Lyapunov Characterization for ISS of Impulsive Switched Systems

This paper establishes necessary and sufficient conditions for the input-to-state stability (ISS) of impulsive switched systems with both stable and unstable modes by introducing time-varying ISS-Lyapunov functions under relaxed mode-dependent average dwell and leave time constraints, while also providing methods to construct decreasing Lyapunov functions and guarantee ISS even with unknown switching signals.

The Gist

Here is an explanation of the paper "Lyapunov Characterization for ISS of Impulsive Switched Systems," translated into simple language with creative analogies.

The Big Picture: A Chaotic Journey

Imagine you are driving a car on a very strange road. This road has two types of terrain:

1. Smooth Hills (Stable Flows): When you drive here, the car naturally slows down and settles.
2. Bumpy Ramps (Unstable Flows): When you drive here, the car speeds up and gets wilder.

Now, imagine a mischievous driver (the Switching Signal) who constantly switches between these terrains. Sometimes they switch smoothly; other times, they hit a pothole that instantly launches the car into the air (Impulses/Jumps).

The Question: Can you guarantee that no matter how crazy the driver switches between smooth hills and bumpy ramps, the car will eventually stay under control and not crash?

This paper answers that question for complex systems (like robots, power grids, or autonomous vehicles) that behave this way.

---

The Core Concept: Input-to-State Stability (ISS)

In the real world, your car isn't just affected by the road; it's also affected by wind gusts (external disturbances).

• ISS (Input-to-State Stability) is a fancy way of saying: "Even if the wind blows hard and the driver switches roads randomly, the car will eventually slow down and stay within a safe distance from the center of the road."

The paper asks: How do we prove a system like this is safe?

The Old Way vs. The New Way

The Old Way (The "Strict Rulebook"):
Previous research said, "To be safe, the car must spend a fixed amount of time on the smooth hills before switching to the bumpy ramps."

• Problem: This is too rigid. Real life isn't that predictable. Sometimes you need to switch quickly; sometimes you need to wait longer. Also, previous rules only worked if all the roads were smooth, or if the jumps were always gentle.

The New Way (The "Smart Navigator"):
The authors of this paper created a smarter navigation system. They introduced two new tools:

1. Mode-Dependent Average Dwell Time (MDADT): Instead of a fixed timer, this is a "budget." It says, "You can spend time on the bumpy ramps, but only if you balance it out by spending enough time on the smooth hills later." It's like a diet: you can eat cake, but you have to run extra miles to balance it out.
2. Mode-Dependent Average Leave Time (MDALT): This is the reverse. It says, "If you are stuck on a bumpy ramp, you must leave it frequently enough to prevent disaster."

The Magic Tool: The "Energy Backpack" (Lyapunov Functions)

To prove the car is safe, mathematicians use a tool called a Lyapunov Function. Think of this as an Energy Backpack the car wears.

• The Goal: The backpack should get lighter (lose energy) over time. If the backpack gets lighter, the car is slowing down and stabilizing.
• The Problem: In this chaotic system, the backpack might get heavier when the car hits a bumpy ramp or jumps.

The paper introduces two types of backpacks:

1. The "Non-Decreasing" Backpack (The Flexible One)

This backpack is allowed to get heavier sometimes.

• How it works: It accepts that when you hit a bumpy ramp, the backpack gets heavy. But, as long as the "Smart Navigator" (the MDADT/MDALT rules) ensures you spend enough time on smooth hills later, the average weight over time goes down.
• The Result: If you can find this flexible backpack, the system is safe.

2. The "Decreasing" Backpack (The Strict One)

This backpack never gets heavier. It always shrinks, even when the car jumps.

• The Breakthrough: The paper proves that if the system is safe, a "Strict Backpack" must exist. This is a huge deal because it means the "Strict Backpack" is the ultimate proof of safety. It's a "necessary and sufficient" condition. If you can't find a Strict Backpack, the system is unsafe.

The "Translation" Trick

Here is the most creative part of the paper. The authors realized that finding a "Strict Backpack" is hard, but finding a "Flexible Backpack" is easier.

They developed a mathematical recipe to turn a "Flexible Backpack" into a "Strict Backpack."

• The Analogy: Imagine you have a flexible backpack that gets heavy sometimes. The authors show you how to add a "counter-weight" (a correction term) to the backpack. This counter-weight is calculated based on how long you've been on the smooth vs. bumpy roads.
• The Result: Even though the original backpack fluctuated, the new backpack (with the counter-weight) strictly shrinks every single second. This allows engineers to use the easier-to-find "Flexible" rules to prove the "Strict" safety guarantee.

Why This Matters (The "So What?")

1. It's More Realistic: It handles systems where some parts are naturally stable and others are naturally unstable.
2. It Handles Unknowns: The paper provides a way to guarantee safety even if you don't know exactly when the driver will switch roads, as long as they follow the general "budget" rules.
3. It's Constructive: For linear systems (like simple robots), they provide a checklist (called LMIs) that engineers can run on a computer. If the checklist passes, the robot is guaranteed to be stable, even with random switching and wind gusts.

Summary in One Sentence

This paper proves that for complex, switching systems with sudden jumps, you can guarantee safety by balancing "good times" and "bad times" using a flexible energy measurement, and then mathematically converting that flexible measurement into a strict, unshakeable proof of stability.

Technical Summary

Here is a detailed technical summary of the paper "Lyapunov Characterization for ISS of Impulsive Switched Systems" by Ahmed, Bachmann, and Trenn.

