A METHODOLOGY FOR MANUFACTURING PROCESS SIGNATURE ANALYSIS

Improvement of control systems entails collection of more information about the process and/or more effective use of that information. We present manufacturing process signature analysis to construct a relationship between collected information (process signatures) and the quality of process output, which can be used for on-line monitoring and control. The general procedure consists of feature extraction, feature selection, and classification. Extraction of large sets of features from signatures is straightforward, and several classification schemes are available, with neutal networks being the most general and powerful. Feature selection, however, is generally quite difficult for complex data structures. We present several feature extraction methods and show that neural networks can be useful in choosing different feature sets. Using a data set from an automated solder joint inspection system, we demonstrate the unique capabilities of neural networks for both feature selection and classification, using more traditional statistical classification techniques as a benchmark.