LabVIEW implementation with principal components analysis for multivariate statistical process monitoring

Access full-text files




Wang, Qi (master of science in electrical and computer engineering), 1980-

Journal Title

Journal ISSN

Volume Title



A major technical challenge facing the manufacturing and process control industries is the need to improve production consistency and provide early warning of process faults and malfunctions. This is particular important in highly demanding situations where the process is subject to varying material properties, changing market needs and specifications, and fluctuating operating conditions due to equipment or process degradation. Multivariate Statistical Process Control based on the techniques of Principal Component Analysis (PCA) provides tools for on-line performance monitoring. PCA generate reduced dimension data set according to the original data with fewer number of variables while still maintain the capability of describing the embedded variation causes. Several techniques that support the PCA process control model such as Hotelling's T², Squared Prediction Error and combined fault detection indices are adopted to identify if data has fault information; in addition, fault reconstruction via optimization and contribution plots methods play roles in finding the causes or locations of the faulty data. With hope of introducing those techniques to the manufacturing line with the least cost and most convenience of use, computer applications implemented by LabVIEW with interactive functionalities and graphical tools are illustrated. Both the methodologies and the computer implementation will be introduced in this thesis


LCSH Subject Headings