introduction - disc-cps15.imtlucca.itdisc-cps15.imtlucca.it/pdf/heemels.pdf · hybrid systems...
TRANSCRIPT
Where innovation starts
Resource-aware control
Maurice Heemels
Zandvoort, June 2015
2/52
/w
Standard digital control loop
Actuator SensorPhysical System
Controller
−→ All control tasks executed periodically and triggered by time
Introduction
3/52
/w
Resource-aware control
• Resource-constrained control systems
– Computation time on embedded systems
– Actuator limitations (strain)
– Network utilization in NCS
– Battery power in WCS
• Time-triggered periodic control: Inefficient usage of resources
Introduction4/52
/w
Periodic or Aperiodic: That’s the question!• Paradigm shift: Periodic control −→ Aperiodic control
• Only act when needed: bringing feedback in resource utilization
Actuator SensorPhysical System
Controller
−→
Actuator SensorPhysical System
Controller
Introduction
5/52
/w
Paradigm shift: Periodic control −→ Aperiodic control
• Event-triggered control:
u(t) = K(x(tk)), when t ∈ [tk, tk+1)
tk+1 = inf{t > tk | C(x(t), x(tk)) > 0}
Actuator SensorPhysical System
Controller
[1] Arzen, IFAC WC’99 [2] Astrom & Bernhardsson, IFAC WC’99 [3] Heemels et al, CEP’99
Introduction5/52
/w
Paradigm shift: Periodic control −→ Aperiodic control
• Event-triggered control:
u(t) = K(x(tk)), when t ∈ [tk, tk+1)
tk+1 = inf{t > tk | C(x(t), x(tk)) > 0}
Actuator SensorPhysical System
Controller
• Example event-triggering condition
‖x(t)− x(tk)‖ > σ‖x(t)‖
Introduction
6/52
/w
Paradigm shift: Periodic control −→ Aperiodic control
• Event-triggered control: reactive
u(t) = K(x(tk)), when t ∈ [tk, tk+1)
tk+1 = inf{t > tk | C(x(t), x(tk)) > 0}
• Self-triggered control: proactive
u(t) = K(x(tk)), when t ∈ [tk, tk+1)
tk+1 = tk +M(x(tk))
Introduction7/52
/w
• Basic setup state-feedback ETC: ‖x(t)− x(tk)‖ > σ‖x(t)‖• Hybrid systems
• Challenges
– Performance/Robustness w.r.t. disturbances & Zeno-freeness– Output-based (& Decentralized)
• Alternative event-triggered controllers
– Relative, absolute and mixed event generators– Periodic event-triggered control– Time regularisation– Dynamic event generators
• Application to vehicle platooning
• Conclusions & What’s next?
Outline
8/52
/w
• Linear system
x(t) = Ax(t) + Bu(t)
• Linear state feedback
u(t) = Kx(t), t ∈ R>0
Actuator SensorPhysical System
Controller
• Ideal loop: x(t) = (A + BK)x(t)
• Sampled-data control with execution times tk, k ∈ N (ZOH)
u(t) = Kx(t) = Kx(tk), t ∈ [tk, tk+1)
• Perturbation perspective: implementation-induced error
e(t) = x(tk)− x(t) for t ∈ [tk, tk+1)
x(t) = Ax(t) + BKx(tk) = (A + BK)x(t) + BKe(t)
[1] Tabuada, Event-triggered real-time scheduling of stabilizing control tasks, TAC 2007
Basic ETC setup9/52
/w
• Perturbation perspective:
x(t) = Ax(t) + BKx(tk) = (A + BK)x(t) + BKe(t)
• SinceA + BK Hurwitz, quadratic Lyapunov function V (x) = x>Px s.t.
d
dtV 6 −a2‖x(t)‖2 + ‖e(t)‖2
• Crux: Guarantee ‖e(t)‖ 6 ρa · ‖x(t)‖ with 0 < ρ < 1 s.t.
d
dtV 6 −a2‖x(t)‖2 + ‖e(t)‖2 6 −(1− ρ2)a2‖x(t)‖2
• Guarantee for Global Exponential Stability
tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > ρa · ‖x(t)‖}
Basic ETC setup
10/52
/w
• Summary of event-triggered setup:
– Linear systemx(t) = Ax(t) + Bu(t)
– Execution times tk, k ∈ N
tk+1 = inf{t > tk | ‖x(tk)− x(t)‖ > ρa · ‖x(t)‖}– Control law:
u(t) = Kx(tk), t ∈ [tk, tk+1)
• Global exponential stability (GES)
Event-triggered control11/52
/w
• Summary of event-triggered setup:
– Linear systemx(t) = Ax(t) + Bu(t)
– Execution times tk, k ∈ N
tk+1 = inf{t > tk | ‖x(tk)− x(t)‖ > ρa · ‖x(t)‖}– Control law:
u(t) = Kx(tk), t ∈ [tk, tk+1)
• Global exponential stability (GES)
• Question: Which important issue should we still verify?
