[滑模控制] (5) 基于分层滑模的吊车控制
[滑模控制] (5) 基于分层滑模的吊车控制
本博客需要一些现代控制理论中Lyapunov稳定性的一些理论知识。
您需要对滑模控制有一个初步的了解,可以参考:
[滑模控制] (1) 二阶系统的简单滑模控制器设计
1 前言
分层滑模主要适用的是欠驱动系统,即输入个数小于需要控制的状态数目,比如下面为例吊车,只有一个控制输入,但是需要同时控制吊车在轨道的位置和下摆的角度。笔者对于分层滑模的了解也不够深入,属于初学阶段,不过恰巧曾经简单学习过分层滑模且应用在了吊车上,因此在此将展示本人的控制器设计。
笔者这里学习的主要是王伟的一些中文文献,他在国内做欠驱动系统控制研究算是很早的。
2 吊车动力学模型
如上图所示一个桥式吊车,假设负载可以看做质点,在X-Y平面内运动,同事忽略台车与轨道之间的摩擦力以及绳子长度L的变化,则利用Euler-Lagrange方法可以得到其动力学模型(其实是抄后面参考文献的):
{
(
m
+
M
)
x
¨
+
m
L
(
θ
¨
cos
θ
−
θ
˙
2
sin
θ
)
=
u
x
¨
cos
θ
+
L
θ
¨
+
g
sin
θ
=
0
\left\{ \begin{matrix} \left( m+M \right)\ddot{x}+mL\left( \ddot{\theta }\cos \theta -{{{\dot{\theta }}}^{2}}\sin \theta \right)=u \\ \ddot{x}\cos \theta +L\ddot{\theta }+g\sin \theta =0 \\ \end{matrix} \right.
{(m+M)x¨+mL(θ¨cosθ−θ˙2sinθ)=ux¨cosθ+Lθ¨+gsinθ=0其中
M
M
M表示台车质量,
m
m
m表示负载质量,
x
x
x表示台车水平位移,
θ
\theta
θ表示吊绳摆角,
L
L
L为绳长。
令
x
1
=
x
{{x}_{1}}=x
x1=x,
x
2
=
x
˙
{{x}_{2}}=\dot{x}
x2=x˙,
x
3
=
θ
{{x}_{3}}=\theta
x3=θ,
x
4
=
θ
˙
{{x}_{4}}=\dot{\theta }
x4=θ˙分别为吊车系统中的台车水平移动距离,台车水平速度,吊绳摆角,吊绳摆角角速度,考虑外部干扰
d
1
(
t
)
{{d}_{1}}\left( t \right)
d1(t)、
d
2
(
t
)
{{d}_{2}}\left( t \right)
d2(t),即有:
{
x
˙
1
=
x
2
x
˙
2
=
g
1
(
X
)
+
b
1
(
X
)
u
+
d
1
(
t
)
x
˙
3
=
x
4
x
˙
4
=
g
2
(
X
)
+
b
2
(
X
)
u
+
d
2
(
t
)
\left\{ \begin{matrix} {{{\dot{x}}}_{1}}={{x}_{2}} \\ {{{\dot{x}}}_{2}}={{g}_{1}}\left( X \right)+{{b}_{1}}\left( X \right)u+{{d}_{1}}\left( t \right) \\ {{{\dot{x}}}_{3}}={{x}_{4}} \\ {{{\dot{x}}}_{4}}={{g}_{2}}\left( X \right)+{{b}_{2}}\left( X \right)u+{{d}_{2}}\left( t \right) \\ \end{matrix} \right.
⎩
⎨
⎧x˙1=x2x˙2=g1(X)+b1(X)u+d1(t)x˙3=x4x˙4=g2(X)+b2(X)u+d2(t)其中:
{
g
1
(
X
)
=
m
L
θ
˙
2
sin
θ
+
m
g
sin
θ
cos
θ
M
+
m
sin
2
θ
b
1
(
X
)
=
1
M
+
m
sin
2
θ
g
2
(
X
)
=
−
(
m
+
M
)
g
sin
θ
+
m
L
θ
˙
2
sin
θ
cos
θ
(
M
+
m
sin
2
θ
)
L
b
2
(
X
)
=
−
cos
θ
(
M
+
m
sin
2
θ
)
L
\left\{ \begin{matrix} {{g}_{1}}\left( X \right)=\frac{mL{{{\dot{\theta }}}^{2}}\sin \theta +mg\sin \theta \cos \theta }{M+m{{\sin }^{2}}\theta } \\ {{b}_{1}}\left( X \right)=\frac{1}{M+m{{\sin }^{2}}\theta } \\ {{g}_{2}}\left( X \right)=-\frac{\left( m+M \right)g\sin \theta +mL{{{\dot{\theta }}}^{2}}\sin \theta \cos \theta }{\left( M+m{{\sin }^{2}}\theta \right)L} \\ {{b}_{2}}\left( X \right)=-\frac{\cos \theta }{\left( M+m{{\sin }^{2}}\theta \right)L} \\ \end{matrix} \right.
