基于盾构掘进参数的LVQ神经网络地层识别

以苏州4号线2标及2号线东延伸线5标地铁工程为背景,分析了盾构机的掘进参数：千斤顶推力、推进速度、刀盘扭矩、螺旋机转速和同步注浆量在不同地层条件下的变化规律。提出了基于盾构机掘进参数的学习向量量化（Learning Vector Quantization,LVQ）神经网络地层识别方法。建立了以盾构机五个掘进参数作为输入,地层特性编码为输出的数学模型,通过每种地层100组训练样本对模型进行训练,通过57步训练,训练样本误差控制在0.1以内,并用每种地层50组检验样本进行检验,地层总体识别率达到82.7%。

Identification of Strata with LVQ Neural Network Based on Shield Tunneling Parameters

Considering the subway projects of NO.2 line and NO. 4 line in Suzhou, the changing rule of tunneling parameters under different stratum conditions is analyzed. The tunneling parameters include cylinder thrust force, advancing velocity, cutterhead torque, rotating speed of screw conveyor and synchronous grouting quantity. A stratum recognition method based on tunneling parameters of TBM and LVQ neural network is proposed, and a mathematical model with the input of five tunneling parameters and output of stratum coding is built. Each stratum has 100 training samples, and the model error of training samples is limited below 0.1 through 57 step training. Fifty samples are selected for each stratum to test this model, and the overall recognition rate reaches 82.7%.