A Methodology for Manufacturing Process Signature Analysis (Classic Reprint) - Softcover

Steven D. Eppinger

 
9781332269860: A Methodology for Manufacturing Process Signature Analysis (Classic Reprint)

Synopsis

This book presents a methodology for manufacturing process signature analysis, which aims to enhance process control systems through data analysis. The author argues that improving control systems involves collecting more information about the process and using it effectively. The book proposes a three-step analysis method, which consists of feature extraction, feature selection, and classification, to establish a relationship between the collected information (process signatures) and the quality of the process output. This relationship can be used for online monitoring and control. The book covers the advantages and limitations of various signature analysis tools, ranging from statistical process control to artificial neural networks. The author emphasizes the importance of feature selection in reducing the complexity of classification problems and discusses different feature selection schemes. One of the key contributions is the use of the first layer weights in a neural network topology to identify the most useful components of the input signature. The book concludes by exploring future research directions, highlighting the need for accurate quality metrics, effective sensor selection, and efficient neural network training algorithms. Overall, this book provides a comprehensive overview of process signature analysis and its application in manufacturing process monitoring and control.

"synopsis" may belong to another edition of this title.

Other Popular Editions of the Same Title