基于多源数据融合的盾构掘进地层参数动态识别及沉降控制研究

城市复杂环境土压平衡盾构掘进控制不当易造成地表沉陷、塌方和邻近建筑损坏。优化盾构掘进力学参数以实现地表沉降有效控制，是城市隧道安全施工的关键之一。项目拟联合钻孔、土工试验、地层变形实测等数据，从地层分布与岩土参数变异性量化、多源数据融合与反演、掘进参数可靠度优化等方面，开展融合多源数据的盾构掘进地表沉降预测与掘进参数优化研究。首先，基于机器学习分类和稀疏贝叶斯回归技术，利用钻孔和土工试验数据，建立地层分布和岩土参数多元相关三维条件随机场，实现地层不确定性和参数空间变异性表征；结合该随机场模型，建立土压平衡盾构掘进随机有限差分数值模型，利用非线性卡尔曼滤波融合地层变形实测数据，反演获得地层分布和岩土参数最优估计，进而实现沉降准确预测；最后基于可靠度理论，建立容许沉降控制的关键掘进力学参数优化方法，实现沉降精细化控制。研究成果将为掘进参数优化与沉降控制提供科学依据，丰富盾构掘进智能控制理论。

The earth-pressure-balance shield tunnelling with unreasonable operational parameters in the complex urban environment is likely to cause surface subsidence and damage to nearby buildings. Optimizing the mechanical parameters of a shield machine to effectively control ground settlements is important for safe construction of urban tunnel engineering. The research project tries to fuse borehole, geotechnical tests, and monitoring data, through subsurface stratigraphy and geotechnical parameter uncertainty quantification, multi-source data fusion and inversion and tunneling mechanical parameter reliability optimization, in order to establish an efficient method to predict tunneling-induced ground deformations and optimize the machine mechanical parameters. First, based on the machine learning classification method and sparse Bayesian regression technology, using borehole and geotechnical test data, the subsurface stratigraphy model associated with multivariate-correlated three-dimensional conditional random fields of geotechnical parameters are established to characterize cross-correlation and spatial variability of soil parameters. A finite-difference numerical model is then used to simulate earth pressure balance shield constructions, with incorporating the established subsoil model into the numerical model. Two unscented Kalman filter models are coupled to fuse in site monitoring deformation of ground deformation, in an attempt to give the optimal estimation of subsoil model, thereby reducing subsurface stratigraphy and geotechnical parameter uncertainty and achieving accurate settlement predictions. Finally, the method of reliability based design optimization is used to determine the key tunneling mechanical parameters, under the constraint of allowable settlement standard. This research aims to enrich the safety controlling theory of shield tunnel constructions and provide a scientific basis for the optimization of tunneling parameters and the control of ground surface settlements.