Using artificial neural networks to predict the distresses of flexible pavement

Abstract EN

Road networks have a very important role in the economic development of any country.

therefore, pavement management systems (PMS) are considered significant in managing the condition of highway networks efficiently to determine the required maintenance in a timely manner at the lowest cost.

this research aims to use artificial neural networks (ANNS) in the possibility of predicting the pavement condition index of the flexible pavement using the LTPP database.

this research included the use of many inputs and data of flexible asphalt pavement containing thickness of the layers (A.C-Base-Subbase-Subgrade), temperature, humidity, precipitation, freezing, age, traffic volumes, and distress.

the total number of sections that were used in the artificial neural network model was 862 sections for nine states of the united states of America.

all sections were divided into two categories; the first, overlay was not used to treat the distress.

these sections take symbols (SPS-1&3&8) according to LTPP.

the second sections were subjected to overlay layer as a result of the maintenance that was performed on it and was of the type (SPS-5) and in this type two cases were monitored; in the first case; the distresses before and after the maintenance operation using an overlay layer, the second case; the distresses were monitored after the maintenance process using an overlay layer was increased due to distresses.

the Matlab program was used to build the ANN model based on the collected data to predict the pavement condition index (PCI).

the results indicated that the model showed an excellent prediction for the pavement condition index.

the correlation coefficient (r^2) was more than 80% approximately for all sections.

this result indicated that the ANN models can be used efficiently to predict the pavement condition and contribute to building an effective plan for future road maintenance work.