Bridge displacement prediction method based on ARIMA model

In order to solve the problem of low displacement prediction accuracy of key components in bridge engineering, a bridge displacement prediction algorithm based on improved ARIMA model is proposed. Firstly, the non-stationary displacement data is decomposed into multiple layers by wavelet transform to obtain the detail coefficient and the approximation coefficient. Then the differential autoregressive moving average model (ARIMA) is used to predict and reconstruct the multi-layer detail coefficient and the highest layer approximation coefficient. Then, the residual part of the reconstructed sequence is predicted by the optimized support vector machine model (SVM). Finally, the results of each model are added to obtain the predicted sequence of the original displacement data. The simulation results show that the prediction accuracy of the bridge is better than that of the traditional prediction model through the combination of ARIMA and SVM.