Datasets:
tensorkelechi
/
arxiv-cstext

Dataset card Data Studio Files Community
1
text
stringlengths
0
89.3k
lz
N Kfz
NΦν
N0wk
lz
N Kfz
N lz
N Kfz
N
βxΦe
N0wk βuΦν
N0wk
lz
N Kfz
N
and by combining the above we get the value function
decrease
V
Nxk1V
Nxk lz
0 v
0 lz
N Kfz
N
lfAKfz
Nlfz
N
βtotwk
where we collected all Kfunctions in
βtotwk βfΓwk
NX
i1βxΦe
i0wkβuΦν
i0wk
41Due to compactness of XandU the variables Φe
i0Φν
i0
and Γare bounded Therefore 41 is a sum of
compositions of Kfunctions eg βfΓwk
βfΓwk which we can rewrite as a single K
function βtot Finally we notice that lz
0 v
0
lxk κMPCxk γ1xk and lfAKfz
N
lfz
N lz
N Kfz
N due to Assumption 1 and
therefore obtain
V
Nxk1V
Nxk γ1xkβtotwk
Since βtot is aKfunction V
N fulfills decrease con
dition 40b and because V
Nxlx κMPCxx
Xfeas the optimal value function is lower bounded
by the Kfunction γ1 by assumption ie V
Nx
γ1xx X feas Using the facts that lf is upper
bounded the state and input constraints are compact
and monotonicity of V
N we can conclude that V
N
is upper bounded by a Kfunction γ2 inXfeasdue
to 3 Propositions 215 216 Therefore V
N also ful
fills 40a and is thus an ISSLyapunov function in Xfeas
Definition 5 Hence we have shown ISS of system 35
in closedloop with 33 which implies ISS of the origi
nal system 1 with SLTMPC controller 33 2
Remark 3 Contrary to 2932 we do not treat the com
ponents of the combined uncertainty ηdwsepa
rately where the state dependent model uncerainty dis
assumed to have a stability margin that is incorporated
inγ1x32 Remark 3 Instead we use 35 with
a single disturbance and the standard ISS proof tech
nique 3 Appendix B6 Intuitively we treat 35as
xAxh
B Ini
v
0
p
0
σ
1wk
in the proof where v
0p
0is an extended input and σ
1
is an optimized but bounded disturbance gain Also note
thatΦe
i0Φν
i0 and Γin41are optimization variables
that vary each time 32is solved Therefore they affect
the size of the region around the origin where the closed
loop system converges to Adding a regularizer on these
variables to 32a thus allows tuning of this convergence
region