Main

WMODEL - APMonitor Option

Main.OptionWmodel History

Show minor edits - Show changes to output

June 08, 2017, at 11:03 PM by 10.5.113.159 -
Changed line 13 from:
See also [[Main/OptionWmeas|WMEAS]],[[http://apmonitor.com/do/index.php/Main/EstimatorObjective|Estimation Objective Tuning]]
to:
See also [[Main/OptionWmeas|WMEAS]], [[http://apmonitor.com/do/index.php/Main/EstimatorObjective|Estimation Objective Tuning]]
June 08, 2017, at 11:03 PM by 10.5.113.159 -
Changed lines 5-6 from:
%width=50px%Attach:cv.png
to:
%width=50px%Attach:cv.png [[Main/DbsVariable|Local Options]] | %width=30px%Attach:apm.png [[Main/DbsGlobal|Global Options]]
Changed line 13 from:
See also [[http://apmonitor.com/do/index.php/Main/EstimatorObjective|Estimation Objective Tuning]]
to:
See also [[Main/OptionWmeas|WMEAS]],[[http://apmonitor.com/do/index.php/Main/EstimatorObjective|Estimation Objective Tuning]]
June 02, 2017, at 06:14 AM by 45.56.3.173 -
Added lines 12-13:

See also [[http://apmonitor.com/do/index.php/Main/EstimatorObjective|Estimation Objective Tuning]]
June 01, 2017, at 06:41 AM by 45.56.3.173 -
Added lines 1-11:
(:title WMODEL - APMonitor Option:)
(:keywords WMODEL, Optimization, Estimation, Option, Configure, Default, Description:)
(:description Objective function weight on model value:)

%width=50px%Attach:cv.png

 Type: Floating Point, Input
 Default Value: 2.0d0
 Description: Objective function weight on model value

A weighting factor to penalize deviation of current model predictions from prior model predictions. This is used in estimation applications (APM.IMODE=2, 5, or 8) where the penalty from a prior model prediction is a “forgetting factor” that approximates an infinite estimation horizon or favors prior predictions. The infinite estimation horizon approximation is especially useful for systems that have weakly observable or unobservable states. A higher WMODEL can also help to reduce the aggressiveness of the estimator in aligning with the measurements by balancing with a penalty against shifting too far from the prior predictions. The WMODEL value should never be equal to or larger than the WMEAS value for APM.EV_TYPE=1 (l1-norm). A WMODEL value higher than WMEAS will ignore measured values in favor of matching prior model predictions.