# 自抗扰控制中的扩张状态观测器收敛性分析1

11 篇文章 0 订阅

(1) { x ˙ 1 ( t ) = x 2 ( t ) , &ThickSpace; x 1 ( 0 ) = x 10 , x ˙ 2 ( t ) = x 3 ( t ) , &ThickSpace; x 2 ( 0 ) = x 20 , ⋮ x ˙ n ( t ) = f ( t , x 1 ( t ) , x 2 ( t ) , … , x n ( t ) ) + w ( t ) + u ( t ) , &ThickSpace; x n ( 0 ) = x n 0 , y ( t ) = x 1 ( t ) , \left\{\begin{aligned} &amp;\dot{x}_1(t)=x_2(t),\;x_1(0)=x_{10},\\ &amp;\dot{x}_2(t)=x_3(t),\;x_2(0)=x_{20},\\ &amp;\vdots\\ &amp;\dot{x}_n(t)=f(t,x_1(t),x_2(t),\ldots,x_n(t))+w(t)+u(t),\;x_n(0)=x_{n0},\\ &amp;y(t)=x_1(t), \end{aligned}\right.\tag{1}

∣ u ∣ + ∣ f ∣ + ∣ w ˙ ∣ + ∣ ∂ f ∂ t ∣ + ∑ i = 1 n ∣ ∂ f ∂ x i ∣ ≤ c 0 + ∑ j = 1 n c j ∣ x j ∣ k , \left\vert u\right\vert+\left\vert f\right\vert+\left\vert\dot{w}\right\vert+\left\vert\frac{\partial f}{\partial t}\right\vert +\sum\limits_{i=1}^{n}\left\vert\frac{\partial f}{\partial x_i}\right\vert\leq c_0+\sum\limits_{j=1}^{n}c_j\left\vert x_j\right\vert^k,

(2) { x ˙ 1 ( t ) = x 2 ( t ) , &ThickSpace; x 1 ( 0 ) = x 10 , x ˙ 2 ( t ) = x 3 ( t ) , &ThickSpace; x 2 ( 0 ) = x 20 , ⋮ x ˙ n ( t ) = x n + 1 ( t ) , &ThickSpace; x n ( 0 ) = x n 0 , x ˙ n + 1 ( t ) = L ˙ ( t ) , &ThickSpace; x n + 1 ( 0 ) = L ( 0 ) , y ( t ) = x 1 ( t ) , \left\{\begin{aligned} &amp;\dot{x}_1(t)=x_2(t),\;x_1(0)=x_{10},\\ &amp;\dot{x}_2(t)=x_3(t),\;x_2(0)=x_{20},\\ &amp;\vdots\\ &amp;\dot{x}_n(t)=x_{n+1}(t),\;x_n(0)=x_{n0},\\ &amp;\dot{x}_{n+1}(t)=\dot{L}(t),\;x_{n+1}(0)=L(0),\\ &amp;y(t)=x_1(t), \end{aligned}\right.\tag{2}

(3) { x ^ ˙ 1 ( t ) = x ^ 2 ( t ) + ϵ n − 1 g 1 ( y ( t ) − x ^ 1 ( t ) ε n ) , x ^ ˙ 2 ( t ) = x ^ 3 ( t ) + ϵ n − 2 g 2 ( y ( t ) − x ^ 1 ( t ) ε n ) , ⋮ x ^ ˙ n ( t ) = x n + 1 ( t ) + g n ( y ( t ) − x ^ 1 ( t ) ε n ) + u ( t ) , x ^ ˙ n + 1 ( t ) = 1 ϵ g n + 1 ( y ( t ) − x ^ 1 ( t ) ε n ) , \left\{\begin{aligned} &amp;\dot{\hat{x}}_1(t)=\hat{x}_2(t)+\epsilon^{n-1}g_1\left(\frac{y(t)-\hat{x}_1(t)}{\varepsilon^n}\right),\\ &amp;\dot{\hat{x}}_2(t)=\hat{x}_3(t)+\epsilon^{n-2}g_2\left(\frac{y(t)-\hat{x}_1(t)}{\varepsilon^n}\right),\\ &amp;\vdots\\ &amp;\dot{\hat{x}}_n(t)=x_{n+1}(t)+g_n\left(\frac{y(t)-\hat{x}_1(t)}{\varepsilon^n}\right)+u(t),\\ &amp;\dot{\hat{x}}_{n+1}(t)=\frac{1}{\epsilon}g_{n+1}\left(\frac{y(t)-\hat{x}_1(t)}{\varepsilon^n}\right), \end{aligned}\right.\tag{3}

