A spatiotemporal dual kalman filter for the estimation of states and distributed inputs in dynamical systems

Abstract

The fusion of Gaussian Process (GP) models with the Dual Kalman Filter (DKF) is proposed in this contribution, with the aim of recursively estimating the state and distributed input in dynamical systems using a limited number of vibration response measurements. To do so, the distributed system excitation is represented by a GP model, enabling thus the implementation of a space-time filtering approach for the input process, which is then coupled with a Kalman filter for the state estimation of system dynamics. The proposed approach is assessed in terms of a simulated case study on the finite element model of a wind turbine blade under operational conditions, which is excited by a distributed drag and lift load along the length of the blade.