Datasets:
tensorkelechi
/
arxiv-cstext

Dataset card Data Studio Files Community
1
text
stringlengths
0
89.3k
0 v
N1p
p
0 p
N1 and λ λ
0 λ
M1denote the
optimal solution of 46 for initial state xk We then
use the candidate ˆλ0 1 ˆλm 0 m 1 M1 for
the convex combination variables which trivially fulfills
constraint 46h and construct the remaining candidate
sequences as
ˆziz
i1M1X
m0λ
mΦem
i10wk i0 N152a
ˆzNAKfz
NM1X
m0λ
mΓmwk 52bˆviv
i1M1X
m0λ
mΦνm
i10wk i0 N252c
ˆvN1Kfz
NM1X
m0λ
mΦνm
N0wk 52d
ˆpip
i1M1X
m0λ
mΣm
i20wk i0 N252e
ˆpN1M1X
m0λ
mΞm
0wk 52f
with ΓmAΦem
N0BΦνm
N0 Ξm
0 Next we show that
candidates 52 satisfy all constraints using the proof of
Theorem 3 Note that 52 are equivalent to the vector
valued candidates in 37 and the tubes stored in M0are
constructed equivalently to 38 apart from the convex
combination of the system responses and disturbance fil
ters However since the candidates and the tubes use
the same convex combination we can treat each contri
bution to the convex combination individually For ex
ample to show ˆ zi Z0
i i 0 N1 we get
M1X
m0λ
mz
i1 Φem
i10wkM1M
m0λ
mX F iΦem
shift
and we can show z
i1 Φem
i10wk X F iΦem
shift
for each mindividually As shown in the proof of Theo
rem 3 tubes 48 satisfy this inclusion by construction
The same argument holds for the other constraints and
thus we have shown that candidate 52 satisfies all con
straints in 46 which proves recursive feasibility 2
For robust stability we note that costs l lf in 46a
are only functions of the nominal trajectories and inde
pendent of the convex combination variables Therefore
the following stability result is a direct consequence of
Theorem 4 and Proposition 5
Corollary 6 Given that Assumption 1 holds system 1
subject to constraints 3and in closedloop with 47
is ISS in Xfeasfor any admissible sequence of combined
disturbances η where Xfeasis the set of all states xk
for which 46is feasible
Remark 5 In case heuristic FMkenters the cost
of46 eg as a regularizer ISS as in Corollary 6 cannot
be proven However we can prove inputtostate prac
tical stability ISpS 29 since the regularization term
will be constant with respect to the state and disturbance
6 Numerical Results
We first compare the recursively feasible SLTMPC 32
to the method presented in 25 on a two dimensional
14000102030405
u1D716u1D434080910Coverage of RCI
u1D4415
000102030405
u1D716u1D434
u1D44110Model uncertainty in u1D434
00010203040506
u1D70Eu1D464
u1D4415
00010203040506
u1D70Eu1D464
u1D44110