, variety of measurement mathematical models are illustrated by example. The four steps to evaluate expanded uncertainty are indicated as following; First, to get type A standard uncertainty, measurement mathematical models of single, double, multiple, design of experiment and serial autocorrelation are shown. Second, to solve type B standard uncertainty measurement mathematical models of empirical probability distributions and multivariate are presented. Third, type A and B combined uncertainty, considering sensitivity coefficient, linearity and correlation are discussed. Lastly, expanded uncertainty, considering degree of freedom for type A, B uncertainty and coverage factor are presented with uncertainty budget. SPC control chart to control expanded uncertainty is shown.

• Journal of the Korean Data and Information Science Society
• /
• 제22권4호
• /
• pp.803-809
• /
• 2011

Properties of multivariate Shewhart and CUSUM charts for monitoring variance-covariance matrix, specially focused on correlation coefficient components, are investigated. The performances of the proposed charts based on control statistic Lawley-Hotelling $V_i$ and likelihood ratio test (LRT) statistic $TV_i$ are evaluated in terms of average run length (ARL). For monitoring correlation coe cient components of dispersion matrix, we found that CUSUM chart based on $TV_i$ gives relatively better performances and is more preferable, and the charts based on $V_i$ perform badly and are not recommended.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 Proceedings of the Korea Automatic Control Conference, 11th (KACC); Pohang, Korea; 24-26 Oct. 1996
• /
• pp.18-18
• /
• 1996

This paper reviews some of the most prominent and promising areas of chemical process control both in relations to batch and continuous processes. These areas include the modeling, optimization, control and monitoring of chemical processes and entire plants. Most of these areas explicitly utilize a model of the process. For this purpose the types of models used are examined in some detail. These types of models are categorized in knowledge-driven and datadriven classes. In the areas of modeling and optimization, attention is paid to batch reactors using the Tendency Modeling approach. These Tendency models consist of data- and knowledge-driven components and are often called Gray or Hybrid models. In the case of continuous processes, emphasis is placed in the closed-loop identification of a state space model and their use in Model Predictive Control nonlinear processes, such as the Fluidized Catalytic Cracking process. The effective monitoring of multivariate process is examined through the use of statistical charts obtained by the use of Principal Component Analysis (PMC). Static and dynamic charts account for the cross and auto-correlation of the substantial number of variables measured on-line. Centralized and de-centralized chart also aim in isolating the source of process disturbances so that they can be eliminated. Even though significant progress has been made during the last decade, the challenges for the next ten years are substantial. Present progress is strongly influenced by the economical benefits industry is deriving from the use of these advanced techniques. Future progress will be further catalyzed from the harmonious collaboration of University and Industrial researchers.

• Korean Chemical Engineering Research
• /
• 제45권1호
• /
• pp.32-38
• /
• 2007

통계적인 공정 제어 기법을 회분식 공정에 적용하여, 일반적인 회분식 공정의 데이터를 통해 보다 빠르고, 손쉽게 공정의 상태를 진단할 수 있는 시스템을 구현해 보았다. 대표적인 회분식 공정의 하나인 반도체 식각공정과 반회분식 스타이렌-부타디엔 고무 생산 공정의 데이터를 이용하여 공정 변수와 공정의 상태간의 연관 관계를 규명할 수 있는 모델을 수립하였으며, 이 모델의 출력(output) 결과를 이용해 통계적 공정 제어 차트를 구성하고, 시간에 따른 공정의 추이를 분석해 이상을 판별해 보았다. 회분식 공정의 다축(multi-way) 데이터를 두개의 축으로 만드는 펼치기(unfolding) 과정을 거쳤으며, 모델링 방법으로는 Support Vector Regression 및 Partial Least Square 등의 다변량 회귀분석 방법을 이용하였다. 또한 에러차트 및 변수 기여도 차트(variable contribution chart)를 이용해 이상의 세기, 형태 및 이상 데이터에 대한 각 변수들의 기여도를 계산해 보았으며, 그 결과 이상의 발생 유무 및 발생시점 뿐만아니라 이상의 세기 및 원인 까지 진단해 볼 수 있는 우수한 성능을 보이는 것을 확인할 수 있었다.