A knowledge-based system for spinning management

Abstract

This paper describes the development of a knowledge-based system to support and improve the spinning process. It is divided in three parts. The first part describes a statistical study using sixteen raw cottons and three kinds of yarn. A set of mathematical models describes the relationships between cotton fiber and yarn properties. Eighteen different models have been identified according to different control processes (HVI motion control and Spinlab and individual instruments) and different production processes (ring and rotor spinning, OE RU 14 Spincomat and Autocoro 117). The second part details the implementation of the knowledge-based system using the programming languages PROLOG and C. The knowledge base uses a framework shell (flex or forward logical expert systems) that interfaces with PROLOG. This prototype can be used either to predict the characteristics of the yarn according to those of the available raw materials, or to select the raw materials adequate to produce yarn with specific characteristics. The last section presents some conclusions of the work.