摘要:
数十年间，地震学家及地震工程学家通力合作，为包括地震构造特征、地震活动性、震源特性、地震动预测模型及场地效应等多个关键问题的解决提供了支撑，形成了地球科学与工程科学交叉具有独特性的工程地震学科，并取得了系统的应用性研究成果。作为工程地震学重要分支的强震动地震学得到了迅猛发展，为地震区划和工程抗震研究奠定了坚实基础，为城乡建设和核电、交通、能源等多类型行业的发展提供了地震安全性的保障。近年来，随着算力、算法及算料（数据）等人工智能关键要素的大力发展，使得进一步实现强震动地震学与信息学科交叉成为可能，也迅速成为本领域的热点问题。论文首先分析了强震动地震学研究进展与关键问题，探讨了其与人工智能交叉的框架。而后，从知识嵌入、数据-知识融合及知识发现三个层面，综述了行业研究成果，重点介绍：（1）地震波动相关的控制方程与边界、初始条件物理嵌入理论与求解方法；（2）数据与物理机制联合驱动的人工智能地震动模型构建理论与方法；（3）强震动人工智能生成模型等。最后，讨论了目前强震动地震学与人工智能交叉研究亟需解决的关键问题，并对未来的发展方向进行了展望。
Abstract:
Over the past few decades, seismologists and earthquake engineers have collaborated closely, providing support for the resolution of multiple key issues, including seismic tectonic characteristics, seismic activity, source characteristics, ground - motion models, and site effects. This has led to the formation of the engineering seismology discipline, a unique intersection of earth science and engineering science. Systematic applied research results have been achieved. In particular, strong - motion seismology, an important branch of engineering seismology, has witnessed rapid development, laying a solid foundation for seismic zoning and engineering anti - seismic research and ensuring seismic safety for urban and rural construction as well as the development of various industries such as nuclear power, transportation, and energy. In recent years, with the significant development of key elements in artificial intelligence, such as computing power, algorithms, and data, it has become possible to further integrate strong - motion seismology with information science, which has quickly become a hot topic in this field. This paper first analyzes the research progress and key issues in strong - motion seismology and explores the framework for its intersection with artificial intelligence. Then, from three aspects of knowledge embedding, data - knowledge fusion, and knowledge discovery, it reviews the research results in the industry, with a focus on: (1) The theory and solution methods for the physical embedding of governing equations, boundaries, and initial conditions related to seismic wave propagation; (2) The theory and methods for constructing artificial - intelligence - based ground - motion models driven by a combination of data and physical mechanisms; (3) Artificial - intelligence - based strong - motion generation models. Finally, it discusses the key issues that urgently need to be addressed in the current cross - research between strong - motion seismology and artificial intelligence and presents prospects for future development.
点击查看大图
下载:
