Model Confidence Band block

The Model confidence band block calculates model alarming limits based on the accuracy of the trained model.

Return to Overview of blocks

Model Confidence Band block

Description

The Model Confidence Band block calculates model alarming limits based on the accuracy of the model.  The design of this block makes it possible to connect the output port to the "AlarmLimits" port and the "ModelOutput" port of the model alarming block.

diagram showing the Model confidence band block

Block Type

Rules and Model block

Input/Output ports

The input port of the Model confidence band block must contain an output field from a Nonlinear model block

The output port will contain the Setpoint, UpperLimit and LowerLimit fields

Functions performed on tags

  • On the values - The value of the selected model output field is copied to the output field of the Model confidence band block and renamed SetPoint (SP).  Setpoint and Model Output (the same field) are also used in the upper and lower limits calculations.  The UpperLimit and LowerLimit are  calculated using the two equations shown below. As stated above Setpoint = Model Output.  The model output is, obviously, obtained from the model for which these limits are being calculated.  The model accuracy weight (a value between 0 and 1) is specified during the configuration of the Model confidence band block and allows the user to weight the contribution of the model accuracy makes to the calculation.  The overall model accuracy parameter is obtained from the Nonlinear model block.

and

    Note : The Model confidence band block MUST be situated in a superblock containing a model block (to obtain the model accuracy) and for the calculation to have any meaning the input to the Model confidence band block must be the output from the same model block.

  • On the timestamp - The execute time is stamped on the upper and lower limit outputs.  The set point will have the time stamp of the configured Model Output field.

  • On the quality - The output fields have the quality of the selected Model Output field.

Return to top

Example

The purpose of the Model confidence band block is to determine limit values for a model output.  It does this by calculating a band around a the model output that makes use of both the model accuracy and a user-defined model accuracy weight.  In the diagram below; the green line is the upper limit, the red line is the model output and the blue line is the lower limit.

 

diagram showing the output fields of the Model confidence band block

The output from the Model confidence band block is typically used by the model alarming block as external setpoint limits.  >> More on external limits in the Model alarming block


Related topics:

  

CSense 2023- Last updated: June 24,2025