SSA与BP神经网络在地铁沉降监测中的应用

Application of SSA and BP Neural Network in Settlement Monitoring of Subway

邻近区域施工,致使地铁沉降呈现复杂的非线性变化。对此,采用奇异谱分析(SSA)和BP神经网络对地铁结构进行分析与预测。通过SSA重建趋势序列和周期序列,分析地铁结构变化的趋势与周期波动;利用BP神经网络对重建趋势序列与时间序列分别进行预测。以上海9号线地铁沉降监测数据为例,提取趋势序列与周期序列进行分析及预测,实验证明了利用SSA对地铁监测序列进行分析以及利用BP神经网络对成分序列进行预测的可行性。

Due to the construction in adjacent area, the subway settlement shows complicated nonlinear changes. For this, the paper analyzed and predicted the subway structure by means of the singular spec trum analysis （SSA and BP neural network. The trend and periodic fluctuation of subway structure change are analyzed by SSA reconstruction trend sequence and periodic sequence, and the reconstructed trend sequence and time series are predicted by BP neural network. Taking the subway settlement monito ring data of Shanghai I.ine 9 as an example, the trend sequence and periodic sequence are extracted for a nalysis and prediction. The experiment proves the feasibility of using SSA to analyze the subway monito ring sequence and BP neural network to predict the component sequence.