Review on artificial intelligence-based large language models for geological hazards
-
摘要:
当下，大语言模型技术迭代速度极快，正加速融入地质灾害防治领域。它不断拓展应用场景，突破了以往在数据分析和复杂建模能力上的局限，革新了传统研究范式。为进一步推动大语言模型等AI技术在地质灾害智慧防治方面取得新突破，文章全面梳理了大语言模型技术的演进特点，以及在多个领域的应用情况。首先，论述了小样本学习、多模态数据融合、模型轻量化与迁移应用，以及专家知识嵌入与人机协同等关键技术及其应用于地质灾害隐患智慧识别的主要思路与研发重点方向。在此基础上还提出基于“应用场景、关键问题、作用机制、数据模态、样本特征、模型研发、专家知识、人机协同”等核心要素的“AI+地质灾害”研究框架、技术思路与典型应用场景，凸显出AI技术在地质灾害领域中处理多维多尺度非线性复杂关系建模时的重要价值。通过以上分析，以促进大语言模型等AI技术从数据、模型、知识等更深层次，在更多场景中融入地质灾害防治工作，更好地借助AI技术，推动我国防灾减灾工作朝着精准化、智能化方向发展。
-
关键词:
- DeepSeek大语言模型 /
- “AI+地质灾害”研究范式 /
- 隐患智慧识别 /
- 数据-模型-知识协同 /
- 地质灾害智慧防治
Abstract:Currently, the technology of large language models is evolving rapidly and accelerating its integration in geological disaster prevention and control. It has been expanding the application scenarios and breaking the limitations in data analysis and complex modeling capabilities as well as innovating the traditional research paradigm. To further promote new breakthroughs in AI technologies in the intelligent prevention and control of geological disasters, this article reviews the evolution characteristics of large language model technology and the application scenarios in multiple fields, and also discusses the key technologies including small sample learning, multimodal data fusion, lightweight model and transfer application, as well as expert knowledge embedding and human-computer collaboration, which are also the main ideas and research focus directions for achieving intelligent identification of geological disaster hazards. The article also proposes an "AI + geological disasters" research framework, technical ideas and typical application scenarios based on core elements including "application scenarios, key issues, mechanism of action, data modalities, sample characteristics, model development, expert knowledge, and human-computer collaboration". This highlights the important value of AI technology in geological disasters research in solving the dealing with multi-dimensional, multi-scale, nonlinear and complex relationship modeling problems. The purpose of this article is to promote AI technologies to integrate into geological disaster prevention and control work at a deeper level, from data, models, and knowledge, and also better leverage AI technology to promote the development of disaster prevention and mitigation towards a greater precision and intelligence.
-
-
表 1 LLMs研发策略与关键技术描述
Table 1. LLMs development strategies and core technical components
思路 策略 主要技术 数据
高效
利用强化数据制备与
噪声清洗利用Attention-Driven Temporal Filtering、Self-Supervised Denosing、Multimodal Joint Denoising、生成对抗网络（generative adversarial network，GAN）及其变种（如CycleGAN）等技术，对视频帧、音频信号、时序数据等多模态数据进行噪声清洗 强化对多模态数据的融合对齐与集成训练 采用数据特征级对齐、语义级对齐、模态间映射和联合学习的策略，利用CLIP-Style等技术，将不同模态的数据映射到统一多维空间内；通过对比学习、掩码预测等多模态自监督学习方法提升融合效果 强化样本生成与
特征增强利用物理驱动的合成数据生成方法，减少对真实数据的依赖；应用Conditional GAN生成特定条件下的数据，扩展训练样本的覆盖度 模型
减重
优化减少模型参数 利用PEFT、Adapters和Prefix Tuning等方法，在保持模型性能的同时减少需要训练的参数 优化任务分配 基于MoE、Sparse Sparse MoE等模型架构，通过动态路由机制将任务分配给特定专家模块 加速模型推理 利用Flash Attention优化技术或线性注意力机制（如Performer、Linformer等）技术，提升注意力机制效率 缩小模型体积 利用INT8量化技术、模型剪枝和知识蒸馏，在精度损失可接受范围内压缩模型参数长度 强化模型迁移 利用低秩自适应（low-rank adaptation，LoRA）、元学习等技术，利用跨领域小样本特征对模型参数进行微调，适应新任务 知识蒸馏技术 利用机器学习技术，将体积庞大、结构复杂、具有庞大参数的模型知识迁移到轻量化模型中，在保持较高性能的同时，显著降低模型的计算复杂度和存储需求 人机
协同
融合强化专家知识嵌入 利用RLHF或交互式机器学习（interactive machine learning，IML）方法，整合专家经验，嵌入模型训练计算，优化模型决策逻辑 构建人机协同
决策体系利用众包理念吸纳专家参与制备高质量样本数据，实现专家介入修正模型偏差；
通过协同过滤和集体智慧等理念提升决策准确性
下载: 导出CSV
-
[1] WANG Hanchen,FU Tianfan,DU Yuanqi,et al. Scientific discovery in the age of artificial intelligence[J]. Nature,2023,620(7972):47 − 60. doi: 10.1038/s41586-023-06221-2
[2] LACHOWYCZ S. Utility of artificial intelligence in geoscience[J]. Nature Geoscience,2024,17(10):953 − 955. doi: 10.1038/s41561-024-01548-5
[3] GOH E,GALLO R J,STRONG E,et al. GPT-4 assistance for improvement of physician performance on patient care tasks:A randomized controlled trial[J]. Nature Medicine,2025.