⎩
⎨
⎧g1(X)=M+msin2θmLθ˙2sinθ+mgsinθcosθb1(X)=M+msin2θ1g2(X)=−(M+msin2θ)L(m+M)gsinθ+mLθ˙2sinθcosθb2(X)=−(M+msin2θ)Lcosθ
3 分层滑模控制器设计
易知
b
1
>
0
{{b}_{1}}>0
b1>0,设有
∣
d
1
(
t
)
∣
≤
σ
1
\left| {{d}_{1}}\left( t \right) \right|\le {{\sigma }_{1}}
∣d1(t)∣≤σ1、
∣
d
2
(
t
)
∣
≤
σ
2
\left| {{d}_{2}}\left( t \right) \right|\le {{\sigma }_{2}}
∣d2(t)∣≤σ2,
0
<
σ
3
≤
∣
b
1
∣
≤
σ
4
0<{{\sigma }_{3}}\le \left| {{b}_{1}} \right|\le {{\sigma }_{4}}
0<σ3≤∣b1∣≤σ4,
∣
b
2
∣
≤
σ
5
\left| {{b}_{2}} \right|\le {{\sigma }_{5}}
∣b2∣≤σ5 。
该四阶系统可以看做是2个二阶子系统耦合,首先考虑第一层滑模面:
{
s
1
=
c
1
x
1
+
x
2
c
1
>
0
s
2
=
c
2
x
3
+
x
4
c
2
>
0
\left\{ \begin{matrix} \begin{matrix} {{s}_{1}}={{c}_{1}}{{x}_{1}}+{{x}_{2}} & {{c}_{1}}>0 \\ \end{matrix} \\ \begin{matrix} {{s}_{2}}={{c}_{2}}{{x}_{3}}+{{x}_{4}} & {{c}_{2}}>0 \\ \end{matrix} \\ \end{matrix} \right.
{s1=c1x1+x2c1>0s2=c2x3+x4c2>0易得:
{
s
˙
1
=
c
1
x
2
+
g
1
(
X
)
+
b
1
(
X
)
u
+
d
1
s
˙
2
=
c
2
x
4
+
g
2
(
X
)
+
b
2
(
X
)
u
+
d
2
\left\{ \begin{matrix} {{{\dot{s}}}_{1}}={{c}_{1}}{{x}_{2}}+{{g}_{1}}\left( X \right)+{{b}_{1}}\left( X \right)u+{{d}_{1}} \\ {{{\dot{s}}}_{2}}={{c}_{2}}{{x}_{4}}+{{g}_{2}}\left( X \right)+{{b}_{2}}\left( X \right)u+{{d}_{2}} \\ \end{matrix} \right.
{s˙1=c1x2+g1(X)+b1(X)u+d1s˙2=c2x4+g2(X)+b2(X)u+d2在不考虑干扰的情况下,等效控制律为:
{
u
e
q
1
=
c
1
x
2
+
g
1
(
X
)
b
1
(
X
)
u
e
q
2
=
c
2
x
4
+
g
2
(
X
)
b
2
(
X
)
\left\{ \begin{matrix} {{u}_{eq1}}=\frac{{{c}_{1}}{{x}_{2}}+{{g}_{1}}\left( X \right)}{{{b}_{1}}\left( X \right)} \\ {{u}_{eq2}}=\frac{{{c}_{2}}{{x}_{4}}+{{g}_{2}}\left( X \right)}{{{b}_{2}}\left( X \right)} \\ \end{matrix} \right.
