Computationally Efficient Methods for Estimating Unknown Input Forces of Structural Systems

TitleComputationally Efficient Methods for Estimating Unknown Input Forces of Structural Systems
Publication TypeConference Proceedings
Year of Conference2020
AuthorsVan Nguyen, H., D. Ranasinghe, A. Skvortsov, and S. Arulampalam
Conference Name23rd International Conference on Information Fusion
Date Published07/2020
PublisherIEEE
Conference LocationVirtual Conference
KeywordsBayes estimations, Ensemble Augmented Kalman filters (EnAKF), Mass-spring-damper system, Minimum variance unbiased (MVU) filters, Non-linear filtering
Abstract

We consider the problem of estimating unknown input forces on structural systems using only noisy acceleration measurement data. This is an important task for condition monitoring, for example, to predict fatigue damage in a structure's body or to reduce transmission of vibrations in marine vessels. In this paper, we propose a new idea to estimate an input force with a sinusoidal form by formulating a force identification problem without a direct feed-through system. Consequently, the minimum variance unbiased (MVU) filter can be implemented coupled with a fast Fourier transform algorithm to estimate unknown input forces accurately in real-time. Moreover, when the input force is completely unknown, the ensemble sampling method combined with an augmented Kalman filter can be formulated to significantly reduce computation time. Experimental results confirm the effectiveness of our proposed methods and show that the formulations investigated outperform other state-of-the-art methods in terms of computational cost whilst not compromising estimation performance.