Aiming at the international frontier issue of structural deformation in hydropower facilities located in high mountains and valleys under extremely complex conditions during long-term operation, and based on the analysis of the causes of dam deformation characteristics and the study of the long-term operational characteristics of hydropower station dams, a generative intelligent analysis method combining large language models (LLMs) with Markov probabilistic logic generative neural networks has been proposed. This method leverages the powerful semantic understanding and generative capabilities of LLMs to optimize the feature extraction and correlation analysis of hydropower station dam deformation data, thereby enhancing the model accuracy in addressing the complex interactions of multiple factors under extreme conditions. This method was applied to the main project of the Dadu River Pubugou Hydropower Station dam infrastructure, which is affected by extreme weather, geological disasters, and sudden changes in turbulent flow of the Dadu River. It is used for the precise prediction and early warning of potential major hazards, such as the displacement of the dam’s main structure, special skeletons, and large-scale equipment. Extensive experimental verification has shown that under extremely complex conditions, the Markov probabilistic logic generative neural network combined with large language models and knowledge graphs can accurately identify the non-uniform deformation characteristics of hydropower station dam infrastructure and uncover the interrelationships between the deformation processes of various dam parts due to internal force interactions. Compared with the knowledge graph embedding methods, the overall accuracy of the results was significantly improved. This breakthrough provides a reference for the structural design and material selection of various parts for constructing of future hydropower stations in China.
本节介绍了用于分析水电站异常情况的三个关键技术。首先是实时监测网络的搭建,通过分布于大坝不同部位的传感器和通信设备收集位移、形变等数据构建一个复杂的监测网络对大坝各部位实施状况进行监测;然后是马尔可夫概率逻辑生成式神经网络的结构和应用,通过将大坝位移、形变和地壳运动数据相结合,以三元组的形式进行联合分布建模,并通过变分期望最大化算法进行训练;最后是BERT(Bidirectional Encoder Representations from Transformers)模型在智能分析方法中的优化,通过大语言模型与水电站监测数据以及马尔可夫概率逻辑神经网络相结合,提高网络对变形数据的理解能力以及对复杂因素关联性的挖掘能力,为大坝形变分析提供更高精度的预测和推理支持。
WANGS W, BAOT F. Characteristic analysis of deformation time-varying effect of super high concrete dams[J]. Journal of Changzhou University (Natural Science Edition), 2018, 30(6): 79-86. DOI: 10.3969/j.issn.2095-0411.2018.06.012(Ch ).
ZHIB, ZHAOW B, DENGZ J, et al. Detection technology for concrete crack depth using steady-state surface waves [C]//Proceedings of the 2019 Annual Conference of the Chinese Hydraulic Engineering Society, Volume 3. 2019: 241-245. DOI:10.26914/c.cnkihy.2019.097156(Ch ).
JIANGF T, CHENG M. Ultrasonic detection of crack depth on dam concrete surface[J]. Water Resources and Hydropower Engineering, 1981, 12(8): 55-58 (Ch).
CHENGB, YANY Y. Measurement of surface crack width based on digital image processing[J]. Journal of Chifeng University (Natural Science Edition), 2018, 34(10):96-98. DOI:10.13398/j.cnki.issn1673-260x.2018.10.032(Ch ).
[11]
PRASANNAP, DANAK J, GUCUNSKIN, et al. Automated crack detection on concrete bridges[J]. IEEE Transactions on Automation Science and Engineering, 2016, 13(2): 591-599. DOI:10.1109/TASE.2014.2354314 .
[12]
FENGC C, ZHANGH, WANGH R, et al. Automatic pixel-level crack detection on dam surface using deep convolutional network[J]. Sensors, 2020, 20(7): 2069. DOI:10.3390/s20072069 .
ZHOUY X, HUANGB, YUY J, et al. Domain information self-augmenting extreme random tree model applied to time series data prediction[J]. Journal of Wuhan University (Natural Science Edition), 2025,71(2): 253-265. DOI:10.14188/J.1671-8836.2023.0183(Ch ).
HUANGB, KANGF, TANGY. A real-time detection method for concrete dam cracks based on an object detection algorithm[J]. Journal of Tsinghua University (Science and Technology), 2023, 63(7): 1078-1086. DOI: 10.16511/j.cnki.qhdxxb.2023.26.013(Ch ).
OU B, WUB B, YUANJ, et al. LSTM-based deformation prediction model of concrete dams[J]. Advances in Science and Technology of Water Resources, 2022, 42(1): 21-26. DOI: 10.3880/j.issn.10067647.2022.01.003(Ch ).
CHENB, CHENW N, GONGX X, et al. Prediction model for deformation of concrete dams in cold regions considering hysteresis and reduction effects from environment[J]. Chinese Journal of Geotechnical Engineering, 2021, 43(12): 2315-2322. DOI: 10.11779/CJGE202112019(Ch ).
HEJ P. Theory and application of dam safety monitoring[M]. Beijing: China Water & Power Press, 2010(Ch). DOI: 10.4028/www.scientific.net/amm.36.182 .
[23]
QUM, TANGJ. Probabilistic logic neural networks for reasoning[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems. New Yoek: ACM, 2019: 7712-7722. DOI: 10.5555/3454287.3454980 .
HUANGB, WUS A, WANGW G, et al. KG-LLM-MCom: A survey on integration of knowledge graph and large language model[J]. Journal of Wuhan University (Natural Science Edition), 2024, 70(4): 397-412. DOI: 10.14188/j.1671-8836.2024.0040(Ch ).