赵光祖, 王亚旭, 李尧,等.基于优化BP神经网络的TBM性能预测[J].河南理工大学学报（自然科学版）,2020,39(5):139-145.

ZHAO G Z, WANG Y X, LI Y,et al.Prediction of TBM performance based on optimized BP neural network[J].Journal of Henan Polytechnic University(Natural Science) ,2020,39(5):139-145.

基于优化BP神经网络的TBM性能预测

赵光祖1, 王亚旭1, 李尧1, 徐受天2, 陈帅2

1.山东大学 岩土与结构工程研究中心，山东 济南 250061;2.中铁工程装备集团有限公司，河南郑州 450016

摘要:由于隧道掘进机（tunnel boring machine， TBM）掘进速度与机器参数、岩体参数之间的非线性关系复杂，难以准确预测，为了构建可靠的TBM性能预测模型，分析TBM掘进速度的主要影响因素，提出应用模拟退火算法（SA）和遗传算法（GA）优化BP神经网络的TBM性能预测模型，并使用吉林引松供水工程的TBM数据库对GA-BP模型和SA-BP模型进行训练测试。结果表明，与传统BP神经网络方法相比，优化后的模型预测泛化性更好，且精度明显提高。优化后的BP神经网络能在一定程度上克服易陷入局部最优的缺陷，应用于TBM性能预测具 有良好表现。

关键词:隧道掘进机;掘进速度;岩体参数;神经网络

doi:10.16186/j.cnki.1673-9787.2020.5.20

基金项目:国家重点基础研究发展计划项目（2015CB058101 ）；国家自然科学基金重点项目（51739007 ）；国家重点研发计划专题项目 （2016YFC0401805 ）；山东省重点研发计划项目（Z135050009107）；山东大学基本科研业务费专项项目（2018GN020）

收稿日期:2019/11/31

修回日期:2020/02/09

出版日期:2020/09/15

Prediction of TBM performance based on optimized BP neural network

ZHAO Guangzu1, WANG Yaxu1, LI Yao1, XU Shoutian2, CHEN Shuai2

1.Geotechnical and Structural Engineering Research Center， Shandong University， Jinan  250061 ， Shandong， China;2.China Railway Engineering Equipment Group Co. ，Ltd. ，Zhengzhou  450016 ，Henan， China

Abstract:Due to TBM penetration velocity having complex nonlinear relationship with machine parameters and rock mass parameters， it is difficult to predict the TBM performances accurately. In order to construct a reliable TBM performance prediction model， the main influencing factors of TBM penetration velocity were discussed， and TBM performance prediction models were proposed based on BP neural network optimized by simulated annealing algorithm and genetic algorithm， and GA-BP model and the SA-BP model were trained and tested based on the TBM database of Jilin Songhua River Water Supply Project. Compared with the traditional BP neural network， prediction of the optimized model had better generalization and significantly improved accuracy. The results showed that the BP neural network optimized by simulated annealing and genetic algorithm could overcome the drawback that were easy to fall into local optimum to some extent， and it had a good performance on TBM performance prediction.

Key words:tunnel boring machine;penetrate velocity;rock mass parameter;neural network

基于优化BP神经网络的TBM性能预测_赵光祖.pdf