ملاحظ لوننبرگر هوتعبير عن نظام يستعمل لحساب الحالة في نظام آخر وتكون مداخل الملاحظ هي مداخل ومخارج النظام وتكون مخارج الملاحظ هي حالات النظام المراد فهم حالته. يستعمل ملاحظ لوننبرگر في التحكم عن طريق إرجاع الحالة.
ملاحظ الحالة النمطي
The state of a physical discrete-time system is assumed to satisfy
-
ملاحظ النمط المنزلق
where:
- The
- sgn(z)=[sgn(z1)sgn(z2)⋮sgn(zi)⋮sgn(zn)]{\displaystyle \operatorname {sgn (\mathbf {z )={\begin{bmatrix \operatorname {sgn (z_{1 )\\\operatorname {sgn (z_{2 )\\\vdots \\\operatorname {sgn (z_{i )\\\vdots \\\operatorname {sgn (z_{n )\end{bmatrix
- for the vector z∈Rn{\displaystyle \mathbf {z \in \mathbb {R ^{n .
- The vector H(x){\displaystyle H(\mathbf {x ) has components that are the output function h(x){\displaystyle h(\mathbf {x ) and its repeated Lie derivatives. In particular,
- H(x)≜[h1(x)h2(x)h3(x)⋮hn(x)]≜[h(x)Lfh(x)Lf2h(x)⋮Lfn−1h(x)]{\displaystyle H(\mathbf {x )\triangleq {\begin{bmatrix h_{1 (\mathbf {x )\\h_{2 (\mathbf {x )\\h_{3 (\mathbf {x )\\\vdots \\h_{n (\mathbf {x )\end{bmatrix \triangleq {\begin{bmatrix h(\mathbf {x )\\L_{f h(\mathbf {x )\\L_{f ^{2 h(\mathbf {x )\\\vdots \\L_{f ^{n-1 h(\mathbf {x )\end{bmatrix
- where LfihLie derivative of output function hrelative degree of n, H(x(t))Jacobian linearization of H(x)diffeomorphism.
- The diagonal matrix M(x^){\displaystyle M({\hat {\mathbf {x ) of gains is such that
- M(x^)≜diag(m1(x^),m2(x^),…,mn(x^))=[m1(x^)m2(x^)⋱mi(x^)⋱mn(x^)]{\displaystyle M({\hat {\mathbf {x )\triangleq \operatorname {diag (m_{1 ({\hat {\mathbf {x ),m_{2 ({\hat {\mathbf {x ),\ldots ,m_{n ({\hat {\mathbf {x ))={\begin{bmatrix m_{1 ({\hat {\mathbf {x )&&&&&\\&m_{2 ({\hat {\mathbf {x )&&&&\\&&\ddots &&&\\&&&m_{i ({\hat {\mathbf {x )&&\\&&&&\ddots &\\&&&&&m_{n ({\hat {\mathbf {x )\end{bmatrix
- where, for each i∈{1,2,…,n {\displaystyle i\in \{1,2,\dots ,n\ , element mi(x^)>0{\displaystyle m_{i ({\hat {\mathbf {x )>0 and suitably large to ensure reachability of the sliding mode.
- The observer vector V(t){\displaystyle V(t) is such that
- V(t)≜[v1(t)v2(t)v3(t)⋮vi(t)⋮vn(t)]≜[y(t){m1(x^)sgn(v1(t)−h1(x^(t))) eq{m2(x^)sgn(v2(t)−h2(x^(t))) eq⋮{mi−1(x^)sgn(vi−1(t)−hi−1(x^(t))) eq⋮{mn−1(x^)sgn(vn−1(t)−hn−1(x^(t))) eq]{\displaystyle V(t)\triangleq {\begin{bmatrix v_{1 (t)\\v_{2 (t)\\v_{3 (t)\\\vdots \\v_{i (t)\\\vdots \\v_{n (t)\end{bmatrix \triangleq {\begin{bmatrix \mathbf {y (t)\\\{m_{1 ({\hat {\mathbf {x )\operatorname {sgn (v_{1 (t)-h_{1 ({\hat {\mathbf {x (t)))\ _{\text{eq \\\{m_{2 ({\hat {\mathbf {x )\operatorname {sgn (v_{2 (t)-h_{2 ({\hat {\mathbf {x (t)))\ _{\text{eq \\\vdots \\\{m_{i-1 ({\hat {\mathbf {x )\operatorname {sgn (v_{i-1 (t)-h_{i-1 ({\hat {\mathbf {x (t)))\ _{\text{eq \\\vdots \\\{m_{n-1 ({\hat {\mathbf {x )\operatorname {sgn (v_{n-1 (t)-h_{n-1 ({\hat {\mathbf {x (t)))\ _{\text{eq \end{bmatrix
- where sgn(⋅)signum function defined for scalars, and {… eq{\displaystyle \{\ldots \ _{\text{eq denotes an "equivalent value operator" of a discontinuous function in sliding mode.
