Lijun Liu

It is still a challenging problem to detect crack damage in beam bridges under moving loads using limited sensor measurements with noise pollution. This study proposes a baseline-free crack damage localization method for beam bridges by combining the discrete wavelet transform (DWT) with independent component analysis (ICA). By considering the acceleration responses of the beam bridge structure at two locations under moving loads and the influence of measurement noises, the noisy acceleration responses are decomposed by wavelet transform, and the high-frequency parts of the wavelet coefficients from the two responses are utilized for signal reconstruction. Subsequently, the reconstructed signals from the two sensors as input, the FastICA algorithm is employed to separate independent components from the mixed signals. Finally, independent component signals containing crack damage information can be obtained, and the crack damage of the beam bridge is localized through the position of singular values in these independent component signals. Numerical simulations of localizing single and multiple cracks in a simply supported beam under moving loads are used to demonstrate that the method exhibits strong robustness against measurement noises and remains effective under various moving load configurations.

Many existing damage detection methods based on structural vibration responses require responses from undamaged structures as the baseline for reference comparison, and the deployment of many measurement sensors on structures limits their practical application. In this paper, a moving hammer excitation and two movable strain sensors are employed in coordination. By extracting the Wavelet Packet Energy (WPE) and relative entropy from adjacent strain responses, a baseline-free distributed damage detection index is established for beam-type structures. Specifically, the analyses of damage indexes for single damage, non-adjacent damage, and adjacent damage are presented, respectively. The performance of the proposed method is validated by numerical simulation cases of structural damage detection of a simply supported beam and the cable-stayed Haiwen bridge girder with various damage scenarios. Results indicate that the relative entropy index demonstrates high sensitivity to damage location and strong robustness to various environmental noises. Moreover, an experimental structural damage detection of an overhanging beam validates the performance of the proposed method. Therefore, an economical and effective decentralized damage detection method using two movable sensors and relative entropy is proposed for baseline-free beam-type structures.

Structural system identification provides a fundamental basis for the damage assessment of civil infrastructure. However, its practical implementation is challenged by inconsistent sensor sampling rates, unknown external excitations, and sparsely distributed structural damage. To this end, a multirate extended Kalman filter with unknown input (MREKF‐UI) is proposed for the joint identification of structural states, sparse damage, and external loads. The proposed MREKF‐UI constructs a dynamic observation model by fusing multirate acceleration and strain response data acquired synchronously or asynchronously. An l ₁ ‐norm regularization term is incorporated as a sparsity constraint on damage and embedded into the MREKF‐UI filter update step via a pseudomeasurement strategy. The proposed framework formulates full‐field response, sparse damage identification, and unknown load estimation within a unified filtering scheme. Numerical simulations on an overhanging beam and a truss structure are conducted to validate the proposed method under synchronous and asynchronous multirate sampling conditions. The results demonstrate accurate damage localization and quantification, reliable identification of unmeasured displacement and velocity responses, and effective estimation of unknown external excitations. Therefore, the proposed MREKF‐UI provides a feasible framework for intelligent structural diagnosis using multisource heterogeneous sensor data under unmeasured excitation conditions.

The identification of fluctuating wind velocity spectra of high‐rise buildings using partial structural responses is challenging due to the ill‐posed nature of the inverse problem. In this paper, two frequency‐domain physics‐informed deep learning methods are proposed to identify stationary fluctuating wind velocity spectra of high‐rise buildings using partial structural responses. The first method targets uniform wind spectra by exploiting their spatial correlations with a one‐dimensional discrete cosine transform (DCT). Structural partial response spectra are used as the inputs to a physics‐informed convolutional neural network (CNN), and the DCT coefficient vector of the wind spectra serves as the outputs, and this physical relationship is embedded into the physical loss function of the CNN. Furthermore, transfer learning is proposed so that the pretrained parameters for a specific frequency of the physics‐informed CNN are used as the initial values of the network parameters at other frequencies, facilitating the identification of wind spectra across different frequencies. Then, it is extended to study the more complex identification of nonuniform wind velocity spectra of high‐rise buildings, and a physics‐informed transfer learning identification method is proposed based on the two‐dimensional DCT of nonuniform wind spectra. Analogously, structural partial response spectra and the two‐dimensional DCT coefficients are used as the CNN inputs and outputs, respectively. The effectiveness of the proposed methods is validated by identifying the uniform and nonuniform fluctuating wind velocity spectra of the wind‐excited 76‐story benchmark building, respectively.

