|
QuickStudy
- How it works
When
you add QuickStudy to your existing process control system, you will
see a rapid improvement. Your initial input is minimal---simply to identify
the relevant process inputs and outputs, as well as the product parameters
to be modeled and controlled. Modeling is done on-line using real-time
process and product data. Tuning of the control strategy is automatic,
and the response curve can be shaped by parameters you select.
One QuickStudy controller contains up to 16 Predictive Controller Set-up
(PCS) blocks. Each PCS block is self-contained, including modeling,
prediction, control and output generation. One of these blocks is required
for each controlled variable and can accept 16 input variables. Each
PCS block can execute at the frequency appropriate to its variable.
Model
Quickly Takes On Actual Process Dynamics
QuickStudy's modeling algorithm
determines the coefficients of the process models using probability
density functions. After a few sampling periods, the model begins to
represent the actual process dynamics. As the sampling continues, the
model becomes more accurate. Typically, 250 to 1,000 sampling periods
are required for a new model to become accurate enough to assume control---far
less than other systems which often require thousands of samples to
build an accurate model. Adaptation of an existing model is normally
accomplished in 50 to 100 samples with QuickStudy.
As inputs are scanned, the actual value of all process variables are
compared to their predicted values. When a variable doesn't match the
predicted behavior, this information is added to the statistical database.
The model's parameters are able to adapt to the changes in process conditions
and the controller acts, almost instantaneously, to assure that the
process remains at setpoint.
The control function is accomplished in two steps. First, the quadratic
optimization algorithm determines the required control action based
on predicted values of the process variables. Second, the dynamic output
generation algorithm computes the trajectory for optimal approach to
setpoint. These calculations are updated every cycle.
Automically
Triggers Proper Interaction of Variables
In many processes, action
taken to control one variable
will affect other variables. QuickStudy incorporates these interactions
into its models so that a change in one control action automatically
triggers the appropriate changes
in others.
QuickStudy can handle the most complex process dynamics. Its modeling
algorithm can handle high order models and can deal effectively with
long dead-times, integrating characteristics and inverse responses.
And the system's configuration allows the user to design the response
curve to meet the unique needs of each process. So response time is
adjusted automatically to meet the process's tolerance of overshoots
in reaching the setpoint. In addition, QuickStudy is exceptionally stable.
|