{ueq1=b1(X)c1x2+g1(X)ueq2=b2(X)c2x4+g2(X)设计第二层滑模为:
s
=
α
s
1
+
β
s
2
s=\alpha {{s}_{1}}+\beta {{s}_{2}}
s=αs1+βs2整个系统只能有一个控制律,因此要考虑2个子系统的等效控制,总控制量为:
u
=
u
e
q
1
+
u
e
q
2
+
u
s
w
u={{u}_{eq1}}+{{u}_{eq2}}+{{u}_{sw}}
u=ueq1+ueq2+usw
u
s
w
{{u}_{sw}}
usw为切换控制律。
考虑Lyapunov函数:
V
=
1
2
s
2
V=\frac{1}{2}{{s}^{2}}
V=21s2求导:
V
˙
=
s
s
˙
=
s
(
α
s
˙
1
+
β
s
˙
2
)
=
s
(
α
(
c
1
x
2
+
g
1
+
b
1
u
+
d
1
)
+
β
(
c
2
x
4
+
g
2
+
b
2
u
+
d
2
)
)
\begin{aligned} & \dot{V}=s\dot{s}=s\left( \alpha {{{\dot{s}}}_{1}}+\beta {{{\dot{s}}}_{2}} \right) \\ & \begin{matrix} {} \\ \end{matrix}=s\left( \alpha \left( {{c}_{1}}{{x}_{2}}+{{g}_{1}}+{{b}_{1}}u+{{d}_{1}} \right)+\beta \left( {{c}_{2}}{{x}_{4}}+{{g}_{2}}+{{b}_{2}}u+{{d}_{2}} \right) \right) \\ \end{aligned}
V˙=ss˙=s(αs˙1+βs˙2)=s(α(c1x2+g1+b1u+d1)+β(c2x4+g2+b2u+d2))带入
u
u
u得:
V
˙
=
s
(
α
b
1
u
e
q
2
+
β
b
2
u
e
q
1
+
α
d
1
+
β
d
2
+
(
α
b
1
+
β
b
2
)
u
s
w
)
\dot{V}=s\left( \alpha {{b}_{1}}{{u}_{eq2}}+\beta {{b}_{2}}{{u}_{eq1}}+\alpha {{d}_{1}}+\beta {{d}_{2}}+\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right){{u}_{sw}} \right)
V˙=s(αb1ueq2+βb2ueq1+αd1+βd2+(αb1+βb2)usw)设计切换控制律:
u
s
w
=
−
α
b
1
u
e
q
2
−
β
b
2
u
e
q
1
−
ε
s
g
n
(
s
1
)
α
b
1
+
β
b
2
−
k
s
k
>
0
\begin{matrix} {{u}_{sw}}=\frac{-\alpha {{b}_{1}}{{u}_{eq2}}-\beta {{b}_{2}}{{u}_{eq1}}-\varepsilon sgn \left( {{s}_{1}} \right)}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-ks & k>0 \\ \end{matrix}
usw=αb1+βb2−αb1ueq2−βb2ueq1−εsgn(s1)−ksk>0
将上式带入,得:
V
˙
=
s
(
−
ε
s
g
n
(
s
1
)
−
(
α
b
1
+
β
b
2
)
k
s
+
α
d
1
+
β
d
2
)
=
α
s
1
(
α
d
1
+
β
d
2
)
−
α
ε
∣
s
1
∣
−
β
ε
s
2
s
g
n
(
s
1
)
+
β
s
2
(
α
d
1
+
β
d
2
)
−
(
α
b
1
+
β
b
2
)
k
s
2
≤
α
∣
s
1
∣
(
∣
α
d
1
+
β
d
2
∣
−
ε
)
−
s
2
β
(
ε
s
g
n
(
s
1
)
−
(
α
d
1
+
β
d
2
)
)
−
(
α
b
1
+
β
b
2
)
k
s
2
\begin{aligned} & \dot{V}=s\left( -\varepsilon sgn \left( {{s}_{1}} \right)-\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right)ks+\alpha {{d}_{1}}+\beta {{d}_{2}} \right) \\ & \begin{matrix} {} \\ \end{matrix}=\alpha {{s}_{1}}\left( \alpha {{d}_{1}}+\beta {{d}_{2}} \right)-\alpha \varepsilon \left| {{s}_{1}} \right| \\ & -\beta \varepsilon {{s}_{2}}sgn \left( {{s}_{1}} \right)+\beta {{s}_{2}}\left( \alpha {{d}_{1}}+\beta {{d}_{2}} \right)-\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right)k{{s}^{2}} \\ & \begin{matrix} {} \\ \end{matrix}\le \alpha \left| {{s}_{1}} \right|\left( \left| \alpha {{d}_{1}}+\beta {{d}_{2}} \right|-\varepsilon \right)-{{s}_{2}}\beta \left( \varepsilon sgn \left( {{s}_{1}} \right)-\left( \alpha {{d}_{1}}+\beta {{d}_{2}} \right) \right)-\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right)k{{s}^{2}} \\ \end{aligned}
V˙=s(−εsgn(s1)−(αb1+βb2)ks+αd1+βd2)=αs1(αd1+βd2)−αε∣s1∣−βεs2sgn(s1)+βs2(αd1+βd2)−(αb1+βb2)ks2≤α∣s1∣(∣αd1+βd2∣−ε)−s2β(εsgn(s1)−(αd1+βd2))−(αb1+βb2)ks2显然,当有下式时:
{
ε
≥
α
σ
1
+
β
0
σ
2
β
=
{
β
0
s
1
s
2
≥
0
−
β
0
s
1
s
2
<
0
β
0
>
0
α
>
β
0
σ
5
σ
3
>
β
0
b
2
b
1
\left\{ \begin{matrix} \varepsilon \ge \alpha {{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{2}} \\ \beta =\left\{ \begin{matrix} \begin{matrix} \begin{matrix} {{\beta }_{0}} & {{s}_{1}}{{s}_{2}}\ge 0 \\ \end{matrix} \\ \begin{matrix} -{{\beta }_{0}} & {{s}_{1}}{{s}_{2}}<0 \\ \end{matrix} \\ \end{matrix} & {{\beta }_{0}}>0 \\ \end{matrix} \right. \\ \alpha >\frac{{{\beta }_{0}}{{\sigma }_{5}}}{{{\sigma }_{3}}}>\frac{{{\beta }_{0}}{{b}_{2}}}{{{b}_{1}}} \\ \end{matrix} \right.
⎩
⎨
⎧ε≥ασ1+β0σ2β={β0s1s2≥0−β0s1s2<0β0>0α>σ3β0σ5>b1β0b2有:
−
s
2
β
(
ε
s
g
n
(
s
1
)
−
(
α
d
1
+
β
d
2
)
)
≤
−
(
ε
−
∣
α
d
1
+
β
d
2
∣
)
∣
s
2
∣
β
0
≤
0
\begin{aligned} & -{{s}_{2}}\beta \left( \varepsilon sgn \left( {{s}_{1}} \right)-\left( \alpha {{d}_{1}}+\beta {{d}_{2}} \right) \right) \\ & \begin{matrix} {} \\ \end{matrix}\le -\left( \varepsilon -\left| \alpha {{d}_{1}}+\beta {{d}_{2}} \right| \right)\left| {{s}_{2}} \right|{{\beta }_{0}}\le 0 \\ \end{aligned}
−s2β(εsgn(s1)−(αd1+βd2))≤−(ε−∣αd1+βd2∣)∣s2∣β0≤0 有
V
˙
≤
0
\dot{V}\le 0
V˙≤0,即第二层滑模面
s
s
s渐进稳定。
下面证明第一层滑模面稳定,考虑Lyapunov函数:
V
1
=
1
2
s
1
2
{{V}_{1}}=\frac{1}{2}s_{1}^{2}
V1=21s12
求导:
V
˙
1
=
s
1
s
˙
1
=
s
1
(
c
1
x
2
+
g
1
+
b
1
u
+
d
1
)
\begin{aligned} & {{{\dot{V}}}_{1}}={{s}_{1}}{{{\dot{s}}}_{1}} \\ & \begin{matrix} {} \\ \end{matrix}={{s}_{1}}\left( {{c}_{1}}{{x}_{2}}+{{g}_{1}}+{{b}_{1}}u+{{d}_{1}} \right) \\ \end{aligned}
V˙1=s1s˙1=s1(c1x2+g1+b1u+d1)将代入
u
u
u,得:
V
˙
1
=
s
1
(
b
1
u
e
q
2
+
b
1
u
s
w
+
d
1
)
=
s
1
(
b
1
u
e
q
2
+
−
α
b
1
2
u
e
q
2
−
β
b
1
b
2
u
e
q
1
−
ε
b
1
s
g
n
(
s
1
)
α
b
1
+
β