Event-triggered control
11/52
/w
• Linear system
x(t) = Ax(t) + Bu(t)
• Linear state feedback (ZOH)
u(t) = Kx(tk), t ∈ [tk, tk+1)
Actuator SensorPhysical System
Controller
• Execution times: tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > σ‖x(t)‖}• Properties established in [1]:
– Global exponential stability (GES) when σ suff. small– Global positive lower bound on minimal inter-event time (MIET)
inf{tk+1 − tk | k ∈ N} > τmin > 0
• Improved designs for GES/L∞-gain via hybrid system analysis [2]
[1] Tabuada, Event-triggered real-time scheduling of stabilizing control tasks, TAC 2007[2] Donkers, Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain ..., TAC 2012
Basic ETC setup12/52
/w
• Perturbation perspective:
x(t) = Ax(t) + BKx(tk) = (A + BK)x(t) + BKe(t)
• Execution times tk, k ∈ N
tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > σ‖x(t)‖}
• Hybrid system perspective [1,2] based on jump-flow models [3]:
d
dt
[xe
]=
[(A + BK)x + BKe−(A + BK)x−BKe
], when ‖e‖2 6 σ2‖x‖2
[x+
e+
]=
[x0
], when ‖e‖2 > σ2‖x‖2
[1] Donkers, Heemels, Output-Based Event-Triggered Control ..., TAC 2012 & CDC 2010[2] Postoyan, Anta, Nesic, Tabuada, A unifying Lyapunov-based framework ..., CDC-ECC 2011[3] Goebel, Sanfelice, Teel, Hybrid Dynamical Systems, Princeton, 2012.
Hybrid systems (side trip)
13/52
/w
Hybrid system perspective (side trip)
d
dt
[xe
]=
[(A + BK)x + BKe−(A + BK)x−BKe
]when ‖e‖2 6 σ2‖x‖2
[x+
e+
]=
[x0
]when ‖e‖2 > σ2‖x‖2
or compactly with ξ =
[xe
] {ξ = Φξ, when ξ>Qξ 6 0
ξ+ = Jξ, when ξ>Qξ > 0
ETC based on feedback14/52
/w
Hybrid system perspective (side trip)
ξ = Φξ when ξ>Qξ 6 0
ξ+ = Jξ when ξ>Qξ > 0
• Stability analysis using hybrid tools [1,2]: V (ξ) = ξ>Pξ
– ddtV (ξ) < 0 when ξ>Qξ 6 0
– V (Jξ) 6 V (ξ) when ξ>Qξ > 0
• Linear matrix inequalities: if there are α, β > 0 s.t.
– Φ>P + PΦ− αQ ≺ 0
– J>PJ − P + βQ � 0
• Guarantee for GES (extended ideas apply for L∞-gains)
• Never more conservative than perturbation approach [1]
[1] Donkers, Heemels, Output-based event-triggered control with guaranteed L∞-gain, TAC 2012 & CDC 2010[2] Goebel, Sanfelice, Teel, Hybrid Dynamical Systems: Modeling, Stability and Robustness, Princeton, 2012.