The modified observation error can be written in the transformed states e=H(x)−H(x^){\displaystyle \mathbf {e =H(\mathbf {x )-H(\mathbf {\hat {x ) . In particular,
- e˙=ddtH(x)−ddtH(x^)=ddtH(x)−M(x^)sgn(V(t)−H(x^(t))),{\displaystyle {\begin{aligned {\dot {\mathbf {e &={\frac {\operatorname {d {\operatorname {d t H(\mathbf {x )-{\frac {\operatorname {d {\operatorname {d t H({\hat {\mathbf {x )\\&={\frac {\operatorname {d {\operatorname {d t H(\mathbf {x )-M({\hat {\mathbf {x )\,\operatorname {sgn (V(t)-H({\hat {\mathbf {x (t))),\end{aligned
and so
- [e˙1e˙2⋮e˙i⋮e˙n−1e˙n]=[h˙1(x)h˙2(x)⋮h˙i(x)⋮h˙n−1(x)h˙n(x)]⏞ddtH(x)−M(x^)sgn(V(t)−H(x^(t)))⏞ddtH(x^)=[h2(x)h3(x)⋮hi+1(x)⋮hn(x)Lfnh(x)]−[m1sgn(v1(t)−h1(x^(t)))m2sgn(v2(t)−h2(x^(t)))⋮misgn(vi(t)−hi(x^(t)))⋮mn−1sgn(vn−1(t)−hn−1(x^(t)))mnsgn(vn(t)−hn(x^(t)))]=[h2(x)−m1(x^)sgn(v1(t)⏞v1(t)=y(t)=h1(x)−h1(x^(t))⏞e1)h3(x)−m2(x^)sgn(v2(t)−h2(x^(t)))⋮hi+1(x)−mi(x^)sgn(vi(t)−hi(x^(t)))⋮hn(x)−mn−1(x^)sgn(vn−1(t)−hn−1(x^(t)))Lfnh(x)−mn(x^)sgn(vn(t)−hn(x^(t)))].{\displaystyle {\begin{aligned {\begin{bmatrix {\dot {\mathbf {e _{1 \\{\dot {\mathbf {e _{2 \\\vdots \\{\dot {\mathbf {e _{i \\\vdots \\{\dot {\mathbf {e _{n-1 \\{\dot {\mathbf {e _{n \end{bmatrix &={\mathord {\overbrace {\begin{bmatrix {\dot {h _{1 (\mathbf {x )\\{\dot {h _{2 (\mathbf {x )\\\vdots \\{\dot {h _{i (\mathbf {x )\\\vdots \\{\dot {h _{n-1 (\mathbf {x )\\{\dot {h _{n (\mathbf {x )\end{bmatrix ^{{\tfrac {\operatorname {d {\operatorname {d t H(\mathbf {x ) -{\mathord {\overbrace {M({\hat {\mathbf {x )\,\operatorname {sgn (V(t)-H({\hat {\mathbf {x (t))) ^{{\tfrac {\operatorname {d {\operatorname {d t H(\mathbf {\hat {x ) ={\begin{bmatrix h_{2 (\mathbf {x )\\h_{3 (\mathbf {x )\\\vdots \\h_{i+1 (\mathbf {x )\\\vdots \\h_{n (\mathbf {x )\\L_{f ^{n h(\mathbf {x )\end{bmatrix -{\begin{bmatrix m_{1 \operatorname {sgn (v_{1 (t)-h_{1 ({\hat {\mathbf {x (t)))\\m_{2 \operatorname {sgn (v_{2 (t)-h_{2 ({\hat {\mathbf {x (t)))\\\vdots \\m_{i \operatorname {sgn (v_{i (t)-h_{i ({\hat {\mathbf {x (t)))\\\vdots \\m_{n-1 \operatorname {sgn (v_{n-1 (t)-h_{n-1 ({\hat {\mathbf {x (t)))\\m_{n \operatorname {sgn (v_{n (t)-h_{n ({\hat {\mathbf {x (t)))\end{bmatrix \\&={\begin{bmatrix h_{2 (\mathbf {x )-m_{1 ({\hat {\mathbf {x )\operatorname {sgn ({\mathord {\overbrace {{\mathord {\overbrace {v_{1 (t) ^{v_{1 (t)=y(t)=h_{1 (\mathbf {x ) -h_{1 ({\hat {\mathbf {x (t)) ^{\mathbf {e _{1 )\\h_{3 (\mathbf {x )-m_{2 ({\hat {\mathbf {x )\operatorname {sgn (v_{2 (t)-h_{2 ({\hat {\mathbf {x (t)))\\\vdots \\h_{i+1 (\mathbf {x )-m_{i ({\hat {\mathbf {x )\operatorname {sgn (v_{i (t)-h_{i ({\hat {\mathbf {x (t)))\\\vdots \\h_{n (\mathbf {x )-m_{n-1 ({\hat {\mathbf {x )\operatorname {sgn (v_{n-1 (t)-h_{n-1 ({\hat {\mathbf {x (t)))\\L_{f ^{n h(\mathbf {x )-m_{n ({\hat {\mathbf {x )\operatorname {sgn (v_{n (t)-h_{n ({\hat {\mathbf {x (t)))\end{bmatrix .\end{aligned
وبذلك:
- ما دام m1(x^)≥|h2(x(t))|{\displaystyle m_{1 (\mathbf {\hat {x )\geq |h_{2 (\mathbf {x (t))| , the first row of the error dynamics, e˙1=h2(x^)−m1(x^)sgn(e1){\displaystyle {\dot {\mathbf {e _{1 =h_{2 ({\hat {\mathbf {x )-m_{1 ({\hat {\mathbf {x )\operatorname {sgn (\mathbf {e _{1 ) , will meet sufficient conditions to enter the e1=0{\displaystyle e_{1 =0 sliding mode in finite time.
