Response Prediction Model for Structures of Quayside Container Crane Based on Monitoring Data

Abstract

Structural health monitoring and response prediction are of great significance to ensure safety of structures and avoid structural failures. In order to capture changes of structural state in time and improve accuracy of the response prediction, a new predictive model based on dual-tree complex wavelet transform (DTCWT) combined with autoregressive moving average (ARMA) and support vector regression (SVR) is proposed. Firstly, structural monitoring signals are preprocessed and then decomposed by DTCWT. According to variable regulation of the signal characteristics at different scales, ARMA modeling and SVR training are carried out to realize prediction of each scale respectively. Finally the prediction results of each scale are fused as a final prediction model. The predictive model is applied to two case studies of quayside container cranes on the influence of different numbers of samples in prediction results. The experimental prediction results indicate that the proposed model has better prediction performance of the short-term response compared to other prediction methods, especially when the model is trained by a small number of sample data.