An adaptive fuzzy model predictive control system for the textile fiber industry

Abstract

Many process steps in the production of modern fibers and yarns are hallmarked by their high complexity and require thus a great know-how of the operating personnel. To support their work an adaptive fuzzy model predictive control system has been designed whose characteristics are sketched here. The system is build upon an expert specified rule base and comprises a data driven optimization component. Two disparate types of measures are collected and exploited for this: continuous available online measurements stemming from machine sensors and sporadic analyses from laboratory spot tests. Further key feature is an inferential control mechanism that allows for continuous control in absence of the primary values from the lab. © Springer-Verlag Berlin Heidelberg 2010.