The significance of online process monitoring of discrete part manufacturing usmg multivariate analysis is its ability to help the Canadian manufacturing industry compete in the global market. Process monitoring can accomplish this by:​ assessing the state of a machining system for unusual occurrences, moving the part quality prediction upstream, and producing higher volumes of in specification parts for improved profits.

The focus of this research was discrete process monitoring of a turning operation in a laboratory at the McMaster Manufacturing Research Institute (MMRI) and an industrial machining cell at Glueckler Metal Incorporated (GMI). Both applications involved instrumentation of a lathe with current sensors, an accelerometer and thennocouples. Serial port communication between the machine control panel and computer was established to allow for online automated data acquisition. The multivariate latent model applied was principal component analysis to develop correlations among the machining process infonnation. Principal component analysis was successful in identifying the occun'ence of an out of balance spindle, unusual surface finish, changes in depth of cut, and a worn tool in laboratory tests, through the use of simple control plots. Industrial results validated the ability of the system to differentiate machining data from one day to another, and to isolate an unusual piece of barstock that led to slightly below average part dimensions,

The difficulties experienced in the transitioning from laboratory conditions to industrial testing were discussed. This infonnation will allow future researchers to continue adding new process monitoring sensors to the system.

In conclusion, this research demonstrated the ability of online process monitoring of discrete part manufacturing in a laboratory setting;​ and brings the MMRI and GMI closer to having a fully implemented process monitoring system for part quality prediction and machine maintenance.