基于互信息与支持向量回归的盾构掘进载荷预测方法研究

掘进载荷是盾构施工中的重要控制量，直接关系着施工安全与效率。通过对掘进载荷影响因素的分析，建立了一种基于工程实测数据分析的掘进载荷特征选择及预测方法。首先，对工程实测数据进行极值归一化预处理，以降低不同参数间量纲和量级的差异产生的支配性影响；其次，通过参数分析和基于互信息的特征选择选取主要的影响参数作为输入；最后，通过支持向量回归（support vector regression，SVR）建立掘进载荷的预测模型，并结合天津地铁9号线盾构施工工程案例检验其预测表现。结果表明，所建立的掘进载荷预测方法能够在工程实测数据包含的众多影响参数中筛选出少量关键特征，实现对掘进载荷的合理预测。研究结果可以为盾构掘进参数的调控提供参考，也为具有众多参数的工程实测数据的分析提供一种思路。

Research on prediction method of driving load of shield machine based on mutual information and support vector regression

driving load is an important control parameter in shield construction, which is directly related to construction safety and efficiency. Through the analysis of influencing factors of driving load, a feature selection and prediction method of driving load based on the analysis of engineering measured data was established. Firstly, the engineering measured data were extreme valuenormalized and preprocessed to reduce the dominant influence caused by dimensional and magnitude differences between different parameters; secondly, through parameter analysis and feature selection based on mutual information, the main influence parameters were selected as input; finally, the prediction model of driving load was established by support vector regression (SVR), and its prediction performance was tested by the actual shield construction engineering case of Tianjin Metro Line 9. The results showed that the established driving load prediction method could select a small number of key features from many influencing parameters contained in the engineering measured data, and realize the reasonable prediction of driving load. The research results can provide a reference for the regulation of shield tunneling parameters, and also provide an idea for the analysis of engineering measured data with many parameters.