Prediction and Control of Shield Tunneling Attitude Based on Automatic Deep Learning
The combination of automatic deep learning(AutoDL)algorithm and multi-objective optimization algorithm is proposed to realize the data-driven attitude deviation control guidance,which is used for the prediction and control of shield tunneling attitude in order to solve the problems of high execution difficulty,high cost and low efficiency faced in the prediction of the existing shield tunneling attitude,and can be used to automatically and accurately predict the dynamic change trend of shield tunneling attitude with the engineering progress.The multi-objective optimization algorithm is implemented according to the construction state of shield tunneling machine quickly and automatically to search for the optimum strategy,adjust the suitable shield tunneling parameters in real time and reduce the reliance on field operator experience and subjective judgment.Taking the Tunnel Line A Project in Chongming Island-Changxing Island-Pudong New District No.5 Ditch LNG Station Pipeline Project of the natural gas main pipeline network in Shanghai as an example,the superiority of the algorithm framework is demonstrated.The research results are helpful to lower the threshold for deep learning to enter the field of intelligent shield control and promote the development of intelligent shield.
automatic deep learningtime windowshield attitudeparticle swarm optimization algorithmmulti-objective optimization algorithm
NETL
NSTL
万方数据
维普