ETC based on feedback
15/52
/w
Example 1: State feedback control
• Consider x =[
0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖ MIET = 0.025
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTC
Illustrative Example15/52
/w
Example 1: State feedback control
• Consider x =[
0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖ MIET = 0.025
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTC
Illustrative Example
15/52
/w
Example 1: State feedback control
• Consider x =[
0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖ MIET = 0.025
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTCETC
Illustrative Example15/52
/w
Example 1: State feedback control
• Consider x =[
0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖ MIET = 0.025
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 140
200
400
600
time t
number
ofevents
TTCETC
Illustrative Example
16/52
/w
Example 1: Comparison P and HS approach• Consider x =
[0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• Example taken from [1]
• We look for largest σ giving GES: ‖e‖2 6 σ2‖x‖2 [2]
σ2 MIETP: Results from [1] 0.0030 0.0318P: By minimising the L2-gain 0.0273 0.0840Hybrid System 0.0588 0.1136
• PS: via minimising L2-gain: maximise a (note σ = ρa)
V 6 −a2‖x(t)‖2 + ‖e(t)‖2 for x = (A + BK)x + BKe
• ETM:tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸
=e(t)
‖ > ρa · ‖x(t)‖}
[1] Tabuada, TAC ’07 [2] Donkers, Heemels, CDC10 & TAC12
Illustrative Examples17/52
/w
• Linear system
x(t) = Ax(t) + Bu(t)
• Linear state feedback (ZOH)
u(t) = Kx(tk), t ∈ [tk, tk+1)
Actuator SensorPhysical System
Controller
• Execution times: tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > σ‖x(t)‖}• Properties established in [1]:
– Global exponential stability (GES) when σ suff. small– Global positive lower bound on minimal inter-event time (MIET)
inf{tk+1 − tk | k ∈ N} > τmin > 0
• Improved designs for GES/L∞-gain via hybrid system analysis [2]
[1] Tabuada, Event-triggered real-time scheduling of stabilizing control tasks, TAC 2007[2] Donkers, Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain ..., TAC 2012
Summary
18/52
/w
• Performance/Robustness w.r.t. disturbances
• Output-based (& Decentralized)
Challenges19/52
/w
Illustrative example
• Consider x =[
0 1−2 3
]x +
[01
]u and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTCETC
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC
20/52
/w
Illustrative example
• Consider x =[
0 1−2 3
]x +
[01
]u + w and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTCETC
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC21/52
/w
Illustrative example
• Consider x =[
0 1−2 3
]x +
[01
]u + w and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTCETC
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC
21/52
/w
Illustrative example
• Consider x =[
0 1−2 3
]x +
[01
]u + w and u(t) = [1 −4]x(tk)
• TTC: tk = k · 0.025
• ETC: tk = t ⇐⇒ ‖e(t)‖ > 0.05‖x(t)‖
0 2 4 6 8 10 12 140
0.5
1
1.5
time t
‖x(t)‖
TTCETC
0 2 4 6 8 10 12 1410−4
10−3
10−2
10−1
100
time t
inter-eventtimeτ i
TTCETC
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC22/52
/w
Illustrative example
Actuator SensorPhysical System
Controller
• Consider{xp =
[1 −110 −1
]xp +
[11
]u
y = [1 0 ]xpu(t) = −2y(tk)
• ETM: ‖y(t)− y(tk)‖2 > σ2‖y(t)‖2
• Parameter: σ2 = 0.5
Output-based ETC
23/52
/w
Illustrative example
• Minimal inter-event time (MIET) is zero! (Zeno behavior)
Donkers, Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain and Improved
and Decentralised Event-Triggering, TAC 2012
