Abstract:
The plant-model ratio, developed to diagnose model-plant mismatch present in
model-based controllers, inherits the same limitations from the frequency-based analysis that
the method is based on. Nonetheless, the plant-model ratio shows the capacity to counteract the
effect of non-linear dynamics within processes due to the ability to diagnose parametric model-plant mismatches for first-order plus time delay models. The plant-model ratio is developed
before being validated on the Wood-Berry distillation column. One of the prominent limitations
of frequency analysis, the filtering effects of time constant differences, is investigated and
quantified for the Wood-Berry distillation column, showing the effect of time constant differences
on each parametric model-plant mismatch diagnosis.
