Scribd
Upload
37 views

EPE3302 Lecture 9 Linear Quadratic Regulators and Robust Controllers

This document discusses quadratic optimal regulators and the algebraic Riccati equation. It provides two key steps for optimal control design: [1] Solve the algebraic Riccati equation to determine the matrix P, which must be positive definite. [2] Substitute P into K = R^-1B^T P to determine the optimal feedback gain matrix K. The document also briefly mentions that robust control theory accounts for modeling errors and designs control systems that maintain stability and performance in the presence of uncertainties.

Uploaded by

Bryan Kek
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
37 views

EPE3302 Lecture 9 Linear Quadratic Regulators and Robust Controllers

This document discusses quadratic optimal regulators and the algebraic Riccati equation. It provides two key steps for optimal control design: [1] Solve the algebraic Riccati equation to determine the matrix P, which must be positive definite. [2] Substitute P into K = R^-1B^T P to determine the optimal feedback gain matrix K. The document also briefly mentions that robust control theory accounts for modeling errors and designs control systems that maintain stability and performance in the presence of uncertainties.

Uploaded by

Bryan Kek
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

EPE3302

State Space Analysis and Control


in Power Engineering
Dr. Khalid ABIDI
Objectives
The objective of this lecture are as follows:
• Quadratic Optimal Regulators
• Algebraic Riccati Equation
Quadratic Optimal Regulator
An advantage of the quadratic optimal control method over the pole-
placement method is that the former provides a systematic way of
computing the state feedback control gain matrix.
Consider the State Space system given as
𝐱𝐱̇ 𝑡𝑡 = 𝐴𝐴𝐱𝐱 𝑡𝑡 + 𝐵𝐵𝐮𝐮 𝑡𝑡
The Quadratic Optimal Control problem is to find a matrix 𝐾𝐾 such that the
control law
𝐮𝐮 = −𝐾𝐾𝐱𝐱
minimizes the Performance Index
∞
𝐽𝐽 = � 𝐱𝐱 ⊺ 𝑄𝑄𝐱𝐱 + 𝐮𝐮⊺ 𝑅𝑅𝐮𝐮 𝑑𝑑𝑑𝑑
0
Quadratic Optimal Regulator
The matrix 𝑄𝑄 is a 𝑛𝑛 × 𝑛𝑛 positive-definite or positive semi-definite matrix
and the matrix 𝑅𝑅 is a 𝑚𝑚 × 𝑚𝑚 positive-definite matrix. The matrices 𝑄𝑄 and 𝑅𝑅
determine the relative importance of the error and the amount of effort
(energy) spent during the process. In other words if 𝑄𝑄 is relatively larger
than 𝑅𝑅, it means that the primary goal is to reduce the error. On the other
