• Korean Chemical Engineering Research
• /
• 제45권1호
• /
• pp.87-92
• /
• 2007

최근 공정의 이상을 감지하고 진단하기 위한 공정 모니터링 시스템의 개발이 공정 시스템 분야에서 많은 주목을 받고 있다. 공정으로부터 얻어지는 데이터는 공정의 특성에 대한 유용한 정보를 제공하고 이는 공정의 모델링과 모니터링 그리고 제어에 사용된다. 현대의 화학 및 환경 공정은 고차원적인 특성과 변수간의 강한 상관관계와 동특성 그리고 비선형적 특성을 가지고 있어 모델 기반 접근을 통해 공정을 분석하는 것을 쉽지 않다. 이러한 모델 기반 접근의 한계를 극복하기 위해 많은 시스템 엔지니어와 연구자들이 주성분 분석법(principal component analysis, PCA) 또는 부분 최소 자승법(partial least squares, PLS)과 같은 다변량 분석을 접목한 통계 기반 접근법에 초점을 맞추고 있다. 또한 동특성, 비선형성 등과 같은 특성을 가진 공정에 적용하기 위해 많은 다변량 분석법들이 보완되었다. 여기에서는 동적 주성분 분석법(dynamic PCA)과 케노니컬 변수 분석법(canonical variate analysis)을 이용한 결측 데이터의 예측법과 공정 변수의 복원을 통한 센서 오작동의 판별법에 대해 언급해 보고자 한다.

• 한국정보통신학회논문지
• /
• 제24권2호
• /
• pp.192-197
• /
• 2020

대규모 발전소나 화학공정과 같은 다변량 공정은 매우 위험한 환경에서 운전되기 때문에 고장이 발생하면 심각한 인적·물적 손실이 발생할 수 있다. 따라서 시스템의 고장을 사전에 탐지할 수 있는 온라인 모니터링 기술이 필수적이다. 본 논문에서는 세 가지의 다른 다변량 공정 데이터에 ICA를 적용하여 고장탐지를 수행하였고, PCA와 성능을 비교하였다. ICA 기반의 고장탐지 절차는 크게 오프라인 과정과 온라인 과정으로 나뉜다. 오프라인 과정에서는 시스템이 정상일 때 계측된 데이터를 이용하여 고장판별을 위한 문턱 값을 설정한다. 그리고 온라인 과정에서는 실시간으로 계측되는 질의벡터에 대한 통계량을 계산한 후, 계산된 통계량과 사전에 정의된 문턱 값과 비교하여 고장을 판별한다. 본 논문에서 이용한 세 가지의 다변량 공정 데이터에 실험한 결과, ICA 기반 고장탐지 방법이 시스템의 고장을 사전에 탐지하였고, PCA 보다 우수한 고장탐지 성능을 보여주었다.

• Journal of Ginseng Research
• /
• 제43권4호
• /
• pp.508-516
• /
• 2019

Background: Ginsenosides are not only the principal bioactive components but also the important indexes to the quality assessment of Panax ginseng Meyer. Their contents in cultivated ginseng vary with the growth environment and age. The present study aimed at evaluating the significant difference between 36 cultivated ginseng of different cultivation areas and ages based on the simultaneously determined contents of 14 ginsenosides. Methods: A high-performance liquid chromatography (HPLC) coupled with triple quadrupole mass spectrometer (MS) method was developed and used in the multiple reaction-monitoring (MRM) mode (HPLC-MRM/MS) for the quantitative analysis of ginsenosides. Multivariate statistical analysis, such as principal component analysis and partial least squares-discriminant analysis, was applied to discriminate ginseng samples of various cultivation areas and ages and to discover the differentially accumulated ginsenoside markers. Results: The developed HPLC-MRM/MS method was validated to be precise, accurate, stable, sensitive, and repeatable for the simultaneous determination of 14 ginsenosides. It was found that the 3- and 5-yr-old ginseng samples were differentiated distinctly by all means of multivariate statistical analysis, whereas the 4-yr-old samples exhibited similarity to either 3- or 5-yr-old samples in the contents of ginsenosides. Among the 14 detected ginsenosides, Rg1, Rb1, Rb2, Rc, 20(S)-Rf, 20(S)-Rh1, and Rb3 were identified as potential markers for the differentiation of cultivation ages. In addition, the 5-yr-old samples were able to be classified in cultivation area based on the contents of ginsenosides, whereas the 3- and 4-yr-old samples showed little differences in cultivation area. Conclusion: This study demonstrated that the HPLC-MRM/MS method combined with multivariate statistical analysis provides deep insight into the accumulation characteristics of ginsenosides and could be used to differentiate ginseng that are cultivated in different areas and ages.

• 품질경영학회지
• /
• 제34권4호
• /
• pp.102-109
• /
• 2006

This paper investigates the effects of non-normality on the performance of univariate and multivariate cumulative sum(CUSUM) control charts for monitoring the process mean. In-control and out-of-control average run lengths of the charts are examined for the univariate/multivariate lognormal and t distributions. The effects of the reference value and the correlation coefficient under the non-normal distributions are also studied. Simulation results show that the CUSUM charts with small reference values are robust to non-normality but those with moderate or large reference values are sensitive to non-normal data especially to process data from skewed distributions. The performance of the chart to detect mean shift of a process is not invariant to the direction of the shift for skewed distributions.

