Deterioration forecasting of joint members based on long-term monitoring data

To compensate for shortcomings of visual inspection data- based asset management, monitoring data-based asset management has attracted a lot of attention. However, there are few researches to detect abnormalities and extract the progress of deterioration based on long-term monitoring data. In this study, the authors express the time series data obtained through long-term monitoring by the autoregressive moving average with exogenous variables generalized autoregressive conditional heteroskedasticity (ARMAX-GARCH) model, and develop the efficient method to estimate unknown parameters based on Bayesian method. Then, a method to forecast the timing of detailed inspection utilizing the ARMAX-GARCH model is developed. Lastly, this methodology is applied to the data of long-term monitoring targeted at the joint members of viaduct, to evaluate its effectiveness.