TBM掘进速度预测模型研究

Prediction Models of the TBM Penetration Rate

随着机械制造技术的进步,全断面岩石隧道掘进机被广泛应用于深、长、大隧洞的开挖,鉴于全断面岩石隧道掘进机(TBM)对地质条件十分敏感,且其前期投入巨大,采用合适的方法、准确地预测TBM掘进速度对TBM施工的进度安排和成本估计十分重要.基于纽约皇后NO.3隧道153组实测岩体参数(UCS、PSI、DWP、BTS、α)和TBM掘进速度(PR),分别采用BP神经网络和CART算法建立TBM掘进速度预测模型,与已有预测模型对比发现,CART预测模型预测精度更高更易于不同工程相互借鉴,且在部分岩体参数缺失的情况下也能对TBM掘进速度进行有效预测.

With the development of manufacturing technology,the TBM has been widely used in tunneling. Since the performance of TBM is sensitive to the geological condition,an accurately prediction about TBM penetration rate is very important for the construction schedule and cost estimation. In the study,the BP neural network and CART algorithm were used for the prediction of penetration rate based on the measured data of the Queens Water Tunnel # 3. According to the comparison with the existing models,it is found that the CART model is more accurate and easier to be applied in different project. Furthermore,the CART model can be used for the prediction of PR in the case that one or several rock mass parameters are deficient.