1. Problem Statement

The paper addresses the Input-to-State Stability (ISS) analysis of impulsive switched systems. These are hybrid dynamical systems characterized by:

• Continuous flows: Governed by differential equations $\dot{x} = f_{\sigma(t)}(t, x, u)$.
• Discrete jumps: Abrupt state changes governed by $x(t^+) = g_{\sigma(t^-)}(t^-, x(t^-), u(t^-))$.
• Switching: A signal $\sigma(t)$ selects which mode (flow and jump pair) is active.

Key Challenges:

• The system contains modes with both stable and unstable flows.
• Existing literature often provides only sufficient conditions for ISS, typically requiring all modes to be stable or imposing restrictive switching constraints (e.g., fixed dwell times).
• There is a lack of necessary and sufficient conditions (converse theorems) for ISS in this specific class of systems, particularly when dealing with time-varying dynamics and unknown switching signals.

2. Methodology

The authors employ Lyapunov-based methods tailored for hybrid systems. Their approach involves:

A. Switching Signal Constraints

Instead of simple dwell-time constraints, the paper introduces two mode-dependent constraints:

1. Mode-Dependent Average Dwell Time (MDADT): Applied to modes with stable flows ($S$). It limits the frequency of jumps to allow stable flows to dominate.
2. Mode-Dependent Average Leave Time (MDALT): Applied to modes with unstable flows ($U$). It ensures that unstable flows are not active for too long relative to the stabilizing jumps.

B. Lyapunov Function Classes

The authors define two types of time-varying ISS-Lyapunov functions:

1. Non-decreasing ISS-Lyapunov Functions: The Lyapunov value is allowed to increase along trajectories (both during flows and jumps) but is constrained by the switching conditions to ensure overall stability.
2. Decreasing ISS-Lyapunov Functions: A stricter subclass where the Lyapunov value strictly decreases along the solution trajectory.

C. Construction Technique

A novel mathematical technique is proposed to construct a decreasing ISS-Lyapunov function ($W_\sigma$) from a non-decreasing one ($V_\sigma$). This involves:

• Transforming the Lyapunov function using integral functions of the rate functions ($\Phi_p$).
• Introducing a correction term $h_\sigma(p, t)$ that accounts for the "overdue" or "excessive" switching relative to the MDADT/MDALT constraints.
• This construction bridges the gap between sufficient conditions (non-decreasing) and necessary/sufficient conditions (decreasing).

3. Key Contributions

1. Necessary and Sufficient Conditions:

   • The paper proves that the existence of a non-decreasing ISS-Lyapunov function is a sufficient condition for ISS.
   • It proves that the existence of a decreasing ISS-Lyapunov function is a necessary condition for ISS.
   • Conclusion: ISS is equivalent to the existence of either type of function, providing a complete Lyapunov characterization for impulsive switched systems.

2. Generalization of Switching Constraints:

   • The results apply to systems with infinite modes (unlike many prior works restricted to finite sets).
   • The constraints (MDADT/MDALT) are mode-dependent, allowing for broader classes of systems where different modes have different stability requirements.
   • The approach handles simultaneous instability of flows and jumps, which previous methods could not address.

3. Robustness to Unknown Switching:

   • The authors provide a method to guarantee ISS for systems with unknown switching signals (where only the set of allowed mode transitions is known).
   • This is achieved by defining time-invariant candidate functions that evolve beyond their active intervals, ensuring robustness against arbitrary switching sequences within the defined constraints.

4. Linear System Application (LMIs):

   • For linear impulsive switched systems, the authors derive a set of Linear Matrix Inequalities (LMIs).
   • The feasibility of these LMIs guarantees ISS for a class of systems with unknown switching signals, provided the switching satisfies the MDADT/MDALT conditions.

4. Key Results

• Theorem 9: Establishes that if a non-decreasing ISS-Lyapunov function exists satisfying specific inequalities involving MDADT/MDALT, the system is ISS.
• Theorem 13 (Converse Theorem): Proves that if a system is ISS, a decreasing ISS-Lyapunov function must exist. This is the first converse ISS-Lyapunov theorem provided for impulsive switched systems.
• Theorem 15: Provides the explicit formula for constructing the decreasing Lyapunov function $W_\sigma$ from a non-decreasing $V_\sigma$ using the correction term $h_\sigma$.
• Theorem 19: Presents the LMI conditions for linear systems. Specifically, it links the eigenvalues of the system matrices and jump matrices to the dwell/leave times via logarithmic inequalities (e.g., $\frac{\ln \mu_q}{|\eta_p|} \leq \tau_q(1-\delta)$ for stable modes).

5. Significance

• Theoretical Advancement: By establishing the equivalence between ISS and the existence of specific Lyapunov functions (both necessary and sufficient), the paper closes a theoretical gap in hybrid systems control.
• Broader Applicability: The relaxation of constraints (allowing unstable flows and infinite modes) makes the results applicable to more complex real-world systems like robotics, automotive control, and networked systems where instability is inherent in certain modes.
• Practical Utility: The construction method for decreasing Lyapunov functions is valuable because the level sets of such functions directly indicate reachable sets, which is crucial for safety verification.
• Computational Feasibility: The derivation of LMIs for linear systems allows engineers to computationally verify stability and design controllers for systems with uncertain switching patterns.

In summary, this work provides a comprehensive framework for analyzing and ensuring the stability of complex impulsive switched systems, moving beyond sufficient conditions to a complete characterization that accommodates unstable dynamics and uncertain switching.