Deterioration Prediction of Infrastructures with Time Series Data Considering Long Memory Effect

In order to compensate for shortcomings of asset management based on visual inspection data, asset management based monitoring data has got a lot of attention. However, there is little methodology to apply time series data to conduct a decision making on asset management. In addition, long-term monitoring data of infrastructure have long memory effect because the deterioration of gradually progress owing to accumulating various deterioration factors such as traffic load, weathering, anti-freezing agent and etc. In this study, the authors propose ARFIMAX- GARCH (Autoregressive Fractional Integrated Moving average with eXogenous variables- Generalized Autoregressive Conditional Heteroskedaticity) model and attempt to demonstrate the applicability of the proposed model by studying concrete application cases.