An improved extreme gradient boosting approach to vehicle speed prediction for construction simulation of earthwork

Construction simulation is an effective tool to provide schedule plans. Vehicle speed is one of the most significant factors in earthwork construction simulation. However, neglecting the strong correlation with contextual factors, random distribution methods will lead to inaccurate prediction of vehicle speed. To address such issues, an improved extreme gradient boosting (XGBoost) approach to vehicle speed prediction is proposed for earthwork construction simulation. Firstly, to improve the global searching ability, an improved grey wolf optimization algorithm (IGWO) is put forward. Secondly, XGBoost is optimized by IGWO to construct an IGWO-XGBoost model. Then, the prediction model is embedded in the earthwork construction simulation model. The case study proves that the simulation results of the proposed method are more consistent with an actual construction schedule. It is expected that the vehicle speed prediction embedded into a simulation program facilitated an accurate development of schedule plan, thereby improving the efficiency of construction management.