Employing machine learning techniques in monitoring autocorrelated profiles

• DOI: 10.1007/s00521-023-08483-3
• Publication Date: 2023-04-29T17:01:51Z
• Authors: Ali Yeganeh; Arne Johannssen; Nataliya Chukhrova; Saddam Akber Abbasi; Farhad Pourpanah
• Keywords: Statistical Process Control; Statistical Inference

Abstract In profile monitoring, it is usually assumed that the observations between or within each are independent of other. However, this assumption often violated in manufacturing practice, and utmost importance to carefully consider autocorrelation effects underlying models for monitoring. For reason, various statistical control charts have been proposed monitor profiles when between- within-data correlated Phase II, which main aim develop with quicker detection ability. As a novel approach, study aims employ machine learning techniques as instead approaches monitoring between-profile autocorrelations. Specifically, new input features based on conventional chart statistics normalized estimated parameters defined capable adequately accounting between-autocorrelation effect profiles. addition, six extended compared by means Monte Carlo simulations. The simulation results indicate can obtain more accurate charts. Moreover, adaptive neuro-fuzzy inference systems outperform other