A physics-informed simulation-based digital twin framework for condition monitoring and health assessment of automotive suspension systems

Abstract

In the present study, a physics-informed simulation-based digital twin framework for an automotive suspension system exposed to transient and stochastic road excitations is formulated for health monitoring and prognostics. A physics-based quarter-car dynamic model is considered for the development of governing equation of motion and integrated with the state estimation and online damping parameter tracking to synchronize the virtual model with the simulated degrading physical system. Further, the proposed framework is assessed through obtaining the response data under controlled road excitations, where the prescribed road displacement and velocity serve as excitation inputs during the state estimation process. A normalized suspension health index (SHI) is obtained continuously from the estimated damping coefficient to quantify its degradation in real time. Initially, remaining useful life (RUL) of the suspension system is evaluated using the conventional slope-based degradation approach. The numerical instability of the slope-based approach is evident from the presence of artificial spikes and projection of unreliable life. However, this limitation is overcome through developing a cumulative damage-based prognostic of suspension system such that the life degradation is expressed in terms of integral of degradation of intensity of the damping coefficient, obtained through SHI. The comparative analysis under half-cosine bump and stochastic road inputs demonstrates that the proposed damage-based RUL approach provides stable, interpretable, and monotonic prognostic estimates. Further, the robustness of the proposed framework against the various uncertainties in vehicle loading is investigated through performing a mass sensitivity analysis. The various uncertainties such as the variation in the sprung mass depending on passenger occupancy, payload, and operating conditions are considered for the investigation. Further, the results demonstrate that the proposed digital twin framework enhances the robustness and reliability of suspension life prediction through integrating the time-varying damping coefficient, SHI, and cumulative damage modelling. It can be concluded that the proposed simulation-based digital twin framework yields a scalable foundation for the physics-informed health monitoring, prognostic studies and offers a extensible basis for future real-world implementation.