Diagnosing Faulty Products in TFT-LCD Manufacturing Processes by Support Vector Machines with Principal Components Analysis

Thin-film transistor liquid-crystal display (TFT-LCD) manufacturing in Taiwan is booming; and the revenues from the TFT-LCD industry have grown significantly in recent years. One of the main problems in the TFT-LCD manufacturing process is to diagnose faulty products. This study employed support vector machines (SVM) with principal components analysis (PCA) to diagnose root causes in sputtering operations of the TFT-LCD industry. The PCA technique was used to transfer original manufacturing parameters into essential factors; and the SVM model was applied in classifying faulty products. Besides, the genetic and tabu (GA/TS) search algorithms were utilized to select SVM parameters. In terms of classification accuracy and efficiency, simulation results indicated that the SVM with PCA procedure is a feasible and promising way to diagnose faulty products in TFT-LCD manufacturing processes.