Worsted Process Quality Control with KDD-based intelligent Methods

摘要

The textile production is a complex industrial process. Due to absence of an integral mathematical model, quality control in textile process has been hard problem for a long time. In fact, the existing process data which were recorded in large quantities to ensure the ability to trace production steps can also be used to improve the quality itself. The field of data mining (DM) and knowledge discovery from database (KDD) has emerged as a new discipline in engineering and computer science. This paper investigates knowledge discovery methods from the textile industrial database, and presents a novel KDD-based intelligent control model (ICM) for worsted process quality. Firstly, before real production, different raw materials will be logically selected with results from ICM simulation in order to improve quality of product and to minimize production cost. Second, simulating domain expert's experience, abnormal flaws in real production can be removed to a great extent, and process quality can be farther stabilized. The model architecture, experiments and methods are presented in this paper, respectively. The applied cases are demonstrated that the intelligent model to control the worsted quality is promising.