基于重标极差法和神经网络的隧道变形趋势判断

为准确、科学及全面地对隧道变形进行趋势判断和预测分析,将隧道的变形过程划分为中期阶段和长期阶段,利用R/S分析对其位移序列和速率序列进行趋势判断研究,再利用PSO-BP神经网络对各阶段的变形进行预测,将预测结果与R/S分析结果进行对比,验证两者的一致性。利用两个工程实例进行检验,得出各序列的Hurst指数均大于0.5,说明各序列均具有持续变形的长期性,且位移序列的趋势性均大于速率序列的趋势性;同时,变形预测结果也显示隧道后期变形将持续增加,验证了R/S分析的准确性。

Estimation of tunnel deformation based on rescaled range analysis and PSO-BP neural network

In order to judge the tunnel deformation trend accurately, scientifically and comprehensively, the monitoring period of the tunnel is divided into medium-term stage and long-term stage, then rescaled range analysis (R/S) is used to judge the displacement sequence and the rate sequence. PSO-BP neural network is applied to predict the deformation in each stage, and the forecast results are compared with the R/S analysis results to verify their consistency. The test results of two engineering examples show that Hurst index of each sequence is greater than 0.5, indicating each sequence has a long and sustained deformation, and the tendency of displacement sequence is greater than that of the rate sequence. The prediction result illustrates that the tunnel deformation would continue, which verifies the accuracy of R/S analysis.