Intelligent Stochastic Agent-Based Model for Predicting Truck Production in Construction Sites by Considering Learning Effect

Abstract

Predicting truck production in construction projects is one of the basic tasks within project planning and control. This paper presents an original and novel intelligent stochastic agent-based model to maximize truck production at construction sites by considering the impact of learning. The proposed model was developed to overcome limitations of existing models, including a lack of the inclusion of a training mechanism and a reward/penalty framework for truck performance. Ideas of reinforcement learning theory were used. A reward/penalty function was designed based on minimum travel time. Traffic and fuel volume were treated as stochastic variables. A worked example and a real case study are presented to show the applicability and efficiency of the proposed model. The paper shows that the results of the proposed model accurately predict truck production. The paper also shows that the proposed model demonstrates a shorter truck travel time and, thus, higher production compared to the Monte Carlo simulation logic. The method proposed here offers an original contribution to the analysis of truck production and will be of use to practitioners engaged in project planning and control, especially in large earth-moving operations.