APM.EV_TYPE - APMonitor Option

 Type: Integer, Input
 Default Value: 1
 Description: Estimated variable error model type
              1=linear error, abs(model-measured) with slacks
              2=squared error, (model-measured)^2
              3=approximate linear without slack variables

EV_TYPE applies a specific objective function. Linear is an l1-norm, or in other words the solver minimizes the sum of the absolute value of the difference between the CV and the set point. Squared is an l2-norm or sum squared error (SSE), or in other words the solver minimizes the sum of the squared difference between the CV and the set point. l1-norm can be useful when noise or measurement error is expected because it better rejects those. Option 3 is not typically used as an approximate absolute value function that uses a nonlinear function instead of slack variables.

See also CV_TYPE, Estimation Objective Tuning

💬