龙滩水电站地下洞室围岩变形的智能化预测方法  

Intelligent deformation predicting methods for underground chambers of Longtan hydropower station

作　　者：张炎[1,2] 朱珍德[1,2,3] 朱姝[1] 赵长冰[1] 

机构地区：[1]河海大学岩土工程科学研究所,南京210098 [2]河海大学岩土力学与堤坝工程教育部重点实验室,南京210098 [3]中国矿业大学深部岩土力学与地下工程国家重点实验室,徐州221008

出　　处：《水力发电学报》2013年第2期246-251,共6页Journal of Hydroelectric Engineering

基　　金：国家重点基础研究发展计划(973计划):"大型水利水电工程高陡边坡全生命周期性能演化与安全控制(2011CB013500)"第四子课题<复杂环境下工程边坡岩体时效力学特性>(编号:2011CB013504);深部岩土力学与地下工程国家重点实验室开放基金课题:"深埋地下洞室软弱围岩卸荷变形破坏过程细观试验研究(SKLGDUEK0902)"

摘　　要：龙滩水电站地下洞室群规模巨大,洞室纵横交错,设计中还无法全面准确地预计到施工与运行中洞室围岩的变形与稳定情况,对地下洞室群进行系统监控和变形趋势预测很有必要。根据监测数据建立学习样本,构建基于粒子群优化算法的BP神经网络智能化预测模型,并将其应用于龙滩水电站地下主厂房D监测断面多点位移计M4D-5监测数据的模拟与预测。仿真结果表明,该PSO-BP模型预测结果稳定且精度较高,误差百分比大多能控制在2%以内,预测效果较好,模型可行。The underground chambers of Longtan hydropower station are featured with very huge scale and crisscross,and this brings about a great difficulty in estimation of the deformation and stability of their surrounding rock at the construction stage,so systematic monitoring and forecasting of the deformation trend is necessary.In this work,first a learning sample is established according to the monitoring data,and a BP neural network intelligent prediction model was developed adopting the particle swarm optimization algorithm.Then using this model,simulation and prediction is made for comparison with the data collected by the multi-point displacement meter located on the monitoring section D of the main house at this station.Results show that the prediction is stable and accurate,with most of the percentage errors below 2%.

关 键 词：岩石力学 变形预测 BP神经网络 地下硐室 龙滩水电站 粒子群优化算法 

分 类 号：TU454[建筑科学—岩土工程]