A Data Mining Based Prediction Model for Penetration Rate

Tunnel boring machines (TBMs) are increasingly being used to construct tunnels because of their efficiency and safety benefits. Predicting the penetration rate of TBMs can clearly guide project schedule planning and cost control, making boring more efficient and economical. Based on a TBM tunnel of the 2nd stage water transfer project in Northern Xinjiang, the distribution of geological and boring parameters was investigated using mathematical and statistical method. The influence of geological and boring parameters on penetration rate was analyzed, and optimal regression formulas were established. Based on the field data, the applicability of six existing models was discussed, and a multi-factor penetration rate prediction model with geological parameters and boring parameters as independent variables was developed using a clustering method of data mining. In addition, the prediction effectiveness of the model was verified. The research results show that the penetration rate was affected by the geological parameters and boring parameters. Using data mining clustering method, the surrounding rock can be classified according to properties such as uniaxial compressive strength, fragmentation degree, and quartz content. The clustered prediction model based on data mining can effectively predict the penetration rate with an absolute error of less than 10mm/min. In the application of the model, a database of corresponding project and similar projects can be established and combined with the model development method proposed in this paper to develop a clustered penetration rate prediction model that is more suitable for the corresponding engineering problems. © 2021, The Editorial Board of Journal of Basic Science and Engineering. All right reserved.