Difficult multivariate industrial control problems can be solved by combining standard control theory, adaptive computer vision algorithms and intelligent modeling into an overarching generalized control system. Creating the foundations for such a system, to be implemented on an industrial potato peeler, was the scope of this thesis. Computer vision algorithms that provided quantifiable metrics from the peeling process were developed using data gathered at a potato research center. Experiments were performed controlling the steamtime and pressure of the peeler and the size and seasonality of the potatoes. Thermal signatures and optical videos of peeled potatoes were recorded throughout testing. It was found that smaller potatoes are more difficult to peel and changes in pressure do not correlate with changes in peel efficiency. Recommendations were made for the next steps towards intelligent control of industrial potato peeling processes.
Zircon - This is a contributing Drupal Theme Design by WeebPal.
