Application of artificial intelligence to overload control

Abstract

High quality road infrastructure is essential to support economic growth for any landlocked region, confirmed by the fact that 79% of South African goods use road transport. Protection of the road infrastructure is implemented by means of overload control monitoring at Traffic Control Centres (TCCs) on freight corridors linking ports with economic hubs. As these systems lack the available information to support intelligent decision-making, 75% to 85% of statically weighed vehicles are legally loaded, resulting in unnecessary wastage of time and fuel. This paper proposes an intelligent weigh-in-motion (IWIM) algorithm aiming to decrease unnecessary static weighing of vehicles through data sharing between TCCs combined with intelligent data interpretation. Several Artificial Intelligence (AI) models were evaluated for their ability to decrease static weighing of vehicles while not increasing the number of overloaded vehicles allowed to proceed on the corridor. We found that a Random Forest Tree model produced the best performance to differentiate between overloaded and legal vehicles, achieving an average improvement of 65.83% in terms of vehicles to be statically weighed when compared to the current rule-based system. Implementation of the IWIM concept can therefore have a significant positive impact for all stakeholders involved in the freight movement process.