hand, if 𝑅𝑅 is larger than 𝑄𝑄 then it means that minimizing the control effort
is of primary concern.
Definition: A positive definite matrix is a matrix with all the eigenvalues
being positive. A positive semi-definite matrix has at least one zero
eigenvalue and the remaining being positive.
Quadratic Optimal Regulator
To solve the Quadratic Optimal Regulator we substitute the control law
𝐮𝐮 = −𝐾𝐾𝐱𝐱
in the Performance Index
∞
𝐽𝐽 = � 𝐱𝐱 ⊺ 𝑄𝑄𝐱𝐱 + 𝐮𝐮⊺ 𝑅𝑅𝐮𝐮 𝑑𝑑𝑑𝑑
0
∞ ∞
=� 𝐱𝐱 ⊺ 𝑄𝑄𝐱𝐱 + −𝐾𝐾𝐱𝐱 ⊺ 𝑅𝑅 −𝐾𝐾𝐱𝐱 𝑑𝑑𝑑𝑑 = � 𝐱𝐱 ⊺ 𝑄𝑄 + 𝐾𝐾 ⊺ 𝑅𝑅𝑅𝑅 𝐱𝐱𝑑𝑑𝑑𝑑
0 0
Set
⊺ ⊺
𝑑𝑑 ⊺
𝐱𝐱 𝑄𝑄 + 𝐾𝐾 𝑅𝑅𝑅𝑅 𝐱𝐱 = − 𝐱𝐱 𝑃𝑃𝐱𝐱
𝑑𝑑𝑑𝑑
where 𝑃𝑃 is a 𝑛𝑛 × 𝑛𝑛 symmetric positive-definite matrix.
Quadratic Optimal Regulator
Consider the Closed-Loop system given as
𝐱𝐱̇ 𝑡𝑡 = 𝐴𝐴 − 𝐵𝐵𝐵𝐵 𝐱𝐱 𝑡𝑡
such that
⊺ ⊺
𝑑𝑑 ⊺
𝐱𝐱 𝑄𝑄 + 𝐾𝐾 𝑅𝑅𝑅𝑅 𝐱𝐱 = 𝐱𝐱 𝑃𝑃𝐱𝐱 = −𝐱𝐱̇ ⊺ 𝑃𝑃𝐱𝐱 − 𝐱𝐱 ⊺ 𝑃𝑃𝐱𝐱̇
𝑑𝑑𝑑𝑑
= −𝐱𝐱 ⊺ 𝐴𝐴 − 𝐵𝐵𝐵𝐵 ⊺ 𝑃𝑃𝐱𝐱 − 𝐱𝐱 ⊺ 𝑃𝑃 𝐴𝐴 − 𝐵𝐵𝐵𝐵 𝐱𝐱 𝑡𝑡
= −𝐱𝐱 ⊺ 𝐴𝐴 − 𝐵𝐵𝐵𝐵 ⊺ 𝑃𝑃 + 𝑃𝑃 𝐴𝐴 − 𝐵𝐵𝐵𝐵 𝐱𝐱 𝑡𝑡
From this equation we can obtain the relation
𝐴𝐴 − 𝐵𝐵𝐵𝐵 ⊺ 𝑃𝑃 + 𝑃𝑃 𝐴𝐴 − 𝐵𝐵𝐵𝐵 = −𝑄𝑄 − 𝐾𝐾 ⊺ 𝑅𝑅𝑅𝑅 = 𝐴𝐴⊺ 𝑃𝑃 − 𝐾𝐾 ⊺ 𝐵𝐵⊺ 𝑃𝑃 + 𝑃𝑃𝑃𝑃 − 𝑃𝑃𝑃𝑃𝑃𝑃
Quadratic Optimal Regulator
Let 𝐾𝐾 = 𝑅𝑅 −1 𝐵𝐵⊺ 𝑃𝑃 then
−𝑄𝑄 − 𝐾𝐾 ⊺ 𝑅𝑅𝑅𝑅 = 𝐴𝐴⊺ 𝑃𝑃 − 𝐾𝐾 ⊺ 𝐵𝐵⊺ 𝑃𝑃 + 𝑃𝑃𝑃𝑃 − 𝑃𝑃𝑃𝑃𝑃𝑃
such that we obtain
𝐴𝐴⊺ 𝑃𝑃 + 𝑃𝑃𝑃𝑃 − 𝑃𝑃𝑃𝑃𝑅𝑅 −1 𝐵𝐵⊺ 𝑃𝑃 + 𝑄𝑄 = 0
This equation is called the Algebraic Riccati Equation. By specifying the
matrices 𝑄𝑄 and 𝑅𝑅 we can obtain the matrix 𝑃𝑃 that will guarantee that the
feedback gain 𝐾𝐾 = 𝑅𝑅 −1 𝐵𝐵⊺ 𝑃𝑃 will lead to an optimum performance.
Quadratic Optimal Regulator
The Optimal control design problem can be summarized in two steps:
1. Solve the Algebraic Riccati Equation for the matrix 𝑃𝑃. The computed
matrix 𝑃𝑃 must be positive-definite.
2. Substitute 𝑃𝑃 in 𝐾𝐾 = 𝑅𝑅 −1 𝐵𝐵⊺ 𝑃𝑃 which will lead to an optimum
performance
Robust Control
Robust control theory uses the assumption that the models we use in
designing control systems have modeling errors. We shall present an
introduction to this theory in this section. Basically, the theory assumes that
there is an uncertainty or error between the actual plant and its
mathematical model and includes such uncertainty or error in the design
process of the control system.
Robust Control
Systems designed based on the robust control theory will possess the
following properties:
1. Robust stability. The control system designed is stable in the presence of
perturbation.
2. Robust performance. The control system exhibits predetermined
response characteristics in the presence of perturbation.

You might also like