基坑变形混沌特征识别与非线性预测模型研究

Research on Chaotic Feature Recognition and Nonlinear Prediction Model of Foundation Pit Deformation

混沌理论特征识别是进行混沌时间序列分析和预测的前提。普通的线性数学算法已经无解决基坑变形所遇到的问题,为了研究基坑变形监测数据的非线性复杂问题,采用混沌非线性理论方法,首先求取基坑变形时间序列的延迟时间和嵌入维数,其次对基坑监测数据进行相空间重构,最后对比分析加权一阶局域预测模型以及RBF神经网络混沌预测模型的预测结果,实验表明RBF神经网络混沌预测模型预测精度最高,同时也说明了混沌预测模型更适合短期预测。最终证明了RBF神经网络混沌预测模型应用在基坑变形监测中的可行性与有效性。

Chaotic theory feature recognition is the premise of chaotic time series analysis and prediction.Common linear mathematical algo-rithms have failed to solve the problems encountered in the foundation pit deformation.In order to study the nonlinear and complex problems of foundation pit deformation monitoring data,weused the chaotic nonlinear theory method.Firstly,we obtained the delay time and embedding di-mension of time series of foundation pit deformation,and then carried out the phase space reconstruction of foundation pit monitoring data.Final-ly,we compared and analyzed the prediction results of the weighted first-order local prediction model and the chaotic prediction model of RBF neural network.The experimental result shows that the chaotic prediction model of RBF neural network has the highest prediction accuracy,and the chaotic prediction model is more suitable for short-term prediction,which can prove the feasibility and effectiveness of RBF neural network chaotic prediction model in monitoring foundation pit deformation.