块石回填土地铁隧道TBM掘进速度的GA-BPNN预测模型

GA-BPNN Prediction Model for TBM Advance Speed in Metro Tunnelling in Backfills with Rock Blocks

TBM掘进速度的预测能为施工参数的最终确定、施工进度的合理安排、建设成本的预估提供理论依据,但TBM掘进速度常常呈现出高度非线性的特点,且受诸多因素影响,往往难以建立准确的预测模型。基于此,文章依托重庆轨道交通五号线北延伸段块石回填土地铁隧道工程,收集了该段的TBM掘进数据,基于深度学习遗传算法优化的反向传播神经网络GA-BPNN算法,提出了一种掘进速度预测模型,并与传统BPNN算法预测模型进行对比分析。实际工程应用效果表明,通过对样本数据进行预先训练,可精确完成掘进速度的预测;遗传算法优化后的预测模型可以对掘进速度进行较为精确的预测,且较优化前的BPNN算法预测模型在预测结果的绝对误差方面上有20%以上的性能提升。

The prediction of TBM advance speed can provide a theoretical basis for final determination of construction parameter,reasonable arrangement of construction progress and estimation of construction cost.However,TBM advance speed often presents the characteristics of high non-linearity and is affected by many factors.So,it is often difficult to establish an accurate prediction model.Therefore,in this paper,the TBM excavation data were collected on the basis of the Metro Tunnel Project of North Extension Section of Chongqing Rail Transit Line 5 in backfills with rock blocks.Then,based on the back propagation neural network GA-BPNN algorithm optimized by deep learning genetic algorithm,a prediction model of excavation speed was proposed and comparatively analyzed with the prediction model of traditional BPNN algorithm.Practical application results show that the prediction of tunneling speed could be accurately completed by pre-training the sample data;the prediction model optimized by the genetic algorithm could predict the tunneling speed more accurately,and its performance is improved by more than 20%in the absolute error of prediction results while compared with the prediction model of BPNN algorithm before optimization.