ABSTRACT
To compete on a global level, manufacturers must strive to find new solutions to
ensure high product quality while maximizing productivity. One way to address many
of the challenges this creates is to turn increasingly to mechatronic design approach
and automation. To successfully realize automation, a mechatronic system for online
process monitoring is required to take the place of an expert's judgment. This
paper outlines the use of a statistical multivariate technique called Projection to
Latent Structure (PLS), and applies it to the monitoring of a machining process as an
application. This approach is used to integrate machine tool sensory data from a
milling machine. Experiments were conducted on a milling machine under three
conditions, sharp, worn and chatter tools. The score models were tested under
different conditions with the results showing that the proposed technique can be
used for tool wear monitoring and can successfully differentiate between new
process conditions.