1 Lyapunov Direct Method
1.1 Definite
(1)Positive Definite (PD) V ( 0 ) = 0 ; V ( x ) > 0 , ∀ x ∈ D − { 0 } ; V(0)=0;V(x)>0,\forall x\in D- \{0\}; V(0)=0;V(x)>0,∀x∈D−{0};(2)Positive Semi Definite (PSD) V ( 0 ) = 0 ; V ( x ) ≥ 0 , ∀ x ∈ D − { 0 } ; V(0)=0;V(x)\geq 0,\forall x\in D- \{0\}; V(0)=0;V(x)≥0,∀x∈D−{0};(3)Negative Definite (ND) V ( 0 ) = 0 ; V ( x ) < 0 , ∀ x ∈ D − { 0 } ; V(0)=0;V(x)< 0,\forall x\in D- \{0\}; V(0)=0;V(x)<0,∀x∈D−{0};(4)Negative Semi Definite (NSD) V ( 0 ) = 0 ; V ( x ) ≤ 0 , ∀ x ∈ D − { 0 } ; V(0)=0;V(x)\leq 0,\forall x\in D- \{0\}; V(0)=0;V(x)≤0,∀x∈D−{0};
1.2 Lyapunov Direct Method Definition
系统状态空间方程
x
˙
=
f
(
x
)
\dot{x}=f(x)
x˙=f(x),其中,
x
=
0
x=0
x=0是平衡点。即
f
(
x
=
0
)
=
0
f(x=0)=0
f(x=0)=0或者说
x
˙
∣
x
=
0
=
0
\dot{x}|_{x=0}=0
x˙∣x=0=0。
假设存在一个函数
V
V
V:
(1)如果
V
V
V是PD,而
V
˙
\dot{V}
V˙是NSD,则平衡点
x
=
0
x=0
x=0是稳定点;
(2)如果
V
V
V是PD,而
V
˙
\dot{V}
V˙是ND,则平衡点
x
=
0
x=0
x=0是渐近稳定点。
1.3 Lie Derivative
系统状态空间方程: [ x 1 ˙ x 2 ˙ ] = [ a x 1 b x 2 + c o s x 1 ] = f ( x ) \begin{bmatrix} \dot{x_1} \\ \dot{x_2} \end{bmatrix}=\begin{bmatrix} ax_1 \\ bx_2+cosx_1 \end{bmatrix}=f(x) [x1˙x2˙]=[ax1bx2+cosx1]=f(x)假设Lyapunov Function: V = x 1 2 + x 2 2 V=x_1^{2}+x_2^2 V=x12+x22 V ˙ = x 1 x 1 ˙ + x 2 x 2 ˙ = [ 2 x 1 2 x 2 ] [ x 1 ˙ x 2 ˙ ] = [ ∂ V ∂ x 1 ∂ V ∂ x 2 ] [ f 1 f 2 ] = ▽ V f ( x ) = L f V ( x ) \dot{V}=x_1\dot{x_1}+x_2\dot{x_2}=\begin{bmatrix} 2x_1&2x_2 \end{bmatrix}\begin{bmatrix} \dot{x_1} \\ \dot{x_2} \end{bmatrix}=\begin{bmatrix} \frac{\partial V}{\partial x_1} & \frac{\partial V}{\partial x_2} \end{bmatrix}\begin{bmatrix} f_1 \\ f_2 \end{bmatrix}=\bigtriangledown Vf(x)=L_fV(x) V˙=x1x1˙+x2x2˙=[2x12x2][x1˙x2˙]=[∂x1∂V∂x2∂V][f1f2]=▽Vf(x)=LfV(x) L f V ( x ) L_fV(x) LfV(x)即为Lie Derivative。
2 非线性系统稳定性分析——以单摆系统(Pendulum)为例
2.1 Pendulum without Friction
X轴受力分析,由牛顿第二定律:
∑
F
x
=
m
a
x
\sum F_x=ma_x
∑Fx=max即
−
m
g
s
i
n
θ
=
m
L
θ
¨
-mgsin\theta=mL\ddot{\theta}
−mgsinθ=mLθ¨系统动态方程为
θ
¨
+
g
L
s
i
n
θ
=
0
\ddot{\theta}+\frac{g}{L}sin\theta=0
θ¨+Lgsinθ=0设
x
1
=
θ
x_1=\theta
x1=θ,
x
2
=
θ
˙
=
x
1
˙
x_2=\dot{\theta}=\dot{x_1}
x2=θ˙=x1˙,有
{
x
1
˙
=
x
2
x
2
˙
=
θ
¨
=
−
g
L
s
i
n
x
1
\begin{cases} \dot{x_1}=x_2 \\ \dot{x_2}=\ddot{\theta}=-\frac{g}{L}sinx_1 \end{cases}
{x1˙=x2x2˙=θ¨=−Lgsinx1从系统能量的角度分析,
E
n
e
r
g
y
=
K
i
n
e
t
i
c
+
P
o
t
e
n
t
i
a
l
Energy=Kinetic+Potential
Energy=Kinetic+Potential,因此,假设Lyapunov Function:
V
=
K
+
P
=
1
2
m
v
2
+
m
g
h
=
1
2
m
L
2
x
2
2
+
m
g
L
(
1
−
c
o
s
x
1
)
V=K+P=\frac{1}{2}mv^2+mgh=\frac{1}{2}mL^2x_2^2+mgL(1-cosx_1)
V=K+P=21mv2+mgh=21mL2x22+mgL(1−cosx1)李雅普诺夫函数随时间变化:
V
˙
=
L
f
V