- Along the e1=0{\displaystyle e_{1 =0 surface, the corresponding v2(t)={m1(x^)sgn(e1) eq{\displaystyle v_{2 (t)=\{m_{1 ({\hat {\mathbf {x )\operatorname {sgn (\mathbf {e _{1 )\ _{\text{eq equivalent control will be equal to h2(x){\displaystyle h_{2 (\mathbf {x ) , and so v2(t)−h2(x^)=h2(x)−h2(x^)=e2{\displaystyle v_{2 (t)-h_{2 ({\hat {\mathbf {x )=h_{2 (\mathbf {x )-h_{2 ({\hat {\mathbf {x )=\mathbf {e _{2 . Hence, so long as m2(x^)≥|h3(x(t))|{\displaystyle m_{2 (\mathbf {\hat {x )\geq |h_{3 (\mathbf {x (t))| , the second row of the error dynamics, e˙2=h3(x^)−m2(x^)sgn(e2){\displaystyle {\dot {\mathbf {e _{2 =h_{3 ({\hat {\mathbf {x )-m_{2 ({\hat {\mathbf {x )\operatorname {sgn (\mathbf {e _{2 ) , will enter the e2=0{\displaystyle e_{2 =0 sliding mode in finite time.
- Along the ei=0{\displaystyle e_{i =0 surface, the corresponding vi+1(t)={… eq{\displaystyle v_{i+1 (t)=\{\ldots \ _{\text{eq equivalent control will be equal to hi+1(x){\displaystyle h_{i+1 (\mathbf {x ) . Hence, so long as mi+1(x^)≥|hi+2(x(t))|{\displaystyle m_{i+1 (\mathbf {\hat {x )\geq |h_{i+2 (\mathbf {x (t))| , the (i+1){\displaystyle (i+1) th row of the error dynamics, e˙i+1=hi+2(x^)−mi+1(x^)sgn(ei+1){\displaystyle {\dot {\mathbf {e _{i+1 =h_{i+2 ({\hat {\mathbf {x )-m_{i+1 ({\hat {\mathbf {x )\operatorname {sgn (\mathbf {e _{i+1 ) , will enter the ei+1=0{\displaystyle e_{i+1 =0 sliding mode in finite time.
So, for sufficiently large midiffeomorphism (i.e., that its Jacobian linearization is invertible) asserts that convergence of the estimated output implies convergence of the estimated state. That is, the requirement is an observability condition.
In the case of the sliding mode observer for the system with the input, additional conditions are needed for the observation error to be independent of the input. For example, that
- ∂H(x)∂xB(x){\displaystyle {\frac {\partial H(\mathbf {x ) {\partial \mathbf {x B(\mathbf {x )
does not depend on time. The observer is then
- x^˙=[∂H(x^)∂x]−1M(x^)sgn(V(t)−H(x^))+B(x^)u.{\displaystyle {\dot {\mathbf {\hat {x =\left[{\frac {\partial H(\mathbf {\hat {x ) {\partial \mathbf {x \right]^{-1 M(\mathbf {\hat {x )\operatorname {sgn (V(t)-H(\mathbf {\hat {x ))+B(\mathbf {\hat {x )u.
انظر أيضاً
- مرشح كالمان
- مبرهنة ماسرليز
|
هذه الموضوعة تعبير عن بذرة بحاجة للنمووالتحسين؛ فساهم في إثرائها بالمشاركة في تحريرها.
|