Real-Time Fault Detection on Textiles Using Opto-electronic Processing

This work is concerned with the study of an automatic fault detection system for textiles. Opto-electronic data processing makes it possible to detect and characterize faults. The method is based on textile surface analysis by Fourier transform. In the first phase of this project, we present a theoretical study of the principles of optical fault detection. Modeling of a textile sample makes it possible to examine the influence of several parameters in the diffraction spectrum when faults appear. In the second phase, an experimental analysis of this diffraction spectrum by means of a photovoltaic cell matrix presents the parameters that distinguish the normal situation (i.e., when there are no faults) from the degraded situation (when there are defects). These pa rameters are exploited by an analog data processing unit in order to perform real-time fault detection and localization. This system also makes it possible to evaluate the number of yarns per unit length and is an original method for measuring the diameter of yarns in real time as well. From the results, a set of specifications has been established in order to build a dedicated sensor. Experiments on different textiles using a prototype of this sensor running on a cloth inspecting table reveal the efficiency of the detection process.