=
[
∂
V
∂
x
1
∂
V
∂
x
2
]
[
f
1
f
2
]
=
m
g
L
x
2
s
i
n
x
1
−
m
g
L
x
2
s
i
n
x
1
=
0
\dot{V}=L_fV=\begin{bmatrix} \frac{\partial V}{\partial x_1} & \frac{\partial V}{\partial x_2} \end{bmatrix}\begin{bmatrix} f_1 \\ f_2 \end{bmatrix}=mgLx_2sinx_1-mgLx_2sinx_1=0
V˙=LfV=[∂x1∂V∂x2∂V][f1f2]=mgLx2sinx1−mgLx2sinx1=0由于(1)
V
V
V是PD的;(2)
V
˙
\dot{V}
V˙是NSD的;故系统是Stability的。
从物理的角度解释:
V
V
V代表能量,
V
˙
\dot{V}
V˙代表能量随时间变化,因为无摩擦,所以
V
˙
=
0
\dot{V}=0
V˙=0即能量始终保持不变,故系统是Stability,与数学推导一致。
2.2 Pendulum with Friction
X轴受力分析,由牛顿第二定律:
∑
F
x
=
m
a
x
\sum F_x=ma_x
∑Fx=max即
−
m
g
s
i
n
θ
−
k
L
θ
˙
=
m
L
θ
¨
-mgsin\theta-kL\dot{\theta}=mL\ddot{\theta}
−mgsinθ−kLθ˙=mLθ¨系统动态方程为
θ
¨
+
g
L
s
i
n
θ
+
k
m
θ
˙
=
0
\ddot{\theta}+\frac{g}{L}sin\theta+\frac{k}{m}\dot{\theta}=0
θ¨+Lgsinθ+mkθ˙=0设
x
1
=
θ
x_1=\theta
x1=θ,
x
2
=
θ
˙
=
x
1
˙
x_2=\dot{\theta}=\dot{x_1}
x2=θ˙=x1˙,有
{
x
1
˙
=
x
2
x
2
˙
=
θ
¨
=
−
g
L
s
i
n
x
1
−
k
m
x
2
\begin{cases} \dot{x_1}=x_2 \\ \dot{x_2}=\ddot{\theta}=-\frac{g}{L}sinx_1-\frac{k}{m}x_2 \end{cases}
{x1˙=x2x2˙=θ¨=−Lgsinx1−mkx2从系统能量的角度分析,
E
n
e
r
g
y
=
K
i
n
e
t
i
c
+
P
o
t
e
n
t
i
a
l
Energy=Kinetic+Potential
Energy=Kinetic+Potential,因此,假设Lyapunov Function:
V
=
K
+
P
=
1
2
m
v
2
+
m
g
h
=
1
2
m
L
2
x
2
2
+
m
g
L
(
1
−
c
o
s
x
1
)
V=K+P=\frac{1}{2}mv^2+mgh=\frac{1}{2}mL^2x_2^2+mgL(1-cosx_1)
V=K+P=21mv2+mgh=21mL2x22+mgL(1−cosx1)李雅普诺夫函数随时间变化:
V
˙
=
L
f
V
=
[
∂
V
∂
x
1
∂
V
∂
x
2
]
[
f
1
f
2
]
=
−
k
L
2
x
2
2
\dot{V}=L_fV=\begin{bmatrix} \frac{\partial V}{\partial x_1} & \frac{\partial V}{\partial x_2} \end{bmatrix}\begin{bmatrix} f_1 \\ f_2 \end{bmatrix}=-kL^2x_2^2
V˙=LfV=[∂x1∂V∂x2∂V][f1f2]=−kL2x22由于(1)
V
V
V是PD的;(2)对于
V
˙
\dot{V}
V˙,
∀
x
=
[
x
1
0
]
\forall x=\begin{bmatrix} x_1 & 0 \end{bmatrix}
∀x=[x10]均有
V
˙
=
0
\dot{V}=0
V˙=0,因此
V
˙
\dot{V}
V˙是NSD的。故系统是Stability。
从物理的角度解释:
V
V
V代表能量,
V
˙
\dot{V}
V˙代表能量随时间变化,有摩擦,系统应为渐近稳定的,但数学推导结果表明,系统是Stability,矛盾。故引入The Invariance Principle——Lasalle’s Theorem扩展李雅普诺夫第二方法。
3 Lasalle’s Theorem
引入一个system(1):
x
˙
=
f
(
x
)
\dot{x}=f(x)
x˙=f(x)Theorem:The equilibrium point
x
=
0
x=0
x=0 of the autonomous system(1) is asymptotically stable if there exists a function
V
(
x
)
V(x)
V(x) satisfying.
(1)
V
(
x
)
V(x)
V(x) positive definite
∀
x
∈
D
\forall x \in D
∀x∈D, where we assume that
0
∈
D
0\in D
0∈D;
(2)
V
˙
(
x
)
\dot{V}(x)
V˙(x) is negative semi definite in a bounded region,
R
∈
D
R\in D
R∈D;
(3)
V
˙
(
x
)
\dot{V}(x)
V˙(x) does not vanish identically along any trajectory in
R
R
R, other than the null solution
x
=
0
x=0
x=0.
其中,把区域内所有使得
V
˙
(
x
)
=
0
\dot{V}(x)=0
V˙(x)=0的点的集合叫做
R
R
R。