Prediction of Bridge Monitoring Information Chaotic Using Time Series Theory by Multi-step BP and RBF Neural Networks

Abstract

This paper uses time series and chaos theory of phase space reconstruction. First, it monitors information phase space reconstruction parameters from the deflection of the mid-span in Masangxi Bridge. As a result, the delay value is 4, the embedded dimension for 15, the maximum number of predictable of 10. Then, it constructs the multiple-step recursive BP neural network and RBF neural network model and realizes the analysis and prediction of monitoring information based on space reconstruction parameters. As the results show, the BP neural network and RBF neural network are all effective in monitoring information prediction and RBF shares more advantages than the BP in keeping the structural dynamic performance.

Keywords:

• Bridge Monitoring Information
• Chaotic time Series
• Prediction
• Multi-step BP
• RBF
• Neural Network

Acknowledgements

This work was supported by National Natural Science Foundation (51208538, 51278512) and Natural Science Foundation of Chongqing (CSTC2011BA6026).