• Journal of the Korean Data and Information Science Society
• /
• 제24권4호
• /
• pp.919-929
• /
• 2013

We compared two basic methods, combine-accumulate method and accumulate-combine method, using the past quality information in multivariate quality control procedure for monitoring mean vector of multivariate normal process. When small or moderate shifts have occurred, accumulate-combine method yields smaller average run length (ARL) and average time to signal (ATS) than combine-accumulate method. On the other hand, we have found from our numerical results that combine-accumulate method has better performances in terms of switching behavior than accumulate-combine method. In industry, a quality engineer could select one of the two method under the comprehensive consideration about the required time to signal, switching behavior, and other physical factors in the production process.

• 통합자연과학논문집
• /
• 제5권3호
• /
• pp.151-156
• /
• 2012

Because of the equivalence between control chart procedures and hypothesis testing, we propose to use likelihood ratio test (LRT) statistic $Z_i^2$ as the multivariate control statistic for simultaneous monitoring means of the multivariate normal process. Properties and comparisons of the proposed control charts are explored and conducted for matched fixed sampling interval (FSI) and variable sampling interval (VSI) with two sampling interval charts. The result of numerical comparisons shows that EWMA chart with two sampling interval procedure is more efficient than the corresponding FSI chart for small or moderate changes. When large shift of the process has occurred, we also found that Shewhart chart is more efficient than EWMA chart.

• Nuclear Engineering and Technology
• /
• 제37권2호
• /
• pp.177-184
• /
• 2005

The Multivariate State Estimation Technique (MSET) is being used in Nuclear Power Plants for sensor and equipment condition monitoring. This paper presents the use of regularization methods for optimizing MSET's predictive performance. The techniques are applied to a simulated data set and a data set obtained from a nuclear power plant currently implementing empirical, on-line, equipment condition monitoring techniques. The results show that regularization greatly enhances the predictive performance. Additionally, the selection of prototype vectors is investigated and a local modeling method is presented that can be applied when computational speed is desired.

• Communications for Statistical Applications and Methods
• /
• 제30권4호
• /
• pp.369-388
• /
• 2023

In this paper, we develop the two-step procedure that detects and estimates the position of structural changes for multivariate nonstationary time series, either on mean parameters or second-order structures. We first investigate the presence of mean structural change by monitoring data through the aggregated cumulative sum (CUSUM) type statistic, a sequential procedure identifying the likely position of the change point on its trend. If no mean change point is detected, the proposed method proceeds to scan the second-order structural change by modeling the multivariate nonstationary time series with a multivariate locally stationary Wavelet process, allowing the time-localized auto-correlation and cross-dependence. Under this framework, the estimated dynamic spectral matrices derived from the local wavelet periodogram capture the time-evolving scale-specific auto- and cross-dependence features of data. We then monitor the change point from the lower-dimensional approximated space of the spectral matrices over time by applying the dynamic principal component analysis. Different from existing methods requiring prior information on the type of changes between mean and covariance structures as an input for the implementation, the proposed algorithm provides the output indicating the type of change and the estimated location of its occurrence. The performance of the proposed method is demonstrated in simulations and the analysis of two real finance datasets.

• 한국지하수토양환경학회:학술대회논문집
• /
• 한국지하수토양환경학회 2001년도 추계학술발표회
• /
• pp.155-159
• /
• 2001

A study on multivariate statistical classification of ground water hydrographs was conducted. The vast data of national ground water monitoring network (78 sites of alluvium) were used. 6 factors were selected to classify the ground water level change. Factor analysis was proved to be useful tool for classifying vast hydrogeological data.

• Smart Structures and Systems
• /
• 제15권5호
• /
• pp.1329-1344
• /
• 2015

The stiffness of a structure is one of several structural signals that are useful indicators of the amount of damage that has been done to the structure. To accurately estimate the stiffness, an equation of motion containing a stiffness parameter must first be established by expansion as a linear series model, a Taylor series model, or a power series model. The model is then used in multivariate autoregressive modeling to estimate the structural stiffness and compare it to the theoretical value. Stiffness assessment for modeling purposes typically involves the use of one of three statistical model refinement approaches, one of which is the efficient Akaike information criterion (AIC) proposed in this paper. If a newly added component of the model results in a decrease in the AIC value, compared to the value obtained with the previously added component(s), it is statistically justifiable to retain this new component; otherwise, it should be removed. This model refinement process is repeated until all of the components of the model are shown to be statistically justifiable. In this study, this model refinement approach was compared with the two other commonly used refinement approaches: principal component analysis (PCA) and principal component regression (PCR) combined with the AIC. The results indicate that the proposed AIC approach produces more accurate structural stiffness estimates than the other two approaches.