(4) { x ^ ˙ 1 ( t ) = x ^ 2 ( t ) + α 1 ϵ ( y ( t ) − x ^ 1 ( t ) ) , x ^ ˙ 2 ( t ) = x ^ 3 ( t ) + α 2 ϵ 2 ( y ( t ) − x ^ 1 ( t ) ) , ⋮ x ^ ˙ n ( t ) = x n + 1 ( t ) + α n ϵ n ( y ( t ) − x ^ 1 ( t ) ) + u ( t ) , x ^ ˙ n + 1 ( t ) = α n + 1 ϵ n + 1 ( y ( t ) − x ^ 1 ( t ) ) , \left\{\begin{aligned} &amp;\dot{\hat{x}}_1(t)=\hat{x}_2(t)+\frac{\alpha_1}{\epsilon}\left(y(t)-\hat{x}_1(t)\right),\\ &amp;\dot{\hat{x}}_2(t)=\hat{x}_3(t)+\frac{\alpha_2}{\epsilon^2}\left(y(t)-\hat{x}_1(t)\right),\\ &amp;\vdots\\ &amp;\dot{\hat{x}}_n(t)=x_{n+1}(t)+\frac{\alpha_n}{\epsilon^n}\left(y(t)-\hat{x}_1(t)\right)+u(t),\\ &amp;\dot{\hat{x}}_{n+1}(t)=\frac{\alpha_{n+1}}{\epsilon^{n+1}}\left(y(t)-\hat{x}_1(t)\right), \end{aligned}\right.\tag{4}

• λ 1 ∥ y ∥ 2 ≤ V ( y ) ≤ λ 2 ∥ y ∥ 2 \lambda_1\Vert y\Vert^2\leq V(y)\leq\lambda_2\Vert y\Vert^2 λ 3 ∥ y ∥ 2 ≤ W ( y ) ≤ λ 4 ∥ y ∥ 2 \lambda_3\Vert y\Vert^2\leq W(y)\leq\lambda_4\Vert y\Vert^2
• ∑ i = 1 n ∂ V ∂ y i ( y i + 1 − g i ( y 1 ) ) − ∂ V ∂ y n + 1 g n + 1 ( y 1 ) ≤ − W ( y ) \sum\limits_{i=1}^{n}\frac{\partial V}{\partial y_i}(y_{i+1}-g_i(y_1))- \frac{\partial V}{\partial y_{n+1}}g_{n+1}(y_1)\leq -W(y)
• ∣ ∂ V ∂ y n + 1 ∣ ≤ β ∥ y ∥ \left\vert\frac{\partial V}{\partial y_{n+1}}\right\vert\leq\beta \Vert y\Vert

(i) 对每一个正常数 a a lim ⁡ ε → 0 ∣ x i ( t ) − x ^ i ( t ) ∣ = 0 \lim\limits_{\varepsilon\rightarrow 0}\vert x_i(t)-\hat{x}_i(t)\vert=0 t ∈ [ a , ∞ ) t\in[a,\infty) 一致成立；
(ii) lim ⁡ t → ∞ ∣ x i ( t ) − x ^ i ( t ) ∣ ≤ O ( ε n + 2 − i ) \lim\limits_{t\rightarrow \infty}\vert x_i(t)-\hat{x}_i(t)\vert\leq O(\varepsilon^{n+2-i})

