基于泥水平衡盾构掘进参数的PSO-BP神经网络掘进地层识别模型研究

Identification of Tunneling Strata with PSO-BP Neural Network Based on SPB Tunneling Parameter

为解决泥水平衡盾构机在掘进时无法准确地实时识别掘进地层的问题,以珠三角水资源配置工程为例,研究泥水平衡盾构机的盾构推力、掘进速度、刀盘转速、刀盘扭矩在不同地层下的变化规律,提出基于掘进参数的PSO-BP神经网络掘进地层识别方法,建立盾构推力、掘进速度、刀盘转速、刀盘扭矩4种掘进参数为输入集,地层编码为输出集的地层识别模型。工程数据的验证结果表明,该模型在珠三角水资源配置工程数据集上的掘进地层的识别准确率达99.07%,PSO-BP神经网络算法的识别准确率明显高于BP、RF、RBF、CNN等机械学习算法。

To address the issue of inaccurate real-time identification of tunneling strata by slurry pressure balance shield,this study focused on the Pearl River Delta water resource allocation project.The variations of tunnelling parameters such as shield thrust,cutterhead torque,tunneling speed,and cutterhead rotation speed in different strata were analyzed.The method of strata identification based on PSO-BP neural network tunneling parameters was proposed.The strata identification model was established with four tunneling parameters(shield thrust,cutterhead torque,tunneling speed,and cutterhead rotation speed)as input features and strata code as the output set.The model was validated using engineering data.The results demonstrate that the model achieves an identification accuracy of 99.07% on the tunneling layers of the Pearl River Delta water resource allocation project dataset.The identification accuracy of the PSO-BP neural network algorithm significantly outperforms other machine learning algorithms such as BP,RF,RBF,and CNN.