Prediction of hard rock TBM penetration rate using random forests

• Published in: The 27th Chinese Control and Decision Conference (2015 CCDC) pp. 3716 - 3720
• Main Authors: Hu Tao, Wang Jingcheng, Zhang Langwen
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2015
• Subjects: Accuracy; prediction; Prediction algorithms; Predictive models; Random Forests; Regression tree analysis; Rocks; TBM Penetration Rate; Vegetation
• ISSN: 1948-9439
• DOI: 10.1109/CCDC.2015.7162572

Penetration Rate is an important parameter of hard rock tunnel boring machine (TBM) in tunneling project. The prediction accuracy of Penetration Rate has great effect on the successful accomplishment of tunneling project. The aim of this paper is to predict the penetration rate and rank the importance of rock mass properties via Random Forests algorithm. Random Forests is a high accuracy regression algorithm, which is not prone to over fitting and has good tolerance to outliers and noise. A database including actual, measured penetration rates and several rock mass properties are established by using the data collected from a real tunnel project. Based on the database, we use random forests algorithm to model the penetration rate of the tunnel project. The simulation results show that the random forest based prediction model has better predictive accuracy and can sort the features of rock mass properties (UCS, BTS, PSI, DPW and \alpha) by the importance.