(5) Δ ( t ) = d d s f ( s , x 1 ( s ) , … , x n ( s ) ) ∣ s = ε t + w ˙ ( ε t ) = ∂ ∂ t f ( ε t , x 1 ( ε t ) , … , x n ( ε t ) ) + ∑ i = 1 n x i + 1 ( ε t ) ∂ ∂ x i f ( ε t , x 1 ( ε t ) , … , x n ( ε t ) ) + u ( ε t ) ∂ ∂ x n f ( ε t , x 1 ( ε t ) , … , x n ( ε t ) ) + w ˙ ( ε t ) . \begin{aligned} \Delta(t)=&amp;\left.\frac{\mathrm{d}}{\mathrm{d}s}f(s,x_1(s),\ldots,x_n(s))\right|_{s=\varepsilon t}+\dot{w}(\varepsilon t)\\ =&amp;\frac{\partial}{\partial t}f(\varepsilon t,x_1(\varepsilon t),\ldots,x_n(\varepsilon t)) +\sum_{i=1}^{n}x_{i+1}(\varepsilon t)\frac{\partial}{\partial x_i}f(\varepsilon t,x_1(\varepsilon t),\ldots,x_n(\varepsilon t))\\ &amp;+u(\varepsilon t)\frac{\partial}{\partial x_n}f(\varepsilon t,x_1(\varepsilon t),\ldots,x_n(\varepsilon t)) +\dot{w}(\varepsilon t). \end{aligned}\tag{5}

(6) e i ( t ) = x i ( t ) − x ^ i ( t ) , &ThickSpace; η i ( t ) = e i ( ε t ) ε n + 1 − i , &ThickSpace; i = 1 , 2 , … , n + 1 , e_i(t)=x_i(t)-\hat{x}_i(t),\;\eta_i(t)=\frac{e_i(\varepsilon t)}{\varepsilon^{n+1-i}},\;i=1,2,\ldots,n+1,\tag{6}

(7) { η ˙ 1 ( t ) = η 2 ( t ) − g 1 ( η 1 ( t ) ) , η 1 ( 0 ) = e 1 ( 0 ) ε n , η ˙ 2 ( t ) = η 3 ( t ) − g 2 ( η 1 ( t ) ) , η 2 ( 0 ) = e 2 ( 0 ) ε n − 1 , ⋮ η ˙ n ( t ) = η n + 1 ( t ) − g n ( η 1 ( t ) ) , η n ( 0 ) = e n ( 0 ) ε , η ˙ n + 1 ( t ) = − g n + 1 ( η 1 ( t ) ) + ε Δ ( t ) , η n + 1 ( 0 ) = e n + 1 ( 0 ) . \left\{\begin{aligned} &amp;\dot{\eta}_1(t)=\eta_2(t)-g_1(\eta_1(t)),\eta_1(0)=\frac{e_1(0)}{\varepsilon^n},\\ &amp;\dot{\eta}_2(t)=\eta_3(t)-g_2(\eta_1(t)),\eta_2(0)=\frac{e_2(0)}{\varepsilon^{n-1}},\\ &amp;\vdots\\ &amp;\dot{\eta}_n(t)=\eta_{n+1}(t)-g_n(\eta_1(t)),\eta_n(0)=\frac{e_n(0)}{\varepsilon},\\ &amp;\dot{\eta}_{n+1}(t)=-g_{n+1}(\eta_1(t))+\varepsilon\Delta(t),\eta_{n+1}(0)=e_{n+1}(0). \end{aligned}\right.\tag{7}

(8) d d t V ( η ( t ) ) = ∑ i = 1 n ∂ V ∂ η i ( η i + 1 − g i ( η 1 ) ) − ∂ V ∂ η n + 1 g n + 1 ( η 1 ) + ∂ V ∂ η n + 1 ε Δ ≤ − W ( η ) + ε M β ∥ η ∥ ≤ − λ 3 λ 2 V ( η ) + λ 1 λ 1 ε M β V ( η ) . \begin{aligned}\frac{\mathrm{d}}{\mathrm{d}t}V(\eta(t)) =&amp;\sum_{i=1}^{n}\frac{\partial V}{\partial \eta_i}(\eta_{i+1}-g_i(\eta_1)) -\frac{\partial V}{\partial \eta_{n+1}}g_{n+1}(\eta_1) +\frac{\partial V}{\partial \eta_{n+1}}\varepsilon\Delta\\ \leq&amp;-W(\eta)+\varepsilon M\beta\Vert\eta\Vert \leq-\frac{\lambda_3}{\lambda_2}V(\eta)+\frac{\sqrt{\lambda_1}}{\lambda_1}\varepsilon M\beta\sqrt{V(\eta)}.\end{aligned}\tag{8}

