Building a rule set for the fiber-to-yarn production process by means of soft computing techniques

An important aspect of the spinning process is the ability to predict the s pinnability of a yam and its resulting strength based on the fiber quality and machine settings. Currently available fiber-to-yam models are limited t o the so-called "black box" approach, generating an output (spinnability) w ithout containing physical, interpretable information about the process its elf. This paper presents a method to predict the spinnability and strength of a yam with a set of IF-THEN rules. The rule set is automatically generat ed using the available data by means of a new learning classifier system ca lled a fuzzy efficiency-based classifier system (FECS), which enhances the original learning classifier algorithm of Goldberg [5] by defining several rule efficiencies and introducing them into the learning strategy of the sy stem. Furthermore, FECS allows the introduction of continuous (fiber and ya m) parameters, which broaden the application fields considerably in contras t to discrete parameters alone. To this end, the generated rules are expand ed to represent fuzzy classes with corresponding membership degrees toward each fiber-to-yam data sample. Rule efficiencies and the reward mechanism a re modified to account for the membership degree of each data sample. The p aper demonstrates that the resulting prediction accuracy is good and, more importantly, also delivers additional qualitative information about the fib er-to-yarn process behavior. The generated rule set allows almost 100% acce ptable classification of yam strength in three classes. The methodologies d escribed in this paper are conveniently classified as "soft computing."