Structural anomaly detection and nonlinearity assessment by integrating phase angle and video-based analysis via a shaking table test

• A video-based method for anomaly detection due to structural nonlinearity is proposed. • The mono-component instantaneous phase angle index demonstrates high sensitivity to structural damage occurrence. • Video-based methods effectively evaluate structural nonlinearity degree and detect damage, while phase-based methods excel in identifying data anomalies. • A three-story frame model shaking table test validates the efficiency of the proposed video-based damage detection method. • Video-based technology enables post-earthquake evaluation of structural damage conditions. Dynamic loads from seismic events frequently result in structural damage, which can significantly impact serviceability and even reduce the safety of civil structures, potentially rendering them unusable. Such damage often leads to a shift from linear to nonlinear behavior in structural dynamics. Consequently, understanding the accumulation of nonlinearity and accurately detecting its location are essential for evaluating the condition and integrity of the structure. In recent years, video data has been widely utilized for structural response evaluation and damage detection due to its ability to extract structural responses from any location, while phase-based methods have proven highly sensitivity to data anomalies. This study employs phase-based methods to validate the accuracy of the proposed video-based method, further demonstrating its superior capability in identifying the degree of structural nonlinearity. A three-story frame model shaking table test is designed for validation. First, the mono-component instantaneous phase angle index is employed to identify the timing of damage occurrence, verifying the presence of damage and providing a reference for subsequent video-based analysis. Then, an extended node strength network index based on the optical flow method is proposed, enabling the identification and visualization of damaged locations. Furthermore, this proposed algorithm achieves identification of the nonlinearity accumulation process across all test scenarios and quantifies nonlinearity through the index of average pixel point number. This study offers a novel perspective on the nonlinear damage evolution throughout the structural damage process during seismic events.