An Approach Integrating Dimensional Analysis and Field Data for Predicting the Load on Tunneling Machine

摘要

The forecast of machine performance has been widely discussed in recent years, where accurate predictions with common regularity could reduce costs and achieve time-efficient construction. This study uses the case of predicting the load (including the thrust and torque) acting on tunneling machine. Because most data-mining techniques are black-box models, which rely too much on specific data and cannot reveal the mechanics principle. A new approach is developed to deal with this problem, which combines data analysis with mechanics principle through dimensional analysis. Firstly, power relations between the soil properties, operational, structural parameters and the load are obtained. Then dimensionless model is established, which involves the dominant dimensionless groups representing operating status and geological conditions. At last, to overcome the problem of high correlation between independent variables, ridge estimation is performed. The obtained dimensionless model is in accordance with Krause Empirical Model for the power relationship between the load and the cutterhead diameter. Moreover, the earth pressure in the chamber is proportional to the load, which is consistent with previous experimental results. In addition to its prediction ability, the validation data from two urban subway projects show that, the proposed model can correctly indicating the geological variation during excavation.