b
2
−
k
b
1
s
+
d
1
)
=
s
1
(
−
β
b
1
b
2
(
u
e
q
1
−
u
e
q
2
)
−
ε
b
1
s
g
n
(
s
1
)
α
b
1
+
β
b
2
−
k
b
1
s
+
d
1
)
=
−
β
b
1
b
2
(
u
e
q
1
−
u
e
q
2
)
s
1
−
ε
b
1
∣
s
1
∣
+
s
1
(
α
b
1
+
β
b
2
)
d
1
α
b
1
+
β
b
2
−
k
b
1
(
α
s
1
2
+
β
s
1
s
2
)
=
−
β
b
1
b
2
(
u
e
q
1
−
u
e
q
2
)
s
1
−
ε
b
1
∣
s
1
∣
+
s
1
(
α
b
1
+
β
b
2
)
d
1
α
b
1
+
β
b
2
−
k
b
1
(
α
s
1
2
+
β
0
∣
s
1
s
2
∣
)
≤
(
∣
(
α
b
1
+
β
b
2
)
d
1
∣
+
∣
β
0
b
1
b
2
∣
u
e
q
1
−
u
e
q
2
∣
∣
−
ε
b
1
)
∣
s
1
∣
α
b
1
+
β
b
2
−
k
b
1
(
α
s
1
2
+
β
0
∣
s
1
s
2
∣
)
\begin{aligned} & {{{\dot{V}}}_{1}}={{s}_{1}}\left( {{b}_{1}}{{u}_{eq2}}+{{b}_{1}}{{u}_{sw}}+{{d}_{1}} \right) \\ & \begin{matrix} {} \\ \end{matrix}={{s}_{1}}\left( {{b}_{1}}{{u}_{eq2}}+\frac{-\alpha b_{1}^{2}{{u}_{eq2}}-\beta {{b}_{1}}{{b}_{2}}{{u}_{eq1}}-\varepsilon {{b}_{1}}sgn \left( {{s}_{1}} \right)}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-k{{b}_{1}}s+{{d}_{1}} \right) \\ & \begin{matrix} {} \\ \end{matrix}={{s}_{1}}\left( \frac{-\beta {{b}_{1}}{{b}_{2}}\left( {{u}_{eq1}}-{{u}_{eq2}} \right)-\varepsilon {{b}_{1}}sgn \left( {{s}_{1}} \right)}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-k{{b}_{1}}s+{{d}_{1}} \right) \\ & \begin{matrix} {} \\ \end{matrix}=\frac{-\beta {{b}_{1}}{{b}_{2}}\left( {{u}_{eq1}}-{{u}_{eq2}} \right){{s}_{1}}-\varepsilon {{b}_{1}}\left| {{s}_{1}} \right|+{{s}_{1}}\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right){{d}_{1}}}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-k{{b}_{1}}\left( \alpha s_{1}^{2}+\beta {{s}_{1}}{{s}_{2}} \right) \\ & \begin{matrix} {} \\ \end{matrix}=\frac{-\beta {{b}_{1}}{{b}_{2}}\left( {{u}_{eq1}}-{{u}_{eq2}} \right){{s}_{1}}-\varepsilon {{b}_{1}}\left| {{s}_{1}} \right|+{{s}_{1}}\left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right){{d}_{1}}}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-k{{b}_{1}}\left( \alpha s_{1}^{2}+{{\beta }_{0}}\left| {{s}_{1}}{{s}_{2}} \right| \right) \\ & \begin{matrix} {} \\ \end{matrix}\le \frac{\left( \left| \left( \alpha {{b}_{1}}+\beta {{b}_{2}} \right){{d}_{1}} \right|+\left| {{\beta }_{0}}{{b}_{1}}{{b}_{2}}\left| {{u}_{eq1}}-{{u}_{eq2}} \right| \right|-\varepsilon {{b}_{1}} \right)\left| {{s}_{1}} \right|}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-k{{b}_{1}}\left( \alpha s_{1}^{2}+{{\beta }_{0}}\left| {{s}_{1}}{{s}_{2}} \right| \right) \\ \end{aligned}
