A Real - time Shield Attitude Deviation Prediction Method Based on Data Drive

In the process of shield construction, the control effect of deviation between shield machine and design axis is an important factor that affects the quality of tunnel construction. However, up to now, the traditional geometric method is still used in the actual construction process to predict the variation of shield deviation, the influence of various load systems is ignored. Based on the massive construction data generated by shield tunneling and the characteristics of shield tunneling, a real-time prediction method of shield attitude deviation driven by data is proposed in this paper. Based on the massive construction data and the characteristics of shield tunneling, a real - time shield attitude deviation prediction method based on data drive is proposed in this paper. This method firstly reduces noise and stabilizes the original construction data based on empirical mode decomposition algorithm, and then uses driving technology to mine the relationship between shield deviation change and shield machine load and surrounding soil pressure, finally establishes the attitude deviation prediction model. This method has been applied to the construction site of the 9th section of Shanghai line 14 for following experiments. The experimental results have verified that this method has better performance in predicting shield attitude deviation. This work also shows the potential of data-driven technology in realizing shield automatic correction control.