Output-based ETC24/52
/w
• Relative: ‖y − y‖ > σ‖y‖ [1]
• Absolute: ‖y − y‖ > δ [2-4]
• Mixed: ‖y − y‖ > σ‖y‖ + δ [5]
[1] Tabuada, Event-triggered real-time scheduling of stabilizing control tasks, TAC 2007[2] Yook, Tilbury, Soparkar, Trading computation for bandwidth: Reducing communication in
distributed control systems using state estimators, TCST 2002[3] Miskowicz, Send-on-delta concept: An event-based data-reporting strategy, Sensors, 2006[4] Lunze and Lehmann, A state-feedback approach to event-based control, Automatica, 2010[5] Donkers, Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain ..., TAC 2012
Event-triggered control schemes
25/52
/w
Inverted pendulum
Movie ETC in action26/52
/w
Event-separation properties / Zeno-freeness
• Consider x = Ax + Bu + w and u(t) = Kx(tk) = K(x(t) + e(t))
• Execution times:
tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > σ‖x(t)‖ + δ}
→MIET τ (x0, w) dependent on x0 and w: τ (x0, w) = infk∈N (tk+1 − tk)
• Event-separation properties (nominal)
– Global ESP: infx0∈Rn τ (x0, 0) > 0
– Semi-global ESP: for compact X0 ⊂ Rn: infx0∈X0 τ (x0, 0) > 0
– Local ESP: for each x0 ∈ Rn: τ (x0, 0) > 0
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC
26/52
/w
Event-separation properties / Zeno-freeness
• Consider x = Ax + Bu + w and u(t) = Kx(tk) = K(x(t) + e(t))
• Execution times:
tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > σ‖x(t)‖ + δ}
→MIET τ (x0, w) dependent on x0 and w: τ (x0, w) = infk∈N (tk+1 − tk)
• Event-separation properties (robust): there is ε > 0
– Robust global: infx0∈Rn, ‖w‖∞<ε τ (x0, w) > 0
– Robust semi-global: compact X0: infx0∈X0, ‖w‖∞<ε τ (x0, w)> 0
– Robust local: for each x0 ∈ Rn and ‖w‖∞ < ε: τ (x0, w) > 0
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Disturbances in ETC27/52
/w
State-feedback caseETM robust global global robust semi-global semi-global robust local localrelative × X × X × Xabsolute × × X X X Xmixed X X X X X X
Output-feedback case
ETM robust global global robust semi-global semi-global robust local localrelative × × × × × ×absolute × × X X X Xmixed × × X X X X
• Relative triggering fragile. Zero robustness
• Mixed or absolute effective (semi-global)
• However, only practical stability / ultimate boundedness (no GAS)
• Challenge: What about robust global ESP and GAS/L2-gains?
Borgers, Heemels, Event-Separation Properties of Event-Triggered Control Systems, TAC 2014
Overview
28/52
/w
P
ETM
C
ZOH
uwz
x
x
• Guaranteed control performance (L2−gain) from disturbance w tooutput z = q(x,w):
‖z‖L2 6 β(|ξ0|) + γ‖w‖L2 with ‖z‖L2 =
√∫ ∞
0
‖z(t)‖2dt
• Global asymptotic stability (GAS) in absence of disturbances
• Robust positive “minimal inter-event time” (τmiet)
• Reduced communication w.r.t. time-triggered control
Objectives29/52
/w
Cooperative Adaptive Cruise Control
WiFi-p: Feedforward
Radar: Feedback
• String stability: disturbance attenuation along the vehicle stringγ 6 1
‖z‖L2 6 β(|ξ0|) + γ‖w‖L2 with ‖z‖L2 =
√∫ ∞
0
‖z(t)‖2dt
• Communication resources limited→ event-triggered communica-tion
Motivation
30/52
/w
Time regularisation:• Periodic Event-Triggered Control (PETC) [6-9]
tk+1 = inf{t > tk | ‖y(t)− y(t)‖ > σ‖y(t)‖ ∧ t = kh, k ∈ N}
• Enforcing minimal inter-event time [7,9-11]
tk+1 = inf{t > tk+T | ‖y(t)− y(t)‖ > σ‖y(t)‖}
