MECHENG 423 Formula Sheet: Parameter Estimation - Pseudo-Inverse

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 3

MECHENG 423 Formula Sheet

Parameter Estimation – Pseudo-inverse


X w= y w=( X T X )−1 X T y

Eigenvalues

det ( λI − A )=0
Eigenvectors
A x=λ x

Diagonalisation

P−1 AP= Λ
P contains eigenvectors, Λ contains eigenvalues as its diagonal elements

Matrix Exponential – Modal method (if A is diagonlisable)

A=P Λ P−1
e At=P e Λ t P−1

State Space Models


ẋ N ×1= A N × N x N ×1 + BN × M u M ×1
y P ×1=C P × N x N ×1 + D P × M u M ×1

OR

ẋ=f (x ,u)
y=h(x ,u)

Linearisation

0=f (x¿ ,u ¿ )
y ¿ =h( x ¿ , u¿ )

∂f
A=
∂x ¿ |
∂f
B=
∂u ¿ |
∂h
C=
∂x ¿ |
∂h
d=
∂u |
¿

Properties of State Space Models


State Space to Transfer Function

Y (s )
T ( s) = =C ( sI− A )−1 B+ D
U (s)

Time solution
t
x ( t )=e x 0 +∫ e A (t −τ ) Bu ( τ ) dτ
At

Φ=e At state transition matrix

Eigenvalues of matrix A are the poles of the system.

Discrete-Time State-Space Models


x k+1= A d x k + B d uk
y k =C x k + D u k

Ad =e At
T

( )
Bd = ∫ e Aμ dμ B=( e AT −I ) A−1 B
0
Transfer function in Z domain

T ( z )=C ( zI− Ad )−1 B d + D

Eigenvalues of Ad are the poles of T ( z), to be stable, they must be within the unit circle.

Full State Feedback


State regulation

u ( t )=−Kx( t)

ẋ=( A−BK ) x
Pole placements

det ( λ I −( A−BK ) )=( λ− λ1 ) ( λ−λ2 ) …


Controllability

M C =[B AB A 2 B … A N −1 B ]
If M C is full rank, then (A,B) is controllable.

You might also like