Bridge influence line identification method based on enhanced dynamic-static separation and matrix solution under moving vehicle

Bridge Influence Line (BIL), which provides mechanical information directly related to structures, is widely utilized for bridge condition evaluation. However, accurate BIL identification requires decoupling dynamic effects and environmental noise from dynamic responses of bridge. This paper proposes a novel BIL identification method based on enhanced dynamic-static separation and matrix-solving techniques to reliably extract structural influence lines under standard operating vehicle speeds. Firstly, the Adaptive Filtering Empirical Mode Decomposition (AFEMD) for dynamic-static separation is proposed, which enhances conventional Empirical Mode Decomposition (EMD) through low-pass filtering and adaptive cutoff conditions. This method can effectively eliminate dynamic fluctuation components and environmental noise, thereby decoupling a smooth quasi-static response from raw dynamic responses with a peak relative error of less than 1.37 %. Subsequently, the physics-based Tikhonov Regularization Singular Value Decomposition (TRSVD) is established to solve the BIL matrix, where the Tikhonov regularization technique combined with Singular Value Decomposition (SVD) addresses multi-axle vehicle effects and reduces the difficulty of solving the matrix under a peak relative error of less than 1.44 %. Further, the numerical simulations and parametric analysis are validated to study the impacts of vehicle speed and road surface roughness on identification accuracy. Finally, the dynamic field test with a moving vehicle demonstrates that the proposed method can achieve reliable BIL identification under real-world operating vehicle speeds with a peak relative error of less than 2.59 %.