[6] Arzen, A simple event-based PID controller, IFAC 1999[7] Heemels, Sandee, van den Bosch, Analysis of event-driven controllers for linear systems, IJC 2008[8] Heemels, Donkers, Teel, Periodic Event-Triggered Control for Linear Systems, TAC 2013[9] Henningsson, Johannesson, Cervin, Sporadic event-based control of first-order linear stochastic .., Aut. 2008[10] Tallapragada, Chopra, Event-triggered dynamic output feedback control for LTI systems, CDC 2012[11] Tallapragada, Chopra, Event-triggered decentralized dynamic output .. LTI systems, NECSYS 2012
Event-triggered control schemes30/52
/w
• Periodic Event-Triggered Control (PETC)
tk+1 = inf{t > tk | ‖y − y‖ > σ‖y‖ ∧ t = kh, k ∈ N}
• Enforcing minimal inter-event time
tk+1 = inf{t > tk+T | ‖y − y‖ > σ‖y‖}
Output-feedback case
ETM robust global global robust semi-global semi-global robust local localrelative × × × × × ×absolute × × X X X Xmixed × × X X X Xtime-regu X X X X X X
Time regularized ETC
30/52
/w
• Periodic Event-Triggered Control (PETC)
tk+1 = inf{t > tk | ‖y − y‖ > σ‖y‖ ∧ t = kh, k ∈ N}
• Hybrid system analysis: GAS & finite L2-gains [1,2]
• Implementation advantages:
– Guaranteed (reasonable) minimal inter-event time
– Only time-periodic verification of event-triggering conditions
– More in line with time-sliced architectures
[1] Heemels, Donkers, Teel, Periodic Event-Triggered Control for Linear Systems, TAC 2013[2] Heemels, Donkers, Model-based Periodic Event-Triggered Control for Linear Systems, Automatica 2013
PETC31/52
/w
Hybrid systems formulation
d
dt
[ξτ
]=
[Aξ +Bw
1
], when τ ∈ [0, h],
[ξ+
τ+
]=
[J1ξ
0
], when ξ>Qξ > 0, τ = h
[J2ξ
0
], when ξ>Qξ 6 0, τ = h
z = Cξ +Dw
PETC
32/52
/w
Hybrid systems formulation
d
dt
[ξτ
]=
[Aξ +Bw
1
], when τ ∈ [0, h],
[ξ+
τ+
]=
[J1ξ
0
], when ξ>Qξ > 0, τ = h
[J2ξ
0
], when ξ>Qξ 6 0, τ = h
z = Cξ +Dw
• In case w = 0 and interested in stability only
• Discretize at kh, k ∈ N (just before jump) leading to discrete-time PWL system[1,2,3]
ξk+1 =
{eAhJ1ξk, when ξ>k Qξk > 0
eAhJ2ξk, when ξ>k Qξk 6 0
[1] Heemels, Donkers, Teel, Periodic Event-Triggered Control for Linear Systems, TAC 2013[2] Heemels, Donkers, Model-based Periodic Event-Triggered Control for Linear Systems, Automatica 2013[3] Heemels, Sandee, van den Bosch, Analysis of event-driven controllers for linear systems, IJC 2008
PETC33/52
/w
Hybrid systems formulation• Including intersample-behavior, e.g., for L2-gain analysis
d
dt
[ξ
τ
]=
[Aξ + Bw
1
], when τ ∈ [0, h],
[ξ+
τ+
]=
[J1ξ
0
], when ξ>Qξ > 0, τ = h
[J2ξ
0
], when ξ>Qξ 6 0, τ = h
z = Cξ + Dw
PETC
33/52
/w
Hybrid systems formulation
d
dt
[ξ
τ
]=
[Aξ + Bw
1
], when τ ∈ [0, h]
[ξ+
τ+
]=
[φ(ξ)
0
], when τ = h
z = Cξ + Dw
PETC33/52
/w
Hybrid systems formulation
d
dt
[ξ
τ
]=
[Aξ + Bw
1
], when τ ∈ [0, h]
[ξ+
τ+
]=
[φ(ξ)
0
], when τ = h
z = Cξ + Dw
• L2-contractive: There are γ0 ∈ [0, 1) and a K-function β s.t.
‖z‖L2 6 β(|ξ0|) + γ0‖w‖L2 with ‖z‖L2 =
√∫ ∞
0
‖z(t)‖2dt
PETC
34/52
/w
d
dt
[ξ
τ
]=
[Aξ + Bw
1
], when τ ∈ [0, h]
[ξ+
τ+
]=
[φ(ξ)
0
], when τ = h
z = Cξ + Dw
ξk+1 = Adφ(ξk) + Bdvk
rk = Cdφ(ξk)
Main result: The hybrid system is internally stable and L2-contractiveiff the discrete-time nonlinear system is internally stable and `2-contractive.
• `2-contractive: there is γ0 ∈ [0, 1) s.t.
‖r‖`2 6 β(|ξ0|) + γ0‖v‖`2 with ‖r‖2`2 =
∞∑
k=0
|rk|2
[1] Heemels, Dullerud, Teel, L2-gain Analysis for a Class of Hybrid Systems with Applications to Reset andEvent-triggered Control: A Lifting Approach
Lifting-based approach34/52
/w
d
dt
[ξ
τ
]=
[Aξ + Bw
1
], when τ ∈ [0, h]
[ξ+
τ+
]=
[φ(ξ)
0
], when τ = h
z = Cξ + Dw
ξk+1 = Adφ(ξk) + Bdvk
rk = Cdφ(ξk)
Main result: The hybrid system is internally stable and L2-contractiveiff the discrete-time nonlinear system is internally stable and `2-contractive.
