Abstract

This paper studies time-delayed simultaneous input and state estimation to enhance estimation accuracy for systems without direct feedthrough, such as earthquake-excited building structures, using absolute floor acceleration measurements. Rank matching, strong observability, and invertibility conditions are crucial for the stability and convergence of input and state estimation. Real-time approaches have achieved successful estimations when those conditions are satisfied. However, a dynamic system model often does not hold those conditions when using acceleration measurements, leading to significant errors in the estimations. To this end, the authors recently developed an optimal sensor placement algorithm to ensure the system model holds the above conditions to achieve accurate real-time estimation. However, accurate estimation in some cases remains challenging because of incomplete measurements, modeling error, and measurement noise. This paper proposes an extended time-delayed joint input and state estimation algorithm (ETDIS) based on the invertibility matrix. Specifically, by incorporating the prior knowledge of the input, the proposed ETDIS is designed from a Bayesian perspective, considering measurement noise to enhance estimation accuracy. In particular, the innovation is used to obtain the input estimation, which is interconnected with the state space equation for state estimation. ETDIS relaxes the rank-matching condition and is more robust against the lack of conditions. Accurate online input and state estimation with a delay and satisfactory computational cost is achieved by utilizing the proposed ETDIS and limited acceleration measurements. Numerical studies are presented to verify the effectiveness of the proposed method.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call