(9) d d t V ( η ( t ) ) ≤ − λ 3 2 λ 2 V ( η ( t ) ) + λ 1 ε M β 2 λ 1 . \frac{\mathrm{d}}{\mathrm{d}t}\sqrt{V(\eta(t))} \leq -\frac{\lambda_3}{2\lambda_2}\sqrt{V(\eta(t))}+\frac{\sqrt{\lambda_1}\varepsilon M\beta}{2\lambda_1}.\tag{9}

(10) ∥ η ( t ) ∥ ≤ V ( η ( t ) ) λ 1 ≤ λ 1 V ( η ( 0 ) ) λ 1 e − λ 3 2 λ 2 t + ε M β 2 λ 1 ∫ 0 t e − λ 3 2 λ 2 ( t − s ) d s . \Vert\eta(t)\Vert\leq\sqrt{\frac{V(\eta(t))}{\lambda_1}} \leq\frac{\sqrt{\lambda_1V(\eta(0))}}{\lambda_1}\mathrm{e}^{-\frac{\lambda_3}{2\lambda_2}t} +\frac{\varepsilon M\beta}{2\lambda_1}\int_{0}^{t}\mathrm{e}^{-\frac{\lambda_3}{2\lambda_2}(t-s)}\mathrm{d}s.\tag{10}

(11) ∣ e i ( t ) ∣ = ε n + 1 − i ∣ η i ( t ε ) ∣ ≤ ε n + 1 − i ∥ η ( t ε ) ∥ ≤ ε n + 1 − i [ λ 1 V ( η ( 0 ) ) λ 1 e − λ 3 t 2 λ 2 ε + ε M β 2 λ 1 ∫ 0 t ε e − λ 3 2 λ 2 ( t / ε − s ) d s ] . \begin{aligned} \vert e_i(t)\vert&amp;=\varepsilon^{n+1-i}\left\vert\eta_i\left(\frac{t}{\varepsilon}\right)\right\vert \leq\varepsilon^{n+1-i}\left\Vert\eta\left(\frac{t}{\varepsilon}\right)\right\Vert\\ &amp;\leq\varepsilon^{n+1-i}\left[ \frac{\sqrt{\lambda_1V(\eta(0))}}{\lambda_1}\mathrm{e}^{-\frac{\lambda_3 t}{2\lambda_2\varepsilon}} +\frac{\varepsilon M\beta}{2\lambda_1}\int_{0}^{\frac{t}{\varepsilon}}\mathrm{e}^{-\frac{\lambda_3}{2\lambda_2}(t/\varepsilon-s)}\mathrm{d}s\right].\end{aligned}\tag{11}

(12) { η ˙ 1 ( t ) = η 2 ( t ) − α 1 η 1 ( t ) , η 1 ( 0 ) = e 1 ( 0 ) ε n , η ˙ 2 ( t ) = η 3 ( t ) − α 2 η 1 ( t ) , η 2 ( 0 ) = e 2 ( 0 ) ε n − 1 , ⋮ η ˙ n ( t ) = η n + 1 ( t ) − α n η 1 ( t ) , η n ( 0 ) = e n ( 0 ) ε , η ˙ n + 1 ( t ) = − α n + 1 η 1 ( t ) + ε Δ ( t ) , η n + 1 ( 0 ) = e n + 1 ( 0 ) . \left\{\begin{aligned} &amp;\dot{\eta}_1(t)=\eta_2(t)-\alpha_1\eta_1(t),\eta_1(0)=\frac{e_1(0)}{\varepsilon^n},\\ &amp;\dot{\eta}_2(t)=\eta_3(t)-\alpha_2\eta_1(t),\eta_2(0)=\frac{e_2(0)}{\varepsilon^{n-1}},\\ &amp;\vdots\\ &amp;\dot{\eta}_n(t)=\eta_{n+1}(t)-\alpha_n\eta_1(t),\eta_n(0)=\frac{e_n(0)}{\varepsilon},\\ &amp;\dot{\eta}_{n+1}(t)=-\alpha_{n+1}\eta_1(t)+\varepsilon\Delta(t),\eta_{n+1}(0)=e_{n+1}(0). \end{aligned}\right.\tag{12}

