盾构大数据预处理方法研究

Research of Shield Big Data Preprocessing

在盾构法隧道施工过程中,合理设置盾构机运行参数,确保盾构机能够按照一定速度掘进,是整个工程流程控制中非常重要的工作。针对现有的盾构参数预测方法缺少对数据系统性的分析和处理,导致预测效果不及预期的情况,文章提出了一整套新的数据预处理流程。该流程分为数据分析和数据处理两个阶段,第一阶段通过主成分分析、皮尔森相关性系数分析进行特征筛选,第二阶段通过插值平滑、卷积平滑进行数据平滑。经过这两个阶段的原始数据处理后,能够在盾构产生的海量数据中提取出特征更明显、价值更高的数据。通过选取真实盾构机数据集进行对比实验,结果表明,所提出的数据预处理流程,能够有效提高盾构机参数预测模型的准确率。

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 predic tion method for shield parameters,a full set of new data preprocess flow is proposed.This new preprocess flow is di vided into two stages of data analysis and data processing,the first stage involves feature selection by analyzing prin cipal 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 con ducted by selecting real data sets,and the result shows that the proposed data preprocessing flow can effectively im prove the accuracy of shield parameter prediction model.