Unmeasured output disturbance model


outdist = getoutdist(MPCobj)
[outdist,channels] = getoutdist(MPCobj)


outdist = getoutdist(MPCobj) retrieves the model of output disturbances used by the Model Predictive Controller object MPCobj. The output model, outdist, is a discrete-time, delay-free, LTI state-space (ss) object.

This command is most useful when a default output disturbance model is being used and you want to review its structure. This model, in combination with the intput disturbance model (if any), governs how well the controller compensates for unmeasured disturbances and prediction errors. You can override the controller's default behavior using the setoutdist command. For details on the role of disturbance modeling in Model Predictive Control, see Controller State Estimation.

If the controller employs an output disturbance model, it must have ny outputs, where ny is the number of plant outputs defined in MPCobj.Model.Plant. All input signals to outdist are unit-variance white noise. If you specify the output disturbance model, the number of inputs is your choice. For the default, the number of inputs depends upon the number of measured outputs, the input disturbance model (if any), and the need to maintain observability of the controller's state variables. For more details, see "State Estimation".

For details about the overall model used in the MPC algorithm for state estimation purposes, see Controller State Estimation.

[outdist,channels] = getoutdist(MPCobj) also returns the output channels to which integrated white noise has been added by default as an output disturbance model. If you have specified a custom output disturbance model using setoutdist, channels is empty.

Was this topic helpful?