Feedback on a shared big dataset for intelligent TBM PartⅠ:Feature extraction and machine learning methods

This review summarizes the research outcomes and findings documented in 45 journal papers using a shared tunnel boring machine(TBM)dataset for performance prediction and boring efficiency optimization using machine learning methods.The big dataset was col-lected during the Yinsong water diversion project construction in China,covering the tunnel excavation of a 20 km-section with 199 items of monitoring metrics taken with an interval of one second.The research papers were the result of a call for contributions during a TBM machine learning contest in 2019 and covered a variety of topics related to the intelligent construction of TBM.This review com-prises two parts.Part I is concerned with the data processing,feature extraction,and machine learning methods applied by the contrib-utors.The review finds that the data-driven and knowledge-driven approaches in extracting important features applied by various authors are diversified,requiring further studies to achieve commonly accepted criteria.The techniques for cleaning and amending the raw data adopted by the contributors were summarized,indicating some highlights such as the importance of sufficiently high fre-quency of data acquisition(higher than 1 second),classification and standardization for the data preprocessing process,and the appro-priate selections of features in a boring cycle.The review finds that both supervised and unsupervised machine learning methods have been utilized by various researchers.The ensemble and deep learning methods have found wide applications.Part I highlights the impor-tant features of the individual methods applied by the contributors,including the structures of the algorithm,selection of hyperparam-eters,and model validation approaches.

Feedback on a shared big dataset for intelligent TBM Part Ⅱ:Application and forward look

This review discusses the application scenarios of the machine learning-supported performance prediction and the optimization effi-ciency of tunnel boring machines(TBMs).The rock mass quality ratings,which are based on the Chinese code for geological survey,were used to provide"labels"suitable for supervised learning.As a result,the generation of machine prediction for rock mass grades reason-ably agreed with the ground truth documented in geological maps.In contrast,the main operational parameters,i.e.,thrust and torque,can be reasonably predicted based on historical data.Consequently,18 collapse sections of the Yinsong project have been successfully predicted by several researchers.Preliminary studies on the selection of the optimal penetration rate and cost were conducted.This review also presents a summary of the main achievements in response to the initiatives of the Lotus Pool Contest in China.For the first time,large and well-documented TBM performance data has been shared for joint scientific research.Moreover,the review discusses the technical problems that require further study and the perspectives in the future development of intelligent TBM construction based on big data and machine learning.

Real-time prediction of tunnel face conditions using XGBoost Random Forest algorithm

Real-time perception of rock conditions based on continuously collected data to meet the requirements of continuous Tunnel Boring Machine(TBM)construction presents a critical challenge that warrants increased attention.To achieve this goal,this paper establishes real-time prediction models for fractured and weak rock mass by comparing 6 different algorithms using real-time data collected by the TBM.The models are optimized in terms of selecting metric,selecting input features,and processing imbalanced data.The results demonstrate the following points.(1)The Youden's index and area under the ROC curve(AUC)are the most appropriate performance metrics,and the XGBoost Random Forest(XGBRF)algorithm exhibits superior prediction and generalization performance.(2)The duration of the TBM loading phase is short,usually within a few minutes after the disc cutter contacts the tunnel face.A model based on the features during the loading phase has a miss rate of 21.8%,indicating that it can meet the early warning needs of TBM construction well.As the TBM continues to operate,the inclusion of features calculated from subsequent data collection can continuously correct the results of the real-time prediction model,ultimately reducing the miss rate to 16.1%.(3)Resampling the imbalanced data set can effectively improve the prediction by the model,while the XGBRF algorithm has certain advantages in dealing with the imbalanced data issue.When the model gives an alarm,the TBM operator and on-site engineer can be reminded and take some necessary measures for avoiding potential tunnel collapse.The real-time predication model can be a useful tool to increase the safety of TBM excavation.