| Lately, there is an increasing demand for resilient infrastructure assets. To support the documentation of resilience, Structural Health Monitoring (SHM) data is a necessity, as well as traffic loads. Those diagnosis and function data can be the basis for the prognosis of future prediction for the performance of the assets. In this research, the authors present an approach based on nineteen (19) Machine Learning (ML) techniques for the prediction of the future strain values in a Dutch Highway Bridge depending on previous measurements of the strain. For the evaluation of the algorithms the indices Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and R square (R2) were chosen and Fast Fourier Transform (FFT) and Averaged Standardized Values were chosen for the time series pre-processing. The results are extremely promising, with almost every algorithm to predict the fluctuations of strain values and the indexes are quite satisfactory. The results ensure that those who are responsible for the maintenance of the bridge or for its repairs, could use these models to determine which time that should take action. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.