摘要
为实现隧道涌水量的高精度预测,在查阅相关资料的基础上,以多种单项预测模型为基础,构建了隧道涌水量的优化组合预测模型。首先,利用遗传算法、粒子群算法和最小二乘法构建了多种单项优化预测模型;其次,在累加叠加和累乘叠加思路的基础上,提出利用整体组合权值和局部组合权值构建出改进后的组合预测模型,以实现隧道涌水量的逐步优化组合预测;最后,再利用重标极差分析(rescaled range analysis,简称R/S分析)判断隧道涌水量序列的发展趋势,以佐证前述组合模型预测结果的准确性。实例结果表明:各类单项预测模型的参数优化可以有效提高预测精度,且组合预测结果的相对误差小于2%,较单项预测模型具有相对更优的预测效果;同时,在组合预测过程中,累加叠加和累乘叠加的预测精度及稳定性均相当,以累乘叠加的预测效果相对略优,但差异不大;另外,R/S分析结果的Hurst指数大于0.5,得出隧道涌水量呈下降趋势,与预测分析结果一致,验证了前述预测模型的有效性,为隧道涌水量预测研究提供了一种新的思路。
In order to achieve high-precision prediction of tunnel water inflow,an optimal combination prediction model of tunnel water inflow is constructed on the basis of studying relevant data and a variety of single prediction models.Firstly,genetic algorithm,particle swarm optimization and least square method are used to construct a variety of single optimization prediction models.Secondly,based on the idea of multiplicative superposition and cumulative superposition,an improved combination forecasting model is proposed by using global and local combination weights to realize the gradual optimization combination prediction of tunnel water inflow.Finally,the rescaled range analysis(R/S analysis)is used to judge the development trend of the tunnel water inflow series,so as to verify the accuracy of the foregoing combined model prediction results.The results show that the parameter optimization of various single prediction models can effectively improve the prediction accuracy,and the relative error of the combined prediction results is less than 2%,which has a better prediction effect than that of single prediction models.At the same time,in the process of combined prediction,the prediction accuracy and stability of the accumulative superposition and the multiplicative superposition are very close,and the prediction effect of multiplicative superposition is slightly better.In addition,the Hurst index of R/S analysis results is greater than 0.5,which indicates that the tunnel water inflow is decreasing,and it is consistent with the prediction results.The validity of the foregoing prediction model is verified,so a new idea is provided for the study of tunnel water inflow prediction.
作者
周玲
杨广庆
杨青潮
ZHOU Ling;YANG Guangqing;YANG Qingchao(Xi'an Siyuan University,Xizan 710038,China;Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Shandong Transport Vocational College,Weifang 261206,China)
出处
《武汉大学学报（工学版）》
CAS
CSCD
北大核心
2020年第10期875-882,共8页
Engineering Journal of Wuhan University
基金
国家自然科学基金资助项目(编号:51378322,51709175,51879246,51408439)
陕西省2019年度虚拟仿真实验教学项目(单桩竖向抗压静载虚拟仿真实验)
西安思源学院“基于BIM技术虚拟仿真实验教学中心”
“土木专业BIM与虚拟仿真实践课程教学团队”经费资助。
关键词
岩溶隧道
组合预测
涌水量预测
R/S分析
趋势判断
karst tunnel
combination prediction
water inflow prediction
R/S analysis
trend judgment
