Multiple Random Slope and Fixed Intercept Linear Regression Models for Pavement Condition Forecasting

Pavement condition forecasting plays an important role in pavement management. Accurate predictions can help pavement managers in making optimal management plans which keeps the pavements in serviceable conditions over a specific period, as well as saves costs that are spent in the pavement maintenance and rehabilitation. In general, there are two types of prediction methods for pavement condition, which are probabilistic approach and deterministic approach. Probabilistic approach has its advantage in large scale pavement networks overall condition predictions, it focuses on the whole pavement network condition forecasting and provides prediction results in distributions manner. Deterministic approach has its advantage in small pavement section scale predictions; it focuses on specific pavement sections condition forecasting and is able to provide the result of specific pavement sections condition.;This paper attempts to develop a deterministic forecasting approach which not only utilizes the advantage of deterministic approach that is able to provide specific pavement section condition prediction, but also attempts to adopt the advantage of probabilistic approach, which considers the effects that might have universal influence on a pavement networks overall condition. After model simulations, the approach finally obtained on making pavement condition forecasting in this study is a random slope and fixed intercept linear regression approach. Each pavement section is assigned with a specific slope and specific intercept based on its categorical variable values and numerical variable values. The effectiveness of the obtained models is checked by comparing their predictions with the predictions through the existing prediction program ODOTPMIS.