text
stringlengths 0
89.3k
|
|---|
j0
|
d
|
AΦe
|
Nj d
|
BΦν
|
NjΞjW W
|
σNW d 1 nD
|
or in compact notation using Ψd
|
Nj d
|
AΦe
|
Nj
|
d
|
BΦν
|
NjΞj
|
αd
|
KfZfN1M
|
j0Ψd
|
NjWW σNW d 1 nD31
|
Remark 2 Since we derived terminal disturbance in
|
clusion 31 similar to 22 in Section 3 Remark 1
|
also applies to 31 and thus allows to replace σNwith
|
diag σN1 σNnifWis designed as a hyperrectan
|
gle Such a choice for Wthus increases the flexibility
|
of the SLTMPC problem proposed below However de
|
pending on the disturbance sets WD choosing Was a
|
hyperrectangle might be conservative
|
Before formulating the proposed filterbased SLTMPC
|
problem we state some standard assumptions on the
|
nominal cost function
|
Assumption 1 The stage cost function lx uis con
|
vex satisfies l00 0 is uniformly continuous for all
|
x X u U and there exists a Kfunction γ1satisfy
|
inglx uγ1xx X u U The terminal cost
|
lfxis convex positive definite uniformly continuous
|
for all x X f there exists a Kfunction γ2satisfy
|
inglfxγ2xx Xf and it holds that lfx
|
lfx lx K fxwith xAxBKfxx X f
|
andKfx U
|
7The recusively feasible version of the filterbased
|
SLTMPC problem is then formulated as
|
min
|
zvpα0
|
ΦeΦνΣΞlfzN N1X
|
i0lzi vi 32a
|
sti 0 N1
|
z0xk 32b
|
zi1AziBvipi 32c
|
h
|
INnZANZBNi
|
Φe
|
Φν
|
Σ32d
|
Φe
|
Nj1AΦe
|
NjBΦν
|
Nj Ξj32e
|
zi X F iΦe 32f
|
vi U F iΦν 32g
|
zNαZf X F NΦe 32h
|
αKfZf U F NΦν 32i
|
αAKfZfαZfΓW 32j
|
d 1 nD
|
ψd
|
i i1M
|
j0Ψd
|
ijW W σi1W32k
|
αd
|
KfZfN1M
|
j0Ψd
|
NjWW σNW32l
|
where l and lf are stage and terminal costs accord
|
ing to Assumption 1 ψd
|
iand Ψd
|
ijare defined as in 18
|
Ψd
|
Nj d
|
AΦe
|
Nj d
|
BΦν
|
NjΞjZfis an RPI set ac
|
cording to Definition 4 and the resulting MPC control
|
law is given by
|
κMPCxk v
|
0 33
|
where v
|
0is the first element of the optimal solution v
|
of 32
|
41 Theoretical Guarantees
|
In the following we prove recursive feasibility and input
|
tostate stability ISS of the proposed filterbased
|
SLTMPC scheme 32 To do this we first discuss the
|
equivalence of 1 and 8 by considering dynamics 1
|
with the filterbased SLTMPC controller 33 The
|
auxiliary disturbance wk can be computed from the
|
states measured at timesteps k k1 the applied control
|
input and the disturbance filter variables p
|
0 σ
|
1as
|
wk 1
|
σ
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.