Application of Fuzzy Inference Method in Printing Pressure State Expectation System

Abstract

To solve a problem in determining the printing pressure of printing machine for real-world liquid crystal display panel (LCD) production, a Printing Pressure expectation system is proposed based on a fuzzy inference method. In real-world LCD panel production, the recognition of printing pressure conditions and its control is a very important and difficult factor that influences the product quality. It is usually performed by skilled engineers, whose performance highly depends on his tacit knowledge. In the proposed system, a fuzzy inference method is employed to solve the problem. Images of the printing area are observed with cameras, from which abstract features are extracted with image processing. The output of the system is the state of printing pressure, which is divided into 3 states: EXCESSIVE PRESSURE (EP), GOOD PRESSURE (GP), and LOW PRESSURE (LP). Based on the abstract features, the state is estimated with fuzzy membership functions. The shapes of membership functions are determined based on the sampled glasses obtained in actual LCD production line. The experiments are performed with the 2000 glasses that are also printed with actual printing machines, of which the result is compared with that of skilled engineers. It is shown that the proposed system outperforms the quality of skilled engineers. The developed system is installed in actual production line, and it is expected to increase the product quality and production speed, as well as to cut off production costs.