Singular Perturbation: When the Perturbation Parameter Becomes a State-Dependent Function

Tengfei Liu,Zhong-Ping Jiang

2024-06-02

Abstract:This paper presents a new systematic framework for nonlinear singularly perturbed systems in which state-dependent perturbation functions are used instead of constant perturbation coefficients. Under this framework, general results are obtained for the global robust stability and input-to-state stability of nonlinear singularly perturbed systems. Interestingly, the proposed methodology provides innovative solutions beyond traditional singular perturbation theory for emerging control problems arising from nonlinear integral control, feedback optimization, and formation-based source seeking.

Optimization and Control,Systems and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to introduce state - dependent perturbation functions in nonlinear singularly perturbed systems to replace the traditional constant perturbation coefficients, thereby proposing a new system framework. Specifically, the goals of the paper include:

1. **Global Robust Stability and Input - State Stability**: Under the new framework, obtain general results on the global robust stability and input - state stability (ISS) of nonlinear singularly perturbed systems.
2. **Solutions to Emerging Control Problems**: Provide innovative methods to solve the challenges posed by emerging control problems such as nonlinear integral control, feedback optimization, and formation - based source localization.
3. **Combination of Theory and Application**: By introducing state - dependent perturbation functions, expand the application scope of the traditional singular perturbation theory, especially provide useful tools in the controller design of complex systems.

### Main Contributions of the Paper

1. **New System Framework**:
- A generalized form of singularly perturbed system is proposed, in which the perturbation parameters are replaced by state - dependent functions \(\rho_s\) and \(\rho_f\).
- The form of the system is:
\[
\dot{x}=\rho_s(x, z, d)g_s(x, z, d)
\]
\[
\dot{z}=\rho_f(z, x, w)g_f(z, x, w)
\]
- This generalized form allows more flexible adjustment of the time scale of the subsystems, thereby improving the performance of the system.

2. **Stability Analysis**:
- Assume that the boundary - layer subsystem (i.e., the \(z\) - subsystem) is input - state stable for the equilibrium point \(z_e\) and input \(w\).
- Assume that the reduced - order subsystem (i.e., the \(x\) - subsystem) is input - state stable for input \(d\) and error state \(z - z_e\).
- Use Lyapunov - based ISS gains to describe the interaction between the reduced - order subsystem and the boundary - layer subsystem, and derive the monotonicity conditions of the perturbation functions to ensure the ISS property of the entire singularly perturbed system.

3. **Application Examples**:
- **Nonlinear Integral Feedback**: By introducing a nonlinear integral control algorithm, the problem that the traditional integral negative feedback cannot guarantee system stability in some cases is solved.
- **Feedback Optimization**: A gradient - flow algorithm for variable - metric terms is proposed to solve the feedback optimization problem in the boundary - layer system without the usual exponential stability.
- **Formation - Based Source Localization**: A distributed control law is proposed to solve the problem of multiple mobile agents cooperatively searching for signal sources.

### Main Results

- **Theorem 1**: Gives the conditions satisfied by the perturbation functions \(\rho_s\) and \(\rho_f\) to ensure the ISS property of the generalized singularly perturbed system.
- **Theorem 2**: In some cases, there exist constant - valued \(\rho_s\) and \(\rho_f\) such that the system satisfies the ISS property.

### Conclusion

The paper expands the applicable scope of the traditional singular perturbation theory by introducing state - dependent perturbation functions and provides new methods to solve the control problems of complex systems. These methods are not only of great significance in theory but also have broad prospects in practical applications.