Explicit model predictive controller
Model Predictive Control Toolbox
The Explicit MPC Controller block uses the following input signals:
Either measured plant outputs (mo
) or custom state estimate
(x[k|k]
)
Reference or setpoint (ref
)
Measured plant disturbance (md
), if any
The Explicit MPC Controller block uses a lookup table to store the precalculated piecewise-affine optimal control law instead of solving a quadratic programming optimization problem at runtime at each control interval like the MPC Controller block. Given the same MPC problem, the two blocks return the same solution. The Explicit MPC Controller block requires less online computational effort, which is useful for applications that need a short control interval. It has, however, a heavier offline computational effort and a larger memory footprint. Indeed, the combinatorial nature of explicit MPC restricts its usage to applications with relatively few inputs, outputs, and state variables, a short prediction horizon, and few output constraints.
The Explicit MPC Controller supports only a subset of optional MPC features, as outlined in the following table.
Supported Features | Unsupported Features |
---|---|
|
|