Using a Classifier Ensemble for Proactive Quality Monitoring and\n Control: the impact of the choice of classifiers types, selection criterion,\n and fusion process

Abstract

In recent times, the manufacturing processes are faced with many external or\ninternal (the increase of customized product rescheduling , process\nreliability,..) changes. Therefore, monitoring and quality management\nactivities for these manufacturing processes are difficult. Thus, the managers\nneed more proactive approaches to deal with this variability. In this study, a\nproactive quality monitoring and control approach based on classifiers to\npredict defect occurrences and provide optimal values for factors critical to\nthe quality processes is proposed. In a previous work (Noyel et al. 2013), the\nclassification approach had been used in order to improve the quality of a\nlacquering process at a company plant; the results obtained are promising, but\nthe accuracy of the classification model used needs to be improved. One way to\nachieve this is to construct a committee of classifiers (referred to as an\nensemble) to obtain a better predictive model than its constituent models.\nHowever, the selection of the best classification methods and the construction\nof the final ensemble still poses a challenging issue. In this study, we focus\nand analyze the impact of the choice of classifier types on the accuracy of the\nclassifier ensemble; in addition, we explore the effects of the selection\ncriterion and fusion process on the ensemble accuracy as well. Several fusion\nscenarios were tested and compared based on a real-world case. Our results show\nthat using an ensemble classification leads to an increase in the accuracy of\nthe classifier models. Consequently, the monitoring and control of the\nconsidered real-world case can be improved.\n