V˙1=s1(b1ueq2+b1usw+d1)=s1(b1ueq2+αb1+βb2−αb12ueq2−βb1b2ueq1−εb1sgn(s1)−kb1s+d1)=s1(αb1+βb2−βb1b2(ueq1−ueq2)−εb1sgn(s1)−kb1s+d1)=αb1+βb2−βb1b2(ueq1−ueq2)s1−εb1∣s1∣+s1(αb1+βb2)d1−kb1(αs12+βs1s2)=αb1+βb2−βb1b2(ueq1−ueq2)s1−εb1∣s1∣+s1(αb1+βb2)d1−kb1(αs12+β0∣s1s2∣)≤αb1+βb2(∣(αb1+βb2)d1∣+∣β0b1b2∣ueq1−ueq2∣∣−εb1)∣s1∣−kb1(αs12+β0∣s1s2∣)显然,当有下式时:
{
ε
>
(
α
σ
4
+
β
0
σ
5
)
σ
1
+
β
0
σ
4
σ
5
∣
u
e
q
2
−
u
e
q
1
∣
σ
3
α
>
β
0
σ
5
σ
3
\left\{ \begin{matrix} \varepsilon >\frac{\left( \alpha {{\sigma }_{4}}+{{\beta }_{0}}{{\sigma }_{5}} \right){{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{4}}{{\sigma }_{5}}\left| {{u}_{eq2}}-{{u}_{eq1}} \right|}{{{\sigma }_{3}}} \\ \alpha >\frac{{{\beta }_{0}}{{\sigma }_{5}}}{{{\sigma }_{3}}} \\ \end{matrix} \right.
{ε>σ3(ασ4+β0σ5)σ1+β0σ4σ5∣ueq2−ueq1∣α>σ3β0σ5有
V
1
≤
0
{{V}_{1}}\le 0
V1≤0,第一层滑模面
s
1
{{s}_{1}}
s1稳定。
由于
s
s
s和
s
1
{{s}_{1}}
s1都稳定,对任意的初始状态,有
lim
t
→
t
1
s
→
0
\underset{t\to {{t}_{1}}}{\mathop{\lim }}\,s\to 0
t→t1lims→0、
lim
t
→
t
2
s
1
→
0
\underset{t\to {{t}_{2}}}{\mathop{\lim }}\,{{s}_{1}}\to 0
t→t2lims1→0,那么对于
s
2
{{s}_{2}}
s2,有
lim
t
→
max
(
t
1
,
t
2
)
s
2
=
lim
t
→
max
(
t
1
,
t
2
)
1
β
(
s
−
α
s
1
)
→
0
\underset{t\to \max \left( {{t}_{1}},{{t}_{2}} \right)}{\mathop{\lim }}\,{{s}_{2}}=\underset{t\to \max \left( {{t}_{1}},{{t}_{2}} \right)}{\mathop{\lim }}\,\frac{1}{\beta }\left( s-\alpha {{s}_{1}} \right)\to 0
t→max(t1,t2)lims2=t→max(t1,t2)limβ1(s−αs1)→0,即
s
2
{{s}_{2}}
s2也能收敛。
综上所述,上文提出的控制律可以使得如、所描述系统稳定,整理如下:
控制律:
{
u
=
u
e
q
1
+
u
e
q
2
+
u
s
w
u
e
q
1
=
c
1
x
2
+
g
1
b
1
c
1
>
0
u
e
q
2
=
c
2
x
4
+
g
2
b
2
c
2
>
0
u
s
w
=
−
α
b
1
u
e
q
2
−
β
b
2
u
e
q
1
−
ε
s
g
n
(
s
1
)
α
b
1
+
β
b
2
−
k
s
\left\{ \begin{matrix} u={{u}_{eq1}}+{{u}_{eq2}}+{{u}_{sw}} \\ \begin{matrix} {{u}_{eq1}}=\frac{{{c}_{1}}{{x}_{2}}+{{g}_{1}}}{{{b}_{1}}} & {{c}_{1}}>0 \\ \end{matrix} \\ \begin{matrix} {{u}_{eq2}}=\frac{{{c}_{2}}{{x}_{4}}+{{g}_{2}}}{{{b}_{2}}} & {{c}_{2}}>0 \\ \end{matrix} \\ {{u}_{sw}}=\frac{-\alpha {{b}_{1}}{{u}_{eq2}}-\beta {{b}_{2}}{{u}_{eq1}}-\varepsilon sgn \left( {{s}_{1}} \right)}{\alpha {{b}_{1}}+\beta {{b}_{2}}}-ks \\ \end{matrix} \right.
