فهرست نشریات علمی دسترسی آزاد ایران

Recent advancements in Deep Learning (DL) have significantly expanded its application to address myriad challenges in civil and earthquake engineering. However, a notable challenge persists: the scarcity of reliable data pertinent to earthquake engineering, which may compromise the accuracy of DL-derived results. In response to this challenge, Generative Adversarial Networks (GANs) have emerged as a promising solution. Initially conceptualized to improve the training of generative models, GANs have exhibited exceptional performance and adaptability, particularly in image generation, gaining substantial recognition within the academic community. In structural engineering, the generation of synthetic ground accelerograms that conform to a specified target response spectrum is essential for conducting nonlinear dynamic analyses. This paper introduces an effective algorithm for spectral matching, facilitating the generation of numerous artificial, spectrum-compatible earthquake accelerograms from a limited set of ground motion records. The proposed algorithm represents a significant advancement in the field, addressing the critical need for robust and accurate synthetic data in earthquake engineering. Consequently, the integration of GANs into this domain not only enhances the reliability of DL applications but also paves the way for more precise and comprehensive analyses, thereby contributing to the overall progress of civil and earthquake engineering disciplines.