Geological Advance Prediction And Operation Parameters Optimization Based On Data Mining

To ensure the safe construction,it is necessary to carry out geological forecast to predict the geological conditions of the tunnel before and during the tunneling,and to correct the information of rock ballast after tunneling for further research and analysis,because of its concealment during TBM tunneling.There are different geological forecast methods for different stages of TBM construction,limited by their own shortcomings,which lead to the bottleneck of geological prediction accuracy,construction efficiency or prediction timeliness.The operating parameters of TBM should be continuously adjusted to adapt to the current geological conditions,as the known tunnel geology changes.Unsuitable selection of operating parameters will lead to poor loading conditions.Its correctness depends on the professional knowledge of the driver,which requires high professionalism and consumes time.For the disadvantages of traditional TBM driving and geological forecast methods,it is necessary to clarify the rock-machine mapping mechanism and scientifically ensure the safe of TBM driving,with the development of the TBM intelligent and safe driving plan.In view of the shortage of geological forecast methods in different construction stages,for the purpose of identification of surrounding rocks,data mining method was used to establish surrounding rock grades real-time prediction model and surrounding rock grades advance prediction model respectively to realize the intelligent recognition of TBM tunneling geology.Taking the minimum load as the objective,the operation parameters were optimized to ensure the safe tunneling and reduce the professional requirements of driving.The closed-loop safety guarantee from the geological intelligent recognition to the safe driving decision was achieved.The specific research contents are as follows:1)Relying on the revision data of a water diversion project,the geology was divided into five grades of surrounding rock according to surrounding rock classification standard.The KNN,decision tree and DNN algorithms were used to establish the surrounding rock grades real-time prediction model between the tunneling parameters and surrounding rock grades.Real-time identification of surrounding rock grades with classification accuracy of 0.99 was accomplished,which solved the problems of low timeliness and high professionalism caused by the geological analysis and correction in the later stage of the project in terms of surrounding rock grades identification.2)Based on the LSTM algorithm,the parameters advance prediction model was established,and in combination with the surrounding rock grades real-time prediction model,the surrounding rock grades advance prediction model was established.Realizing the function of predicting the surrounding rock grade after 70 minutes in advance under the premise of meeting a certain classification accuracy(0.85)was accomplished.In terms of surrounding rock grades identification,it solved the problems of inaccurate geological prediction and difficult coring in the early stage of the project,as well as the problem of low project efficiency due to the need to shut down for advance geological forecast in the middle stage of the project.3)Ridge regression and DNN regression algorithms were used to establish load prediction models between operating parameters and loads under different surrounding rock grades.The total thrust and cutter head torque prediction MAPEs are below 10% and 29%,respectively.On the premise of ensuring the driving efficiency,the operating parameters were optimized with the decision goal of minimizing the cutter head torque or total thrust.By comparing the ratio of optimized response torque to rated torque and the ratio of optimized response total thrust to maximum total thrust under different surrounding rock grades,the main limiting factors that influence TBM safe driving were selected.The operating parameters corresponding to the main limiting factors were selected as the recommended operating parameters to achieve the purpose of reducing the response values of the main limiting factors at the expense of the secondary factors.Combined with geological prediction,it will guide the TBM driving under different surrounding rocks.