Abstract：
During the construction of shield tunnel, reasonable determination of operating parameters of shield ma? chine is significant regarding process control to ensure that shield machine can drive at a certain speed. Aiming at the unsatisfactory prediction results due to the lacking of systematic data analysis and processing of existing prediction method for shield parameters, a full set of new data preprocess flow is proposed. This new preprocess flow is divided into two stages of data analysis and data processing, the first stage involves feature selection by analyzing principal component and Pearson correlation coefficient, and the second stage involves data smoothing by interpolation smoothing and convolution smoothing. After processing of original data, the data with more obvious features and higher valued can be extracted from the massive data generated by the shield machine. Contrast experiment is conducted by selecting real data sets, and the result shows that the proposed data preprocessing flow can effectively improve the accuracy of shield parameter prediction model.
ZHANG Longguan1 DUAN Wenjun1 ZHUANG Yuanshun1 ZHANG Zhonghua1 LIU Suimei1 ZHANG Feng2
.Research of Shield Big Data Preprocessing[J] MODERN TUNNELLING TECHNOLOGY, 2020,V57(2): 34-41
2020, Vol. 57 