A Diagnostic Expert System for the Coloration of Polyester Materials

Abstract

The resolution of textile related problems across the globe necessitates collaboration amongst interdisciplinary teams of experts where it must be recognized that the perturbation of one part would affect many others. Resolution of problems in the coloration industry, for instance, can be very time consuming and highly challenging as the process is influenced by a large number of, often, interactive variables. In addition, there is a growing realization that the complex problems faced by the industry go beyond the abilities of local individual experts, and that there is a diminishing expertise base to address various challenges. The utilization of a field of artificial intelligence known as expert systems can help address many of the problems in the textile industry more effectively, and economically. Thus the purpose of this study was to develop a knowledge-based diagnostic expert system (ES) to troubleshoot issues in the coloration of polyester materials as the most widely used synthetic fiber. This was accomplished in four stages: selection of a suitable ES development tool, knowledge acquisition, design and development of appropriate modules and finally testing and evaluation of the system. A Windows based version of C Language Integrated Production System (wxfuzzyCLIPS: ver. 1.64) was selected as the development shell. A review of the most commonly occurring problems in the dyeing of polyester in the forms of yarn, knitted and woven fabric with various classes of disperse dyes and employing different dyeing methods and machines was carried out. In addition over fifty coloration experts from several countries were contacted and asked to participate in an electronic technical survey of which fourteen coloration experts (with a range of industrial and/or academic experience from 6 to 45 years) responded. Additionally, the knowledge reported in the literature was incorporated into this database. After the knowledge base of the system was developed, it was verified, validated and evaluated using human experts as well as case studies. The validation and verification process involved debugging, testing and evaluation of various modules. Assessments were based on the comparative performance of human experts and the expert system through several faulty dyed samples. The system should be ideally tested in an industrial setting to ensure a satisfactory performance is achieved as verified by several human experts. The results of testing and evaluation demonstrated good performance of the system when compared to human experts. This system is a powerful tool to diagnose issues related to the coloration of polyester, which can save time, energy and reduce the environmental impact of the sector. It can also be used as a training tool for novice and practical dyers and a decision making aid for seasoned experts as well as management personnel.