Forced Resonance Model Overview

• Forced resonance model is an analytical framework that quantifies dynamic systems' responses near natural frequencies using modal decomposition and reduced order modeling.
• It employs spectral submanifold reduction to transform high-dimensional problems into low-dimensional algebraic systems, achieving significant computational speed-ups.
• Applications include mechanical, structural, and power systems where the model predicts phenomena such as amplitude jumps, isolas, and resonance-induced instabilities.

The forced resonance model refers to the analytical and computational framework for characterizing the response of a dynamical system—typically linear or weakly nonlinear and often high-dimensional—to external periodic or quasi-periodic driving, especially when the drive frequency is close to, or coincides with, internal (natural) frequencies of the system (“resonance”). The model encompasses modal decomposition, reduced-order modeling, response reconstruction, and parametric continuation, with extensions for systems featuring multiple resonances, internal nonlinearities, or parameter uncertainty.

1. System Formulation and Modal Decomposition

Forced resonance models begin with the second-order mechanical (structural or electrical) system, linearized around an equilibrium:

$$M \ddot{x}(t) + C \dot{x}(t) + K x(t) + f(x,\dot{x}) = \epsilon f^{\text{ext}}(\Omega t)$$

where $x \in \mathbb{R}^n$, $M, C, K$ are mass, damping, and stiffness matrices; $f(x,\dot{x})$ is a weakly nonlinear function ($\mathcal{O}(|x|^2, |x||\dot{x}|, |\dot{x}|^2)$); and $\epsilon f^{\text{ext}}(\Omega t)$ is a small-amplitude, typically harmonic, external forcing with frequency $\Omega$. In power systems, this structure arises from linearization of swing equations with sinusoidal perturbations at selected buses (You, 2020).

This continuous system is recast in first-order state-space form:

$\dot{z}(t) = A z(t) + B u(t), \qquad y(t) = C z(t)$

where $z$ collects positions and velocities, $u(t)$ is the external driving, and $(A,B,C)$ encapsulate system dynamics and measurement operators. Modal decomposition yields a basis of eigenvectors and eigenvalues, facilitating projection onto dominant modes (including those involved in resonance). The resonant response, in the weak-damping limit, concentrates near frequencies where the external drive matches the imaginary part of an eigenvalue, leading to modal amplification governed by the transfer function poles (You, 2020, Huang et al., 2018, Delabays et al., 2022).

2. Internal Resonance and Spectral Submanifold Reduction

For large-scale or nonlinear systems, explicit computation of the forced response benefits from dimensional reduction. Spectral submanifold (SSM) theory provides an invariant-manifold-based model order reduction scheme, in which the system’s response is projected onto manifolds tangent to “master” resonant modes (Li et al., 2021, Li et al., 2023). Parameterization yields:

$$z(t) = W_\epsilon(p,\phi), \qquad \dot{p} = R_\epsilon(p, \phi), \qquad \dot{\phi} = \Omega$$

where $W_\epsilon$ embeds the SSM into phase space and $R_\epsilon$ governs reduced dynamics in the modal amplitudes and phases. The forced response curves (FRCs) are obtained as equilibria of the reduced system:

$r_i^p(\rho, \theta, \Omega, \epsilon) = 0$

for each mode $i$, where $q_i = \rho_i e^{i(\theta_i+r_i \Omega t)}$. Nonlinearities and internal resonances are captured via normal-form terms and resonance monomials in $R_\epsilon$. This approach reduces high-dimensional periodic orbit continuation to the solution of low-dimensional algebraic systems, often with $2m$ variables for $m$ resonant mode pairs (Li et al., 2021, Li et al., 2023).

3. Forced Response Curves and Surfaces: Continuation and Features

A key output is the forced response curve (FRC), the set of periodic solutions (response amplitudes) as a function of forcing frequency at fixed amplitude. For nonlinear and amplitude-varying responses, this generalizes to the forced response surface (FRS)—the set of all periodic solutions as a function of both forcing frequency and amplitude. The SSM-based reduction yields an implicit scalar algebraic relation, e.g.,

$$\mathcal{F}(\rho, \Omega, \epsilon) := a(\rho)^2 + [b(\rho)-\Omega]^2 \rho^2 - \epsilon^2 |f|^2 = 0$$

which is continued numerically as a manifold in $(\rho, \Omega, \epsilon)$ (Li et al., 2023).

Within the FRS, critical features include:

• Ridges: loci of local response maxima (highest amplitude).
• Trenches: loci of local response minima.

These features are algorithmically traced by supplementing the implicit function defining the FRS with extremum conditions (vanishing derivative with respect to frequency or amplitude).

The computational advantage is pronounced: ridges and trenches in FRSs for high-dimensional systems (up to $n \sim 10^3$) can be computed orders of magnitude faster via SSM-ROM continuation than with full-system shooting or collocation (Li et al., 2023).

Feature	SSM/ROM method	Full-order method
FRC	$\sim$ seconds–minutes	$\sim$ days
2D FRS	$\sim$ hours	Exponentially expensive
Ridges	$\sim$ seconds	$\sim$ days

4. Practical Implementation and Applications

SSM-based forced resonance models are applied to a variety of mechanical and structural systems: finite-element (FE) models of von Kármán beams, plates, and shallow shells have been reduced to SSMs featuring $m=1$ or $m=2$ dominant mode pairs, enabling FRC and FRS computation, bifurcation analysis, and stability characterization even for models with up to $n \approx 10^5$–$10^6$ DOF. Numerical continuation is performed with packages such as COCO and SSMTool (Li et al., 2021, Li et al., 2023).

The SSM-based framework accommodates:

• Internal resonances and multi-mode interactions via augmented normal forms.
• Nonlinear amplitude effects (bending/jump phenomena, isolas, Neimark–Sacker bifurcations).
• Parameter continuation in both frequency and forcing amplitude, yielding global response surfaces.

5. Limitations, Extensions, and Theoretical Underpinnings

The model rests on assumptions of weak nonlinearity and light damping; convergence of the SSM series is local in amplitude and excitation. Nonresonance conditions on “slave” modes are necessary but are generically satisfied in practical systems. Continuation may miss isolated FRC branches (isolas) unless the augmented system is expanded to capture multi-valued structure. Estimation of safe parameter regimes for SSM validity is an open problem (Li et al., 2021).

The forced resonance model connects to classical modal superposition and backbone curves in linear/weakly nonlinear systems, as well as to sophisticated PDE-based response theory in structural and fluid-structure interaction contexts (Li et al., 2023). It extends the principle that resonance-induced response amplification is governed by the alignment of the forcing frequency with internal modal frequencies or their nonlinear generalizations, and is made computationally tractable via dimension reduction.

6. Computational Advantages and Physical Insights

Reduction to SSMs and low-dimensional ROMs confers dramatic computational gains: typical speed-ups are two to four orders of magnitude compared to full-system periodic orbit continuation. This enables efficient global exploration of response surfaces in high-dimensional, physically-relevant models, and extraction of salient resonance phenomena—amplitude jumps, stability loss, quasi-periodicity, and amplitude plateaus—with fine parameter resolution (Li et al., 2023).

The forced resonance model provides a principled, scalable, and extensible framework for studying, predicting, and controlling resonance phenomena in large-scale mechanical, structural, and electromechanical systems subject to external periodic excitation. It is foundational for applications in vibration analysis, power grid stability, acoustic and electromagnetic meta-materials, precision engineering, and beyond.