Unsupervised feature selection based on fuzzy partition optimization for industrial processes monitoring

Abstract

Industrial processes have enormous volumes of complex and high dimensional data available, with poorly defined domains and redundant, noisy or inaccurate measures with unknown parameters. Therefore, using just relevant and informative variables will decrease the high dimensionality in the data and will facilitate the use of data-based methods for developing monitoring and fault detection systems. In this paper, a new unsupervised feature selection method based on partition optimization for fuzzy clustering based monitoring systems is proposed. Application on monitoring an intensification reactor, the 'open plate reactor (OPR)' is studied. Results show fewer variables are needed to classify process data into accurate functional states.