(13) E = [ − α 1 1 0 ⋯ 0 − α 2 0 1 ⋯ 0 ⋮ ⋮ ⋮ ⋱ ⋮ − α n 0 0 ⋯ 1 − α n + 1 0 0 ⋯ 0 ] , E=\left[\begin{matrix} -\alpha_1 &amp; 1 &amp; 0 &amp; \cdots &amp; 0\\ -\alpha_2 &amp; 0 &amp; 1 &amp; \cdots &amp; 0\\ \vdots &amp; \vdots &amp; \vdots &amp;\ddots &amp;\vdots\\ -\alpha_n &amp; 0 &amp; 0 &amp; \cdots &amp; 1\\ -\alpha_{n+1} &amp; 0 &amp; 0 &amp; \cdots &amp; 0\\ \end{matrix}\right],\tag{13}

P P 为满足Lyapunov方程 P E + E T P = − I PE+E^\mathrm{T}P=-I 的正定矩阵， I I n + 1 n+1 维单位矩阵，则假设3中的函数 V V W W R n + 1 → R \mathbb{R}^{n+1}\rightarrow\mathbb{R} 可以定义为

(14) V ( η ) = η T P η , &ThickSpace;&ThickSpace; W ( η ) = η T η , &ThickSpace;&ThickSpace; ∀ η ∈ R n + 1 . V(\eta)=\eta^\mathrm{T}P\eta,\;\;W(\eta)=\eta^\mathrm{T}\eta,\;\;\forall \eta\in\mathbb{R}^{n+1}.\tag{14}

λ min ⁡ ( P ) ∥ η ∥ 2 ≤ V ( η ) ≤ λ max ⁡ ( P ) ∥ η ∥ 2 , \lambda_{\min}(P)\Vert\eta\Vert^2\leq V(\eta)\leq\lambda_{\max}(P)\Vert\eta\Vert^2,

∑ i = 1 n ∂ V ∂ η i ( η i + 1 − α i η 1 ) − ∂ V ∂ η n + 1 α n + 1 η 1 = − η T η = − ∥ η ∥ 2 = − W ( y ) , \sum_{i=1}^{n}\frac{\partial V}{\partial \eta_i}(\eta_{i+1}-\alpha_i\eta_1)- \frac{\partial V}{\partial \eta_{n+1}}\alpha_{n+1}\eta_1 =-\eta^\mathrm{T}\eta=-\Vert\eta\Vert^2=-W(y),

∣ ∂ V ∂ η n + 1 ∣ ≤ ∥ ∂ V ∂ η ∥ = ∥ 2 η T P ∥ ≤ 2 ∥ P ∥ ∥ η ∥ = 2 λ max ⁡ ( P ) ∥ η ∥ , \left\vert\frac{\partial V}{\partial \eta_{n+1}}\right\vert \leq\left\Vert\frac{\partial V}{\partial \eta}\right\Vert =\Vert2\eta^\mathrm{T}P\Vert\leq 2\Vert P\Vert \Vert\eta\Vert =2\lambda_{\max}(P)\Vert \eta\Vert,

[1]韩京清. 自抗扰控制技术: 估计补偿不确定因素的控制技术[M]. 国防工业出版社, 2008.

[2]Guo B Z, Zhao Z. On the convergence of an extended state observer for nonlinear systems with uncertainty[J]. Systems & Control Letters, 2011, 60(6): 420-430.

[3]Zhao Z L, Guo B Z. Extended state observer for uncertain lower triangular nonlinear systems[J]. Systems & Control Letters, 2015, 85: 100-108.

[4]Khalil H K. Nonlinear systems[M]. Prentice-Hall, 2001.

[5]Freidovich L B, Khalil H K. Performance recovery of feedback-linearization-based designs[J]. IEEE Transactions on automatic control, 2008, 53(10): 2324-2334.

• 11
点赞
• 8
评论
• 36
收藏
• 一键三连
• 扫一扫，分享海报