Using the Final Value Theorem, we can try to design a controller that meets some steady-state specs. However, it’s still difficult to design a controller that satisfies desired transient specs. Some examples of other control design methods:
Limitations of these methods:
- Requires lots of manual tuning of parameters, often based on trial and error to satisfy transient specs.
- May not be feasible to satisfy desired transient plants
With these in mind, we come up with a new method to design controllers. The idea is to:
- Design closed-loop transfer functions to obtain closed-loop stability
- We are designing the closed-loop transfer functions directly instead of designing the controller
- Recover a controller that results in those closed-loop transfer functions.
To do so, we will use a clever change of variables called input-output parameterization (IOP).
Define the IOP equations (feasibility constraints) for variables :
Since the plant is known (constant/not variable), these equations are linear in the variables .
Recall that for closed-loop stability, we just need to check:
Theorem and Proof
IOP Theorem
- a. If results in closed-loop stability, then , , and , satisfy the IOP equations (i)-(iii).
- b. If satisfy the IOP equations (i)-(iii), and if we choose our controller , then , , , and .
Proof.
Part (a):
- Given: results in closed-loop stability
- WTS: satisfy the IOP equations
Proof of part (a):
- results in closed-loop stability
- are BIBO stable [definition of closed-loop stability]
- are stable [theorem from class]
- satisfy equation (iii)
- We have:
which satisfies (i)
- We have which satisfies (ii)
Part (b):
- Given: satisfy the IOP equations (i)-(iii)
- WTS:
Proof of part (b):
- satisfy the IOP equations (i)-(iii)
-
- Because , from equation (i) re-arranged
-
- from equation (iii) re-arranged