A fixed-lag smoother for solving input force and support motion identification problems in structural dynamics
2016 (English)Manuscript (preprint) (Other academic)
In this work, the problem of joint input and state estimation based on measured response in a dynamic system is addressed.A fixed-lag smoother for support motion or input force estimation in linear structural dynamics is proposed. The unknown inputs are estimated in conjunction with the states in an augmented state description of the system. A fixed-lag smoother maximizes the use of information available in the measurements by allowing a small time lag in the estimation, leading to a recursive non-causal filter. This type of filter has not previously been implemented in the field of structural dynamics. The filter is capable of handling any combination of measured displacement, velocity and acceleration; furthermore are both measurement noise and model errors accounted for and simulated as stochastic processes. In this study, estimations are based on sparse noisy measurements of system displacement and model errors are simulated through the use of deliberately incorrect model descriptions. The proposed algorithm is verified by numerical simulations and the qualities of the estimations are compared to those of the augmented Kalman filter.It is shown that with this filter, significant improvements of the input estimations are achieved, especially of inputs located far away from measurement sensors. The improvement consists of noise reduction, phase correction and by a more accurate division of the individual contribution of multiple inputs.
Place, publisher, year, edition, pages
2016. , 26 p.
Joint input-state estimation, Fixed-lag smoother, Force identification, Structural dynamics
Other Mechanical Engineering
Research subject Solid Mechanics
IdentifiersURN: urn:nbn:se:kth:diva-193052OAI: oai:DiVA.org:kth-193052DiVA: diva2:974810
QC 201609282016-09-272016-09-272016-09-28Bibliographically approved