基于人工智能算法的隧道锚承载能力评价  

Bearing Capacity Evaluation of Tunnel-Type Anchorage Based on Artificial Intelligent Algorithm

作　　者：王中豪[1] 郭喜峰[1] 杨星宇 WANG Zhonghao;GUO Xifeng;YANG Xingyu(Chongqing Branch of Yangtze River Scientific Research Institute,Chongqing 400026,China)

机构地区：[1]长江科学院重庆分院,重庆400026

出　　处：《西南交通大学学报》2021年第3期534-540,共7页Journal of Southwest Jiaotong University

基　　金：国家自然科学基金(42002276);湖北省自然科学基金(2019CFB258)。

摘　　要：针对隧道锚承载能力评价合理的解析计算公式缺乏、模型试验测试方法耗时费力、数值模拟可靠性不佳的问题,提出了一种人工智能化隧道锚承载能力预测方法.从隧道锚受力传力过程出发,分析了影响承载能力的因子,确定了承载能力评价指标体系;基于最小二乘支持向量机(least squares support vector machines,LSSVM)强大的学习预测能力和粒子群优化(particle swarm optimization,PSO)算法良好的优化效果,建立了承载能力非线性映射PSO-LSSVM模型;将收集到的17个隧道锚工程案例作为输入样本对模型进行了训练,获得了核函数参数和惩罚系数的最优组合为(1,500).将该模型应用于某大桥隧道锚承载能力的预测,预测结果为10.2P(1P为1倍设计荷载);通过与现场缩尺模型试验和数值模拟方法综合研究确定的承载能力为11.0P对比,结果表明:预测结果略低,但两者结果非常接近,说明该模型的预测结果合理可靠且偏于保守,预测效果较为理想.At present,the tunnel-type anchorage is short of reasonable analytical formula for the evaluation of the bearing capacity,the model test is time-consuming and labor-consuming,and its numerical simulation has poor reliability.To handle the above problems,an artificial intelligence method is presented for predicting the bearing capacity of the tunnel-type anchorage.Starting from its force transmission process,the factors influencing the bearing capacity are analyzed and the evaluation index system of bearing capacity has been determined.Then,given the strong learning prediction ability of least squares support vector machines(LSSVM)and excellent performance of particle swarm optimization(PSO),a PSO-LSSVM model with nonlinear mapping of bearing capacity is established.After training the model with 17 cases of the tunnel-type anchorage as input samples,the optimal combination of kernel parameters and penalty coefficients is determined to be(1,500).Finally,the model is used to predict the bearing capacity of a bridge tunnel-type anchorage and the prediction result is determined as 10.2P.The comparison with the bearing capacity result of 11.0P that is determined by the comprehensive study of the scale model test and numerical simulation method,demonstrate that the predicted result is slightly lower but very close to the result of other method.This also shows that the prediction results of the model are reasonable,reliable and conservative,and the prediction effect is desirable.

关 键 词：隧道锚 承载能力 最小二乘支持向量机 粒子群优化算法 

分 类 号：U45[建筑科学—桥梁与隧道工程]