[4] ZHAO Tianjie,WANG Sheng,QUYANG Chaojun,et al. Artificial intelligence for geoscience:progress,challenges and perspectives[J]. The Innovation,2024(5):100691.
[5] CHEN Min,QIAN Zhen,BOERS N,et al. Collaboration between artificial intelligence and Earth science communities for mutual benefit[J]. Nature Geoscience,2024,17(10):949 − 952. doi: 10.1038/s41561-024-01550-x
[6] REICHSTEIN M,CAMPS-VALLS G,STEVENS B,et al. Deep learning and process understanding for data-driven Earth system science[J]. Nature,2019,566(7743):195 − 204. doi: 10.1038/s41586-019-0912-1
[7] LIU Changchun,NIE Qingwei. A blockchain-based LLM-driven energy-efficient scheduling system towards distributed multi-agent manufacturing scenario of new energy vehicles within the circular economy[J]. Computers & Industrial Engineering,2025,201:110889.
[8] HAN Wei,LI Jun,WANG Sheng,et al. Geological remote sensing interpretation using deep learning feature and an adaptive multisource data fusion network[J]. IEEE Transactions on Geoscience and Remote Sensing,2022,60:4510314.
[9] DUNJKO V,BRIEGEL H J. Machine learning & artificial intelligence in the quantum domain:A review of recent progress[J]. Reports on Progress in Physics Physical Society,2018,81(7):074001. doi: 10.1088/1361-6633/aab406
[10] 窦杰,向子林,许强,等. 机器学习在滑坡智能防灾减灾中的应用与发展趋势[J]. 地球科学,2023,48(5):1657 − 1674. [DOU Jie,XIANG Zilin,XU Qiang,et al. Application and development trend of machine learning in landslide intelligent disaster prevention and mitigation[J]. Earth Science,2023,48(5):1657 − 1674. (in Chinese with English abstract)]
DOU Jie, XIANG Zilin, XU Qiang, et al. Application and development trend of machine learning in landslide intelligent disaster prevention and mitigation［J］. Earth Science, 2023, 48（5）: 1657 − 1674. （in Chinese with English abstract）
[11] 许强. 对地质灾害隐患早期识别相关问题的认识与思考[J]. 武汉大学学报(信息科学版),2020,45(11):1651 − 1659. [XU Qiang. Understanding and consideration of related issues in early identification of potential geohazards[J]. Geomatics and Information Science of Wuhan University,2020,45(11):1651 − 1659. (in Chinese with English abstract)]
XU Qiang. Understanding and consideration of related issues in early identification of potential geohazards［J］. Geomatics and Information Science of Wuhan University, 2020, 45（11）: 1651 − 1659. （in Chinese with English abstract）
[12] 贾雨霏,魏文豪,陈稳,等. 基于SOM-I-SVM耦合模型的滑坡易发性评价[J]. 水文地质工程地质,2023,50(3):125 − 137. [JIA Yufei,WEI Wenhao,CHEN Wen,et al. Landslide susceptibility assessment based on the SOM-I-SVM model[J]. Hydrogeology & Engineering Geology,2023,50(3):125 − 137. (in Chinese with English abstract)]
JIA Yufei, WEI Wenhao, CHEN Wen, et al. Landslide susceptibility assessment based on the SOM-I-SVM model［J］. Hydrogeology & Engineering Geology, 2023, 50（3）: 125 − 137. （in Chinese with English abstract）
[13] LI Wenjuan,FANG Zhice,WANG Yi. Stacking ensemble of deep learning methods for landslide susceptibility mapping in the Three Gorges Reservoirarea,China[J]. Stochastic Environmental Research and Risk Assessment,2022,36(8):2207 − 2228. doi: 10.1007/s00477-021-02032-x
