Fault detection and identification with a new feature selection based on mutual information

Verron, Sylvain; Tiplica, Téodor; Kobi, Abdessamad

doi:10.1016/j.jprocont.2007.08.003

Cited by 143 publications

(78 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[22][23][24][25][26][27][28] Meanwhile, machine learning techniques, e.g., discriminant analysis (DA), neural network (NN), expert systems, support vector machines (SVM), Bayesian belief network (BBN), and mutual information, have been explored to address the complex process monitoring problems with some success. [29][30][31][32][33][34][35][36] Despite a rich body of literature in chemical process monitoring, most of the MSPM approaches rely on the assumption that the normal process data come from a single operating region and follow a unimodal distribution. The popular PCA/ PLS methods, for example, require the operating data to obey unimodal Gaussian distribution approximately so as to guarantee the valid T 2 and SPE control limits.…”

Section: Introductionmentioning

confidence: 99%

Multimode process monitoring with Bayesian inference‐based finite Gaussian mixture models

2008

View full text Add to dashboard Cite

in Wiley InterScience (www.interscience.wiley.com).For complex industrial processes with multiple operating conditions, the traditional multivariate process monitoring techniques such as principal component analysis (PCA) and partial least squares (PLS) are ill-suited because the fundamental assumption that the operating data follow a unimodal Gaussian distribution usually becomes invalid. In this article, a novel multimode process monitoring approach based on finite Gaussian mixture model (FGMM) and Bayesian inference strategy is proposed. First, the process data are assumed to be from a number of different clusters, each of which corresponds to an operating mode and can be characterized by a Gaussian component. In the absence of a priori process knowledge, the Figueiredo-Jain (F-J) algorithm is then adopted to automatically optimize the number of Gaussian components and estimate their statistical distribution parameters. With the obtained FGMM, a Bayesian inference strategy is further utilized to compute the posterior probabilities of each monitored sample belonging to the multiple components and derive an integrated global probabilistic index for fault detection of multimode processes. The validity and effectiveness of the proposed monitoring approach are illustrated through three examples: (1) a simple multivariate linear system, (2) a simulated continuous stirred tank heater (CSTH) process, and (3) the Tennessee Eastman challenge problem. The comparison of monitoring results demonstrates that the proposed approach is superior to the conventional PCA method and can achieve accurate and early detection of various types of faults in multimode processes.

show abstract

Section: Introductionmentioning

confidence: 99%

Multimode process monitoring with Bayesian inference‐based finite Gaussian mixture models

2008

View full text Add to dashboard Cite

show abstract

“…It is well-known that keeping non-informative descriptors in the model increases the classification error rate of new observations. The final goal when using a classifier is to obtain the lowest classification error of observations which did not participate to the construction of discriminant functions with the smallest number of descriptors [25].…”

Section: -P4mentioning

confidence: 99%

On the quality of acoustical measures when evaluating fruits quality

Tiplica

Verron

Grémy-Gros

et al. 2015

Int. J. Metrol. Qual. Eng.

View full text Add to dashboard Cite

Abstract. In this article we show that, without any technological modification, an acoustic firmness sensor used for estimating fruits firmness, can also be used in order to discriminate between different apple varieties. A repetability and reproducibility (R&R) analysis shows that the device is capable to measure not just the mass and the resonance frequency of fruits but also some other key spectral characteristics as well. Thus, 18 key features were identified and used for the classification of apples belonging to 10 different varieties. The prospects of this study are interesting in the context of sorting automatically fruits.

show abstract

“…For example, given a learning sample of a bivariate system with three different known faults as illustrated in the figure 5, we can easily use supervised classification to classify a new faulty observation. A feature selection can be used in order to select only the most informative variables of the problem (Verron et al (2008)). In this study, we will use the Bayesian network as a supervised classification tool.…”

Section: Bayesian Network For Fault Diagnosismentioning

confidence: 99%