Work

Play-Operator-Based Koopman Model for Nonlinear Hysteretic Systems

Year: 2026
Type: conference-paper
Abstract: This paper presents a Play-operator-based Koopman modeling approach for nonlinear hysteretic systems. To characterize path-dependent memory effects, a bank of one-sided Play operators is introduced to generate hysteresis-related features from the input history, which are then incorporated into the lifted-state construction. Based on this lifted representation, a bilinear Koopman-inspired predictor is established and identified directly from input–output data by least squares. To validate the proposed method, experiments are conducted on a pneumatic bellows actuator (PBA) under sinusoidal pressure excitations with different amplitudes. The results show that the identified model can accurately reproduce the system response and maintain satisfactory prediction performance across different excitation levels. These results demonstrate that the proposed method provides an effective data-driven modeling framework for nonlinear hysteretic systems and offers a compact representation for prediction and future control-oriented applications.
Language: English
Open Access status: hybrid
OpenAlex ID: W7204200843
FWCI:
Cites: 13
Cited by:
Related to: 10