• Lifting with verifiable conditions without linearity
• For PETC piecewise linear system −→ contractivity/stability anal-ysis via LMIs using piecewise quadratic Lyapunov functions
[1] Heemels, Dullerud, Teel, L2-gain Analysis for a Class of Hybrid Systems with Applications to Reset andEvent-triggered Control: A Lifting Approach
Lifting-based approach
35/52
/w
P
ETM
C
ZOH
ua = uw
x
xc
P : x =
[0 1
−2 3
]x +
[0
1
]u + w
C : u =[
1 −4]xc
• Periodic Event-Triggered Control (PETC)
tk+1 = inf{t > tk | ‖xc(t)− x(t)‖ > σ‖x(t)‖ ∧ t = kh, k ∈ N}
• Enforcing minimal inter-event time
tk+1 = inf{t > tk+T | ‖xc(t)− x(t)‖ > σ‖x(t)‖}
0 2 4 6 8 10 12 14 16 18 200
0.5
1
1.5
time t
‖x(t)‖
time reg.PETC
0 2 4 6 8 10 12 14 16 18 2010−2
10−1
100
time t
inter-eventtimeτk
time reg.PETC
Time regularisation: Example36/52
/w
• Static event generator: tk+1 := inf{t > tk + T |C(x(t), e(t)) > 0}
Dynamic event generator [1,2,3]
η = Ψ(x, e, η)
tk+1 := inf{t > tk + T | η(t) < 0}
• How to find Ψ and T?
[1] Postoyan et al., “Event-triggered and self-triggered stabilization ...,” CDC 2011[2] Girard, “Dynamic triggering mechanisms for event-triggered control,” TAC 2015[3] Dolk, Borgers, Heemels, “Dynamic Event-triggered Control...,” CDC 2014
Dynamic event-triggered control
37/52
/w
• Perturbation perspective:
x(t) = Ax(t) + BKx(tk) = (A + BK)x(t) + BKe(t)
• SinceA + BK Hurwitz, quadratic Lyapunov function V (x) = x>Px
d
dtV 6 −a2‖x(t)‖2 + ‖e(t)‖2
• Crux: Guarantee ‖e(t)‖ 6 ρa · ‖x(t)‖ with 0 < ρ < 1 s.t.
d
dtV 6 −a2‖x(t)‖2 + ‖e(t)‖2 6 −(1− ρ2)a2‖x(t)‖2
• Guarantee for Global Exponential Stability
tk+1 = inf{t > tk | ‖x(tk)− x(t)︸ ︷︷ ︸=e(t)
‖ > ρa · ‖x(t)‖}
• Zeno-free: There is T > 0 such that tk+1 − tk > T for all k ∈ N.
Recap: Design relative triggering38/52
/w
• SinceA + BK Hurwitz, quadratic Lyapunov function V (x) = x>Px
d
dtV 6 −a2‖x(t)‖2 + ‖e(t)‖2
• Now consider η = Ψ(x, e, η) and LF U(x, η) = V (x) + η [2] :
d
dtU 6 −a2‖x‖2 + ‖e‖2 + Ψ
• To get ddtU 6 −(1− ρ2)a2‖x‖2 − εη for some ε > 0 we require
−a2‖x‖2 + ‖e‖2 + Ψ = −(1− ρ2)a2‖x‖2 − εηand thus η = Ψ(x, e, η) = ρ2a2‖x‖2 − εη − ‖e‖2• Now tk+1 := inf{t > tk + T | η(t) < 0}, η(0) = 0 and t0 = 0 :
– η(t) > 0 for t ∈ R>0 and thus U positive definite
– ddtU 6 −(1− ρ2)a2‖x‖2 − εη and thus GES
• Never triggers before the static version!!
Basic design dETM
39/52
/w
• Static event generator: tk+1 := inf{t > tk + T |C(x(t), e(t)) > 0}
Dynamic event generator [1,2,3]
η = Ψ(x, e, η)
tk+1 := inf{t > tk + T | η(t) < 0}
• [1,2] design for w = 0 (no disturbances)
• Recently, [3] new design methodology for output-based decentral-ized triggering under disturbances (Lp-gain)
[1] Postoyan et al., “Event-triggered and self-triggered stabilization ...,” CDC 2011[2] Girard, “Dynamic triggering mechanisms for event-triggered control,” TAC 2015[3] Dolk, Borgers, Heemels, “Dynamic Event-triggered Control...,” CDC 2014 and TAC?
