Shield Construction Multiobjective Optimization of Surface Settlement Safety Control Based on Machine Learning. Issue 1 (1st August 2022)

• Record Type: Journal Article
• Authors: Feng, Zongbao; Chen, Hongyu; Zeng, Tiemei; Wu, Xianguo; Qin, Yawei; Xu, Wensheng
• Is Part Of: Journal of physics. Volume 2333:Issue 1(2022)
• Publication Date: 2022-08-01
• Subjects: Shield construction parameters -- Surface settlement -- Cutter wear -- Machine learning -- Multiobjective optimization.; Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/2333/1/012018
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: A multi-objective optimization model combining Support Vector Machines (SVM) and Nondominated Sorting Genetic Algorithm-II (NSGA-II) algorithm was established to control the construction parameters of shield and guide the safe construction of shield. This model takes the main shield parameters as the research object and the ground settlement as the control target, and optimizes the control analysis of the construction parameters. In this paper, eight shield parameters controlling surface settlement were selected as input indexes of SVM prediction model, and the nonlinear relationship between shield construction parameters and surface settlement is obtained as NSGA-II fitness function. Then, cutter wear was selected as the second optimization objective, and the constraint range of construction parameters was set for multi-objective optimization. Taking a rail transit project in karst areas as an example, the research results show that the prediction model obtains by training and simulating the measured data of the project using the SVM algorithm has high model accuracy. Based on the SVM-NSGA-II model, the multiobjective optimization effect of surface settlement and cutter wear is significant. Taking the obtained set of optimal Pareto fronts as a reference, the recommended values for the setting range control of shield construction parameters in karst areas are proposed.