Dynamic response identification based on state estimation and operational modal analysis

Abstract

Abstract This paper presents and experimentally validates an augmented Kalman filter extended with a fixed-lag smoother for solving joint state and input estimation problems. Sparse acceleration measurements from a truck side skirt excited by road-induced vibrations from a vibration test track are analysed. The system model is obtained experimentally from an operational modal analysis, reducing modelling errors and avoiding the need for a finite element model and it serves itself as a numerical model. The motion of the truck component is estimated and the results are compared to those of a joint input-state estimation filtering algorithm, in addition to the actual measured motion. Both algorithms are tuned according to a novel process based on minimal a priori information concerning the system states and inputs. The focus of this work is to assess the robustness, performance, and tuning of the algorithms. Two sensor configurations are studied: one where the number of response measurement sensors is high compared to the number of estimated motions and participating modes, and another where the number of response measurements is reduced. Both algorithms perform very well within the first configuration. With a reduced number of response measurements, the fixed-lag smoother is superior to the joint input-state filter in capturing the individual motion of each position on the side skirt.