The identification of road roughness using the dynamic responses of passing vehicles is of practical value. However, with vehicle parameters unknown a priori, unknown vehicle parameters and road roughness are coupled in the identification. In existing methods, vehicle parameters and road roughness are identified based on complex stage identification or mutual iteration. Moreover, road roughness is random in nature, but there is a lack of research on the identification of statistical road roughness. In this paper, an efficient method is proposed for the direct identification of joint vehicle parameters and road roughness field using dynamic response data fusion of three‐dimensional (3D) moving vehicles. First, the extended Kalman filter with unknown inputs is used to identify vehicle parameters, except tire stiffness, and unknown forces in a 3D moving vehicle. Then, the tire stiffness and road roughness are identified based on the property that front and rear vehicle wheels pass same road surface. Furthermore, it is extended to study the identification of joint vehicle parameters and Gaussian random road roughness field. Based on the vehicle parameters identified using the above deterministic method, the Karhunen–Loève (KL) expansion is performed on vehicle responses to obtain their truncated KL components, which are used as the observations of the Kalman filter with unknown inputs to identify the corresponding KL components of random road roughness. Finally, statistics of Gaussian random road roughness are estimated. The proposed deterministic and stochastic methods are tested by numerical simulation of 3D vehicle responses with different road roughness levels.

In this paper, a damage identification method is proposed for bridge structures under unknown seismic excitation using wavelet-based transmissibility functions (TFs) of structural strain responses. Structural strain responses of intact structures under unknown ambient excitation are used as baselines. The wavelet-based TF of structural strain responses of damaged structures under unknown seismic excitation is compared with that of intact structures under ambient excitation to construct a damage index. This damage index effectively eliminates the influence of different excitation, i.e. ambient excitation to intact structures and seismic excitation to damaged structures. Compared with the current TF-based method of structural acceleration or displacement responses, the proposed method avoids the frequency band selection in the frequency domain TF and utilizes the high sensitivity of structural strain responses to local damage. Additionally, it enhances local feature extraction ability by the wavelet transform. The proposed method is first illustrated through numerical damage identification of a simply supported overhanging beam and the main girder and tower of the Haiwen bridge. Then, it is validated by the damage identification of a cable-stayed bridge tower model under laboratory shake table tests. The damage identification results demonstrate that structural damage can be accurately located and damage degrees can be reflected by the damage index constructed from the wavelet-based TF of structural strain responses with the influence of noise, which is beneficial for structural health monitoring of actual bridges.

Road surface roughness inherently exhibits stochastic characteristics due to spatial variability and temporal unpredictability. While moving vehicle response-based methods have been employed for road roughness identification due to their practicality and cost-effectiveness, there is a lack of statistical identification of random road roughness. The difficulty arises from the coupling of the inverse identification problem and the random dynamic responses of a moving vehicle. Moreover, the intrinsic coupling between the unknown vehicle parameters and road roughness requires double loops of iterative procedures to simultaneously identify both vehicle parameters and roughness characteristics. To address these limitations, a novel stochastic approach using a half-car model of a moving vehicle is proposed in this paper for the simultaneous identification of vehicle parameters and the statistical characteristics of Gaussian random road roughness. First, the Karhunen–Loève (K–L) expansion is employed to represent the stochastic dynamic response of moving vehicles while capturing its key statistical features. Specifically, the zeroth-order K–L component of the vehicle response is used as the observation of the extended Kalman filter with unknown inputs to identify the mean of the unknown forces and vehicle parameters, except tire stiffness. Subsequently, by leveraging the fact that the front and rear axles traverse the same road surface, the mean of random road roughness and vehicle tire stiffness are identified. Then, the remaining K–L components are utilized as observations of the Kalman filter with unknown inputs to estimate the K–L components of the random road roughness. Finally, sufficient samples of random road roughness can be generated by the identified K–L components of random road roughness, and the statistical properties such as the variance and power spectral density of random road roughness can be estimated. The effectiveness of the proposed method is validated through numerical simulation studies under varying levels of road roughness and measurement noises, demonstrating the accuracy and robustness in identifying both vehicle parameters and Gaussian random road roughness statistics.

The proper selection of the model error covariance matrix and the measurement noise covariance matrix of Kalman filter is an optimization problem. Some scholars have studied this, but there is relatively little research on the selection of the two covariance matrices for Kalman filters with an unknown input. Recently, the authors proposed a modified directed bat algorithm (MDBA) which introduces the best historical location of individuals and the elimination strategy to effectively prevent falling into local optimal solution. So, two methods are proposed in this paper to optimize the model error covariance matrix and measurement noise covariance matrix of Kalman filter with unknown inputs (KF-UI) and extended Kalman filter with unknown inputs (EKF-UI) by MDBA, respectively. The objective functions are constructed using the measurement vectors and the corresponding estimated values, and MDBA is adopted to optimize the two covariance matrices of KF-UI and EKF-UI. To validate the effectiveness of proposed methods, two simple structure examples and a benchmark example are adopted. The influence of structural parameter uncertainties on KF-UI is also considered. The result shows that the MDBA-optimized KF-UI has a strong convergence and can take into account the effect of parameter uncertainties. Then, the effectiveness of the proposed MDBA-optimized EKF-UI method is validated by comparing it with EKF-UI with empirically selected covariance values through trial-and-error. The identification results showed that the proposed methods achieved better identification accuracy and enhanced convergence compared to KF-UI and EKF-UI with empirical covariance values.