Dynamic event-triggered control40/52
/w
P
ETM
C
ZOH
uwz
x
x
P : x =
[0 1
−2 3
]x +
[0
1
]u + w
C : u =[
1 −4]xc
Case study: L2-gain θ = 4 from input w to state x: τmiet = 9.1 · 10−3
• Dynamic event generator tk+1 := inf{t > tk + τmiet | η(t) < 0}• Static event generator: tk+1 := inf{t > tk + τmiet | Ψ(x, e, τ, η) < 0}
Dynamic ETC: Example
41/52
/w
P
ETM
C
ZOH
uwz
x
x
P : x =
[0 1
−2 3
]x +
[0
1
]u + w
C : u =[
1 −4]xc
Case study: L2-gain θ = 4 from input w to state x: τmiet = 9.1 · 10−3
• Dynamic event generator tk+1 := inf{t > tk + τmiet | η(t) < 0}• Static event generator: tk+1 := inf{t > tk + τmiet | Ψ(x, e, τ, η) < 0}
0 2 4 6 8 10 12 14 16 18 200
0.5
1
1.5
time t
|x(t)|
static ETMdynamic ETM
0 2 4 6 8 10 12 14 16 18 20
10−2
10−1
100
time t
inter-eventtimeτk
staticdynamic
Dynamic ETC: Example42/52
/w
0 1 2 3 4 5 6 7 8
·10−2
2
4
6
8
10
τmati / τmiet / τavg
L 2-gainθ
τmatiτmietτavg,staticτavg,dynamic
Dynamic ETC: Example
43/52
/w
Cooperative Adaptive Cruise Control
WiFi-p: Feedforward
Radar: Feedback
• String stability: disturbance attenuation along the vehicle string
– Lp-gain6 1
• Communication resources limited→ event-triggered communica-tion
Motivation44/52
/w
45/52
/w
• Headway time: 0.6 seconds
• MIET: 0.07 seconds
−→MOVIE
CACC46/52
/w
CACC
47/52
/w
CACC48/52
/w
• Event-triggered control: A new resource-aware control paradigm
• Several ETC algorithms discussed with their own tools (hybrid)
• Challenges
– Performance / Robustness w.r.t. disturbances– Output-based & decentralized event generators– Constrained systems (MPC)– Implementation and Applications– Better than periodic time-triggered control– Improved analysis and design tools: MIET, average inter-execution
times, Lp-gain, etc.
• Many interesting practical and theoretical issues open in this appealingresearch field
• More info: http://www.heemels.tue.nl
Conclusions
49/52
/w
• Collaborators
– Duarte Antunes, Niek Borgers, Florian Brunner, Victor Dolk, TijsDonkers, Tom Gommans, Heico Sandee, ...
– Frank Allgöwer, Adolfo Anta, Geir Dullerud, Kalle Johansson,Dragan Nesic, Romain Postoyan, Paulo Tabuada, Andy Teel,Paul van den Bosch, ...
• Financial support
• More info: http://www.heemels.tue.nl
Acknowledgements50/52
/w
• Arzen, A simple event-based PID controller, IFAC World Congress, 1999.
• Astrom, Bernhardsson Comparison of periodic and event based sampling for first order stochastic systems,IFAC World Congress 1999
• D.P. Borgers and W.P.M.H. Heemels, Event-separation properties of event-triggered control systems, IEEETransactions on Automatic Control, 59(10), p. 2644-2656, 2014.
• V.S. Dolk, D.P. Borgers, W.P.M.H. Heemels, Dynamic Event-triggered Control: Tradeoffs Between Transmis-sion Intervals and Performance, IEEE Conference on Decision and Control (CDC), pp. 2764-2769, 2014.
• V.S. Dolk, D.P. Borgers, W.P.M.H. Heemels, Output-based and Decentralized Dynamic Event-triggered Controlwith Guaranteed Lp-gain Performance and Zeno-freeness, prov. accepted in IEEE Transactions on AutomaticControl, 2015.
• M.C.F. Donkers and W.P.M.H. Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain andImproved and Decentralised Event-Triggering, IEEE Transactions on Automatic Control, 57(6), p. 1362-1376,2012.
• M.C.F. Donkers and W.P.M.H. Heemels, Output-Based Event-Triggered Control with Guaranteed L∞-gain andImproved Event-Triggering, IEEE Conference on Decision and Control (CDC) 2010, Atlanta, USA, p. 3246-3251.
