Speaker
Description
The growing interest in pulse-like ground motions, especially occurring near faults, stems from their potential to inflict significant structural damage due to distinct directivity and fling effects. This study seeks to enhance the identification and classification of these impulsive ground motions by integrating traditional data-processing methods with advanced machine learning techniques. Although traditional data-processing methods are effective, they often require intensive calculations. In this research, traditional methods will be utilized to identify pulse-like ground motions, followed by application of Machine Learning techniques for classification purposes. The methodology involves training various classification algorithms on an extensive data set of ground motion records and comparing their performance in identifying pulse-like signals and categorizing them accurately. This study assesses the reliability and accuracy of these models, emphasizing their capability to address the computational challenges inherent in traditional approaches. Experimental results indicate that classification algorithms can reliably identify pulse-like ground motions and predict their pulse periods with high accuracy, offering a viable alternative to conventional methods. This advancement holds significant implications for seismic hazard analysis and earthquake engineering, suggesting that the integration of Machine Learning and conventional methods can contribute to the development of more resilient structures capable of withstanding near-fault seismic events.
[email protected] |