深圳大学学报（理工版）2024，Vol.41Issue(1)：50-57,8.DOI:10.3724/SP.J.1249.2024.01050
基于多层感知机技术的地铁盾构施工参数预测
Prediction of subway shield construction parameters based on multi-layer perceptron
摘要
Abstract
Shield construction technology has been widely used in subway construction,and reasonable prediction of shield tunneling parameters is of great practical significance for improving construction safety and reducing operational difficulties.Taking a section of the Hangzhou Airport Express Underpass Tunnel as the engineering background,this study uses the friction angle,cohesion force,compression modulus and heaviness of the soil layer within the diameter of the tunnel as well as the burial depth of the tunnel cover,the preset blade speed and propulsion speed of the shield machine as inputs,and the slurry volume,slurry pressure,soil discharge volume,total thrust force and blade torque during the shield construction as outputs,and establishes a prediction model of shield tunneling parameters based on a multi-layer perceptron.By comparing the predictive performance of the model under different combinations of hyper-parameters on the dataset,a suitable shield construction parameter prediction model is selected for the project.The model predictive effect is verified using the measured data,and the predicted values are consistent with the overall variation pattern of the measured data,with an average error within 20%.The established multi-layer perceptron model provides reasonable predictions with good accuracy and can be applied to predict shield tunneling parameters under the conditions of composite strata.关键词
岩土工程/多层感知机/盾构掘进参数/复合地层/预测模型/K折验证Key words
geotechnical engineering/multi-layer perceptron/shield tunneling parameters/mixed ground/prediction model/K-fold verification分类
建筑与水利引用本文复制引用
李文乾,吴云桓,吴兢业,陈治怀,谢森林,胡安峰..基于多层感知机技术的地铁盾构施工参数预测[J].深圳大学学报（理工版）,2024,41(1):50-57,8.基金项目
National Natural Science Foundation of China(51978612,52378419) （51978612,52378419）
National College Students'Innovation and Entrepreneurship Training Program(202210335036) （202210335036）
Zhejiang Xinmiao Talents Program(2023R401189) 国家自然科学基金资助项目(51978612,52378419) （2023R401189）
国家级大学生创新创业训练计划资助项目(202210335036) （202210335036）
浙江省大学生科技创新活动计划资助项目(2023R40 1189) （2023R40 1189）
