A Novel Recurrent Input Shaping with Particle Filter-calibrated Parameter for Robust Vibration Control

Vibration control in underactuated nonlinear systems remains a critical challenge in modern control engineering. Data-driven input shaping has been widely adopted as an effective and easily deployable feedforward control technique, while its performance heavily relies on precise system parameters. However, existing methods predominantly depend on single-mode models and suffer from parameter inaccuracies, leading to suboptimal control performance. To address these limitations, this paper proposes a Particle filter-calibrated Recurrent Input shaping (PRI) model based on two-fold ideas: 1) utilizing a particle filter to calibrate the initial system parameters, and 2) introducing a novel recursive input shaping method to efficiently optimize impulse parameters in multimodal systems. Extensive experimental results demonstrate that the proposed PRI model achieves significant improvements in both optimization accuracy and control performance compared to state-of-the-art approaches.