[14] 王惠卿,肖锐铧,刘艳辉,等. 近20年地质灾害气象风险预警研究文献图谱可视化分析[J]. 中国地质灾害与防治学报,2024,35(2):10 − 20. [WANG Huiqing,XIAO Ruihua,LIU Yanhui,et al. Visualization analysis of research literature on early warning of geo-hazards based on meteorological factors in the past 20 years[J]. The Chinese Journal of Geological Hazard and Control,2024,35(2):10 − 20. (in Chinese with English abstract)]
WANG Huiqing, XIAO Ruihua, LIU Yanhui, et al. Visualization analysis of research literature on early warning of geo-hazards based on meteorological factors in the past 20 years［J］. The Chinese Journal of Geological Hazard and Control, 2024, 35（2）: 10 − 20. （in Chinese with English abstract）
[15] 揭鸿鹄,蒋水华,常志璐,等. 融合多源信息的降雨入渗边坡概率反分析及可靠度预测[J]. 中国地质灾害与防治学报,2024,35(1):28 − 36. [JIE Honghu,JIANG Shuihua,CHANG Zhilu,et al. Probabilistic inverse-analysis and reliability prediction of rainfallinduced landslides for slope with multi-source information[J]. The Chinese Journal of Geological Hazard and Control,2024,35(1):28 − 36. (in Chinese with English abstract)]
JIE Honghu, JIANG Shuihua, CHANG Zhilu, et al. Probabilistic inverse-analysis and reliability prediction of rainfallinduced landslides for slope with multi-source information［J］. The Chinese Journal of Geological Hazard and Control, 2024, 35（1）: 28 − 36. （in Chinese with English abstract）
[16] TONG Bin,LI Yuan,YANG Xudong,et al. The development and application of China national landslide database and information system[J]. Arabian Journal of Geosciences,2021,14(6):448. doi: 10.1007/s12517-021-06825-w
[17] 李振洪,宋闯,余琛,等. 卫星雷达遥感在滑坡灾害探测和监测中的应用:挑战与对策[J]. 武汉大学学报(信息科学版),2019,44(7):967 − 979. [LI Zhenhong,SONG Chuang,YU Chen,et al. Application of satellite radar remote sensing to landslide detection and monitoring:Challenges and solutions[J]. Geomatics and Information Science of Wuhan University,2019,44(7):967 − 979. (in Chinese with English abstract)]
LI Zhenhong, SONG Chuang, YU Chen, et al. Application of satellite radar remote sensing to landslide detection and monitoring: Challenges and solutions［J］. Geomatics and Information Science of Wuhan University, 2019, 44（7）: 967 − 979. （in Chinese with English abstract）
[18] 葛大庆,戴可人,郭兆成,等. 重大地质灾害隐患早期识别中综合遥感应用的思考与建议[J]. 武汉大学学报(信息科学版),2019,44(7):949 − 956. [GE Daqing,DAI Keren,GUO Zhaocheng,et al. Early identification of serious geological hazards with integrated remote sensing technologies:Thoughts and recommendations[J]. Geomatics and Information Science of Wuhan University,2019,44(7):949 − 956. (in Chinese with English abstract)]
GE Daqing, DAI Keren, GUO Zhaocheng, et al. Early identification of serious geological hazards with integrated remote sensing technologies: Thoughts and recommendations［J］. Geomatics and Information Science of Wuhan University, 2019, 44（7）: 949 − 956. （in Chinese with English abstract）
[19] ZHANG Qi,WANG Teng. Deep learning for exploring landslides with remote sensing and geo-environmental data:Frameworks,progress,challenges,and opportunities[J]. Remote Sensing,2024,16(8):1344. doi: 10.3390/rs16081344
[20] WANG Haojie,ZHANG Limin,LUO Hongyu,et al. AI-powered landslide susceptibility assessment in Hong Kong[J]. Engineering Geology,2021,288:106103. doi: 10.1016/j.enggeo.2021.106103
[21] CASAGLI N,INTRIERI E,TOFANI V,et al. Landslide detection,monitoring and prediction with remote-sensing techniques[J]. Nature Reviews Earth & Environment,2023,4(1):51 − 64.
