Inverse Reconstruction of Unsteady Aerodynamic Loads Acting on Railway Vehicles

During normal operation, railway vehicles often endure significant vibrations due to unsteady aerodynamic loads. Precisely quantifying these transient forces offers essential insights for operational safety monitoring and vehicle aerodynamic testing. In this paper, we introduce an innovative inverse method for reconstructing active aerodynamic loads using a limited number of acceleration measurements. This method capitalizes on health monitoring instruments already present on the vehicles, thereby eliminating the necessity for supplementary pressure sensors on the vehicle's exterior surface, as mandated by traditional direct pressure measurement strategies. We develop a Multi-Task Gaussian Processes (MTGP) inverse estimation technique to calculate the conditional probability distribution of loads given the noise-affected acceleration data. The MTGP approach boasts the advantage of analytically forming the posterior of unsteady aerodynamic loads at any time point, as well as offering high reconstruction accuracy. To validate our proposed method, we utilize a numerical example with a 31 DOF railway vehicle model. Aerodynamic loads generated by two trains passing each other are applied to the vehicle model, and acceleration data from the bogies are employed for the inverse reconstruction process. Our results successfully demonstrate the feasibility of reconstructing unsteady aerodynamic loads on railway vehicles, highlighting the potential of our novel approach.