Decentralized modal parameter estimation using time-domain stepwise mode reconstruction

Abstract

Infrastructures are often subjected to dynamic loads such as traffic, wind, and earthquake, resulting in non-stationary operational responses. The stationary conditions used in the traditional modal parameter identification methods will affect the identification accuracy. To address this limitation, this paper proposes a decentralized modal parameter estimation framework for non-stationary conditions, which has key improvement measures. The framework introduces the time-domain stepwise mode reconstruction technique, which effectively reduces the influence of noise and non-stationary effects on the accurate modal parameter estimation by iteratively optimizing the frequency estimation and reconstructing mono-components. It also combines automatic initial condition estimation to eliminate the manual initialization process through adaptive baseline correction and multi-scale peak detection. In addition, the framework adopts a decentralized architecture, which enables independent sensor-level analysis, reduces data transmission requirements, and enhances system flexibility. This method supports scalable implementation because structural responses are analyzed independently on each sensor node, and mono-components are combined to extract modal shapes. The effectiveness of the proposed method is verified by using simulated data from a three-degree-of-freedom system, measured data from a real footbridge, and the ASCE benchmark model. The results show that the framework has a strong ability to deal with non-stationary responses and has the potential to achieve accurate and scalable decentralized modal parameter estimation, making it a powerful tool for structural health monitoring.