Abstract
Sufficient conditions are derived for global asymptotic synchronization in a system of identical nonlinear electrical circuits coupled through linear time-invariant (LTI) electrical networks. In particular, the conditions we derive apply to settings where: i) the nonlinear circuits are composed of a parallel combination of passive LTI circuit elements and a nonlinear voltage-dependent current source with finite gain; and ii) a collection of these circuits are coupled through either uniform or homogeneous LTI electrical networks. Synchronization in these networks is guaranteed by ensuring the stability of an equivalent coordinate-transformed differential system that emphasizes signal differences. The applicability of the synchronization conditions to this broad class of networks follows from leveraging recent results on structural and spectral properties of Kron reduction—a model-reduction procedure that isolates the interactions of the nonlinear circuits in the network. The validity of the analytical results is demonstrated with simulations in networks of coupled Chua's circuits.
Introduction
Synchronization is a fundamental phenomenon where coupled dynamical systems—such as power generators in a grid, laser arrays, or biological neurons—spontaneously align their states. In electrical engineering, ensuring that nonlinear oscillators synchronize is vital for stable power grids and communication systems. Most existing research focuses on simple network structures or assumes the network is purely resistive. However, real-world networks are often "dynamic," containing energy-storage elements like inductors and capacitors, and have "general topologies" with arbitrary connection patterns.
This paper establishes rigorous mathematical conditions for the global synchronization of identical nonlinear circuits connected through complex, dynamic electrical networks. By leveraging Kron reduction and $L_2$ gain analysis, the authors provide a framework to guarantee that oscillators (like Chua’s circuits) will eventually reach the same terminal voltage, regardless of the network's specific topology, provided it meets certain symmetry or uniformity criteria.
Methodology
Circuit Modeling and Kron Reduction
The authors model each nonlinear circuit as a parallel combination of a linear time-invariant (LTI) subsystem and a nonlinear, voltage-dependent current source with a finite gain. This representation allows for a clean separation of the linear dynamics from the nonlinear excitation.
To handle complex topologies, they use Kron reduction, a Schur-complement-based model reduction procedure. This process mathematically "eliminates" interior nodes that don't have circuits attached, resulting in an equivalent network that explicitly shows how the nonlinear circuits interact at the boundary nodes.
Stability Analysis
The system is transformed into "differential coordinates" using a projector matrix. Instead of looking at absolute voltages, the analysis focuses on the differences between voltages. Synchronization is achieved if these differences decay to zero. The authors treat the system as a feedback loop between a linear network and nonlinear circuit elements. Synchronization is guaranteed if the product of the "differential $L_2$ gain" of the nonlinear part and the $H_\infty$ norm of the linear network part is less than one, applying a "small-gain" argument to signal differences.
Key Insights and Results
- Network Symmetry: The paper identifies two broad classes of networks where synchronization is mathematically tractable: Uniform networks (identical per-unit-length impedances) and Homogeneous networks (identical effective impedance between any two terminals).
- Spectral Connectivity: For dynamic networks (with L and C), synchronization must be checked across all nonzero network modes (eigenvalues), rather than just the algebraic connectivity used for resistive networks.
- Simulation Validation: In simulations of Chua’s circuits, when the derived condition was met (Gain < 1), the circuits synchronized perfectly even from different initial states. When the condition was violated, the circuits failed to synchronize, often falling into independent chaotic trajectories.
Conclusion
This work successfully generalizes synchronization results from simple star topologies to any arbitrary topology, provided the lines are uniform or the network is homogeneous. Future directions include extending this work to heterogeneous networks (where circuits and lines differ) and including nonlinear network elements like diodes and rectifiers, which are common in modern power electronics.
Summary updated