Estimate of the impact of rainwater on unsealed road degradation in the Mbanya catchment area using an artificial intelligence model

This article is part of the sustainable development of road infrastructure. It assesses the effects of rainwater on unpaved roads in the Mbanya catchment area between January 2019 and December 2023, and proposes mitigation measures. July was the month when the risk of deterioration on these unsealed roads was highest. The aim of this article is to predict changes in soil moisture levels resulting from rainfall in the Mbanya catchment using a coupled Genetic Algorithm–Artificial Neural Network (GA-ANN) model, and to analyse the impact of these changes on the susceptibility of unpaved roads to damage caused by rainwater. To this end, an experimental study was carried out to determine the initial water content in the soil. Genetic algorithms and neural networks were used to estimate the water content at different depths in the soil. The square correlation coefficient, with a value of around 0.72, and the root mean square error, with a value of around 0.37 on all the test data are performance indicators that demonstrate the high accuracy of the coupled genetic algorithm-neural network model. Porosity was used as input parameter for the model, and the results showed that porosity has an influence on the evolution of the water content in the soil. The pavement construction project is possible on these unsealed roads in the Mbanya watershed, as the soils have a bearing capacity P ≥ 20 MPa. However, the increase in water content can weaken as well as reduce the bearing capacity of the soil. The results of this work contribute to improving soil treatment prior to the design phase of the road construction project, in order to ensure the long-term sustainability of infrastructure in the Mbanya watershed.