⎩
⎨
⎧u=ueq1+ueq2+uswueq1=b1c1x2+g1c1>0ueq2=b2c2x4+g2c2>0usw=αb1+βb2−αb1ueq2−βb2ueq1−εsgn(s1)−ks控制器参数:
{
k
>
0
ε
>
max
(
α
σ
1
+
β
0
σ
2
,
(
α
σ
4
+
β
0
σ
5
)
σ
1
+
β
0
σ
4
σ
5
∣
u
e
q
2
−
u
e
q
1
∣
σ
3
)
β
=
{
β
0
s
1
s
2
≥
0
−
β
0
s
1
s
2
<
0
β
0
>
0
α
>
β
0
σ
5
σ
3
\left\{ \begin{matrix} k>0 \\ \varepsilon >\max \left( \alpha {{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{2}},\frac{\left( \alpha {{\sigma }_{4}}+{{\beta }_{0}}{{\sigma }_{5}} \right){{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{4}}{{\sigma }_{5}}\left| {{u}_{eq2}}-{{u}_{eq1}} \right|}{{{\sigma }_{3}}} \right) \\ \beta =\left\{ \begin{matrix} \begin{matrix} \begin{matrix} {{\beta }_{0}} & {{s}_{1}}{{s}_{2}}\ge 0 \\ \end{matrix} \\ \begin{matrix} -{{\beta }_{0}} & {{s}_{1}}{{s}_{2}}<0 \\ \end{matrix} \\ \end{matrix} & {{\beta }_{0}}>0 \\ \end{matrix} \right. \\ \alpha >\frac{{{\beta }_{0}}{{\sigma }_{5}}}{{{\sigma }_{3}}} \\ \end{matrix} \right.
⎩
⎨
⎧k>0ε>max(ασ1+β0σ2,σ3(ασ4+β0σ5)σ1+β0σ4σ5∣ueq2−ueq1∣)β={β0s1s2≥0−β0s1s2<0β0>0α>σ3β0σ5
4 仿真
系统参数:
{
M
=
1
(
k
g
)
m
=
0.8
(
k
g
)
L
=
0.3
(
m
)
\left\{ \begin{matrix} M=1\left( kg \right) \\ m=0.8\left( kg \right) \\ L=0.3\left( m \right) \\ \end{matrix} \right.
⎩
⎨
⎧M=1(kg)m=0.8(kg)L=0.3(m)控制器参数:
{
c
1
=
0.8
c
2
=
0.4
k
=
2
ε
(
t
)
=
max
(
α
σ
1
+
β
0
σ
2
,
(
α
σ
4
+
β
0
σ
5
)
σ
1
+
β
0
σ
4
σ
5
∣
u
e
q
2
(
t
)
−
u
e
q
1
(
t
)
∣
σ
3
)
β
0
=
0.1
α
=
3.2
\left\{ \begin{matrix} {{c}_{1}}=0.8 \\ {{c}_{2}}=0.4 \\ k=2 \\ \varepsilon \left( t \right)=\max \left( \alpha {{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{2}},\frac{\left( \alpha {{\sigma }_{4}}+{{\beta }_{0}}{{\sigma }_{5}} \right){{\sigma }_{1}}+{{\beta }_{0}}{{\sigma }_{4}}{{\sigma }_{5}}\left| {{u}_{eq2}}\left( t \right)-{{u}_{eq1}}\left( t \right) \right|}{{{\sigma }_{3}}} \right) \\ {{\beta }_{0}}=0.1 \\ \alpha =3.2 \\ \end{matrix} \right.
⎩
⎨
⎧c1=0.8c2=0.4k=2ε(t)=max(ασ1+β0σ2,σ3(ασ4+β0σ5)σ1+β0σ4σ5∣ueq2(t)−ueq1(t)∣)β0=0.1α=3.2初始状态:
{
x
=
2
θ
=
30
∘
\left\{ \begin{matrix} x=2 \\ \theta =30{}^\circ \\ \end{matrix} \right.
{x=2θ=30∘仿真结果如下:
5 参考文献
杨兴明,“基于分层滑模方法的两轮载人自平衡车的运动控制”,合肥工业大学学报,2013
王伟,“基于滑模方法的桥式吊车系统的抗摆控制”,控制与决策,2004
王伟,“桥式吊车系统的分级滑模控制方法”,自动化学报,2004