• A. Girard, Dynamic triggering mechanisms for event-triggered control, IEEE Transactions on Automatic Con-trol, To appear, 2015.
• Goebel, Sanfelice, Teel, Hybrid Dynamical Systems, Princeton, 2012.
• W.P.M.H. Heemels, M.C.F. Donkers, Model-based periodic event-triggered control for linear systems, Auto-matica 49(3), pp. 698-711, 2013.
• W.P.M.H. Heemels, M.C.F. Donkers, and A.R. Teel, Periodic Event-Triggered Control for Linear Systems, IEEETransactions on Automatic Control , 58(4), p. 847-861, 2013.
Literature
51/52
/w
• W.P.M.H. Heemels, G. Dullerud, A.R. Teel, L2-gain Analysis for a Class of Hybrid Systems with Applicationsto Reset and Event-triggered Control: A Lifting Approach, prov. accepted IEEE Transactions on AutomaticControl.
• W.P.M.H. Heemels, R.J.A. Gorter, A. van Zijl, P.P.J. v.d. Bosch, S. Weiland, W.H.A. Hendrix, M.R. Vonder, Asyn-chronous measurement and control: a case study on motor synchronisation, Control Engineering Practice,7(12), 1467-1482, (1999)
• W.P.M.H. Heemels, J.H. Sandee, P.P.J. van den Bosch, Analysis of event-driven controllers for linear systems,International Journal of Control, 81(4), pp. 571-590 (2008).
• Henningsson T, Johannesson E, Cervin A, Sporadic event-based control of first-order linear stochastic sys-tems, Automatica 44, pp. 2890-2895, 2008.
• M. Miskowicz, Send-on-delta concept: An event-based data-reporting strategy, Sensors 6, pp. 49-63, 2006.
• J. Lunze and D. Lehmann, A state-feedback approach to event-based control, Automatica 46, pp. 211-215,2010.
• R. Postoyan, A. Anta, D. Nesic and P. Tabuada, A unifying Lyapunov-based framework for the event-triggeredcontrol of nonlinear systems, CDC (IEEE Conference on Decision and Control), pp 2559?2564, 2011.
• R. Postoyan, P. Tabuada, D. Nesic and A. Anta, Event-triggered and self-triggered stabilization of distributednetworked control systems, CDC (IEEE Conference on Decision and Control), 2011.
• P. Tabuada, Event-triggered real-time scheduling of stabilizing control tasks, IEEE Trans. Autom. Control,vol. 52, no. 9, pp. 1680-1685, 2007.
Literature52/52
/w
• Tallapragada P, Chopra N, Event-triggered decentralized dynamic output feedback control for LTI systems,IFAC workshop on distributed estimation and control in networked systems, pp 31-36, 2012.
• Tallapragada P, Chopra N, Event-triggered dynamic output feedback control for LTI systems, IEEE 51st annualconference on decision and control (CDC), Maui, pp 6597-6602, 2012.
• J.K. Yook and D.M. Tilbury and N.R. Soparkar, Trading Computation for Bandwidth: Reducing Communicationin Distributed Control Systems Using State Estimators, IEEE Trans. Control Systems Technology, 10(4), pp.503-518, 2002.
Recent overviews:
• W.P.M.H. Heemels, K.H. Johansson, and P. Tabuada, An introduction to event-triggered and self-triggeredcontrol, 51st IEEE Conference on Decision and Control 2012, Hawaii, USA, p. 3270-3285
• W.P.M.H. Heemels, K.H. Johansson, and P. Tabuada, Event-Triggered and Self-Triggered Control, Encyclope-dia of Systems and Control, Springer-Verlag London 2014.
Pointers for “better than periodic time-triggered control:”
• D. Antunes and W.P.M.H. Heemels, Rollout Event-Triggered Control: Beyond Periodic Control Performance,IEEE Transactions on Automatic Control 59(12), p. 3296-3311, 2014.
• Astrom, Bernhardsson Comparison of periodic and event based sampling for first order stochastic systems,IFAC World Congress 1999
• T.M.P. Gommans, D. Antunes, M.C.F. Donkers, P. Tabuada, W.P.M.H. Heemels, Self-Triggered Linear QuadraticControl, Automatica 50(4), p. 1279-1287, 2014.
−→More info: http://www.heemels.tue.nl
Literature