Multimodal sensor fusion of inertial, strain, and distance data for state estimation of adaptive structures using particle filtering

This contribution depicts an evaluation of the potential of state estimation techniques in the context of adaptive structures. Those techniques are necessary, as the control algorithms acting on the actuators of the adaptive structures need to know the system's state, but not all states are accessible for measurement. We model and simulate a truss structure, which is equipped with inertial sensors, strain gauges, and an optical scanning system. The latter performs a point-wise scan of the truss structure's surface, which renders the measurement equation time-varying. The truss structure is excited by wind in the simulation. We implement a particle filter in order to perform the multimodal sensor fusion and state estimation. The particle filter shows good results for both an unexcited scenario as well as for a wind excitation scenario. We discuss the results as well as the implications for the active control of adaptive structures.