Modeling of Pavement Maintenance Decisions Using Artificial Intelligence Based on Maintenance Unit.

Recently, all efforts have been directed toward keeping the network functional at a high level by determining the appropriate maintenance or rehabilitation (M & R) treatment. Determining the appropriate M & R strategies for flexible pavements is a complex process and is considered a key component of the Pavement Maintenance Management System (PMMS). Since such a decision system is complex, automated implementation using a pre-trained model via an artificial neural network (ANN) approach is a critical tool for decision-makers. Many studies have been conducted on modeling pavement condition index using ANN to determine the maintenance decision. The Egyptian Code of Practice has recently relied on the maintenance unit (MU) concept for maintenance decision prediction. A few researchers have investigated maintenance decision (MD) predications using the MU modeling by ANN but have not adequately studied Egyptian Code consideration. Therefore, this paper addresses the application of the latest machine learning technique for forecasting the current pavement maintenance decisions based on the MU system according to the Egyptian code considerations to develop a one-step enhanced decision-making tool. A pattern-recognition algorithm (neural network) was applied to 54.3 km of surveyed roads in Minia governorate, Egypt. The results indicated that the ANN model is capable of predicting the MD with a high level of reliability, with a mean square error (MSE) value of 0.02993, 0.03046, and 0.03018, and a percentage error (% E) value of 13.29693, 14.11734, and 13.83215 for the training, validation, and testing datasets, respectively.