221 / 2024-09-14 15:21:04

基于AG-PSO的泥水平衡盾构地表沉降掘进主控参数优化与应用

泥水平衡盾构,地表沉降预测,掘进参数优化,粒子群优化,自动化机器学习

摘要：在泥水平衡盾构施工过程中需要对地表沉降进行监测，进而反馈调整盾构机掘进参数。基于自动化机器学习(AutoML)框架AutoGluon快速构建盾构机刀盘前方地表平均地表沉降预测模型，将所构建的AG模型作为目标函数，利用粒子群优化算法(PSO)对泥水平衡盾构机的气泡舱压力优化调整。未进行超参数优化的AG模型在地表前方平均沉降预测任务能够达到较好的性能，相较超参数优化后的RF、SVM、XGBoost等模型在前方地表沉降预测任务上甚至表现出更高的精度和稳定性。PSO优化调整后的气泡舱压力应用于典型断面的沉降预测结果表明基于AG-PSO的泥水平衡盾构掘进参数优化对前方平均地表沉降控制具有一定效果。构建泥水平衡盾构气泡舱压力实时优化系统应用于实际工程，能够有效减少地表沉降警报，实现泥水平衡盾构掘进参数的优化控制。

Abstract  During slurry shield tunneling, it is essential to monitor ground surface settlement and adjust the tunneling parameters of the shield machine accordingly. By leveraging the automated machine learning (AutoML) framework AutoGluon, a predictive model for the average ground surface settlement in front of the shield machine's cutterhead is rapidly constructed. The AG model is then utilized as an objective function, and the particle swarm optimization (PSO) algorithm is applied to optimize the pressure in the slurry shield machine's bubble chamber. The AG model, without hyperparameter tuning, achieves good performance in predicting the average settlement in front of the shield, and it even outperforms hyperparameter-tuned models like RF, SVM, and XGBoost in terms of accuracy and stability for this task. The optimized bubble chamber pressure, adjusted by PSO, when applied to typical sections, shows that the AG-PSO-based optimization of slurry shield tunneling parameters has a positive effect on controlling the average ground surface settlement. Implementing a real-time optimization system for the slurry shield's bubble chamber pressure in actual engineering projects can effectively reduce ground surface settlement alarms and achieve optimal control of the slurry shield tunneling parameters.