关键词: 混沌时间序列, 混沌识别, 加权一阶局域预测, RBF神经网络混沌预测

Abstract:

Aiming at the problem of pier nonlinear sinking, chaos theory is introduced. The reconstructed by the improved C-C and the G-P algorithm of time series, compared with traditional algorithms,the anti-interference and computational efficiency are improved. The maximum Lvyapunov exponent is obtained to determine whether there is chaos in time series. Finally,a weighted first-order local prediction model and a RBF neural network chaotic prediction model are established according to the obtained parameters to respectively predict and analyze the observed data. The chaotic time prediction results are compared with those of exponential smoothing method. The prediction precision of chaotic time is higher than that of exponential smooth method,and the predicted precision of chaotic model of RBF neural network is the highest, which proves that the predicted precision of chaotic time series is reliable, and can monitor the deformation of the bridge body in real time to avoid disasters.

Key words: chaotic time series, chaotic identification, weighted first order local, RBF neural network model