[22] MOHAN A,SINGH A K,KUMAR B,et al. Review on remote sensing methods for landslide detection using machine and deep learning[J]. Transactions on Emerging Telecommunications Technologies,2021,32(7):e3998. doi: 10.1002/ett.3998
[23] HADID A,CHAKRABORTY T,BUSBY D. When geoscience meets generative AI and large language models:Foundations,trends,and future challenges[J]. Expert Systems,2024,41(10):e13654. doi: 10.1111/exsy.13654
[24] WANG Siqin,HU Tao,XIAO Huang,et al. GPT,large language models (LLMs) and generative artificial intelligence (GAI) models in geospatial science:A systematic review[J]. International Journal of Digital Earth,2024,17(1):1 − 5.
[25] POO M M. Reflections on DeepSeek’s breakthrough[J]. National Science Review,2025,12(3):nwaf044. doi: 10.1093/nsr/nwaf044
[26] CONROY G,MALLAPATY S. How China created AI model DeepSeek and shocked the world[J]. Nature,2025,638(8050):300 − 301. doi: 10.1038/d41586-025-00259-0
[27] BUTLER K T,DAVIES D W,CARTWRIGHT H,et al. Machine learning for molecular and materials science[J]. Nature,2018,559(7715):547 − 555. doi: 10.1038/s41586-018-0337-2
[28] 马雷鸣. 天气预报中的人工智能技术进展[J]. 地球科学进展,2020,35(6):551 − 560. [MA Leiming. Development of artificial intelligence technology in weather forecast[J]. Advances in Earth Science,2020,35(6):551 − 560. (in Chinese with English abstract)]
MA Leiming. Development of artificial intelligence technology in weather forecast［J］. Advances in Earth Science, 2020, 35（6）: 551 − 560. （in Chinese with English abstract）
[29] 周创兵,姚池,杨建华,等. 数智技术赋能库坝系统安全管控探讨[J]. 中国水利,2025(1):7 − 14. [ZHOU Chuangbing,YAO Chi,YANG Jianhua,et al. Safety control of reservoir dam system with digital intelligence technology[J]. China Water Resources,2025(1):7 − 14. (in Chinese with English abstract)] doi: 10.3969/j.issn.1000-1123.2025.01.002
ZHOU Chuangbing, YAO Chi, YANG Jianhua, et al. Safety control of reservoir dam system with digital intelligence technology［J］. China Water Resources, 2025（1）: 7 − 14. （in Chinese with English abstract） doi: 10.3969/j.issn.1000-1123.2025.01.002
[30] 董少群,曾联波,车小花,等. 人工智能在致密储层裂缝测井识别中的应用[J]. 地球科学,2023,48(7):2443 − 2461. [DONG Shaoqun,ZENG Lianbo,CHE Xiaohua,et al. Application of artificial intelligence in fracture identification using well logs in tight reservoirs[J]. Earth Science,2023,48(7):2443 − 2461. (in Chinese with English abstract)]
DONG Shaoqun, ZENG Lianbo, CHE Xiaohua, et al. Application of artificial intelligence in fracture identification using well logs in tight reservoirs［J］. Earth Science, 2023, 48（7）: 2443 − 2461. （in Chinese with English abstract）
[31] RUAN Yixiang,LU Chenyin,XU Ning,et al. An automatic end-to-end chemical synthesis development platform powered by large language models[J]. Nature Communications,2024,15(1):10160. doi: 10.1038/s41467-024-54457-x
[32] 李郎平,兰恒星. 滑坡运动路径复杂度研究:综述与展望[J]. 地球科学,2022,47(12):4663 − 4680. [LI Langping,LAN Hengxing. Complexities of landslide moving path:A review and perspective[J]. Earth Science,2022,47(12):4663 − 4680. (in Chinese with English abstract)]
LI Langping, LAN Hengxing. Complexities of landslide moving path: A review and perspective［J］. Earth Science, 2022, 47（12）: 4663 − 4680. （in Chinese with English abstract）
[33] BHUYAN K,RANA K,FERRER J V,et al. Landslide topology uncovers failure movements[J]. Nature Communications,2024,15(1):2633. doi: 10.1038/s41467-024-46741-7
[34] 云南省人民政府. 云南镇雄“1•22”山体滑坡地质灾害调查报告[R]. 2024. [People’s Government of Yunnan Province. Investigation report on the geological disaster of mountain sliding on January 22nd in Zhenxiong,Yunnan Province[R]. 2024. (in Chinese)]
People’s Government of Yunnan Province. Investigation report on the geological disaster of mountain sliding on January 22nd in Zhenxiong, Yunnan Province［R］. 2024. （in Chinese）
-
图(7)
表(1)
计量
- 文章访问数: 45
- PDF下载数: 5
- 施引文献: 0