Research On Forecasting Truck Capacity Based On Truck Trajectory Data

As one of the main modes of cargo transportation in China,road transportation bears more than 70% of the freight volume.However,the current Chinese road truck market is still dominated by individual truck drivers,high truck empty load rate,and low freight efficiency.The main reason is that these capacity cannot be dispatched in a unified manner,and there is no direct information interaction between the capacity and the demand side,difficulty in finding goods for vehicles,and difficulty in finding vehicles for goods.With the rapid development of mobile internet technology in recent years,a road freight exchange platform has emerged,accumulating a large amount of truck operation data,which provides a new opportunity for the freight system.By mining these data to extract useful information,and developing truck-to-cargo matching business,the above-mentioned problems can be gradually alleviated.For the truck-to-cargo matching platform,as a car-free carrier,it can effectively dispatch the idle capacity that is mastered and made by most individual drivers in real time,which is the key to realizing the platform’s truck-to-cargo matching function.In response to the above problems,this paper starts from the truck capacity side,based on the truck trajectory data,excavates the temporal and spatial distribution characteristics of the truck side capacity,predicts the travel position and loading status of the truck,and helps the truck-to-cargo matching platform to better manage and control these individual capacity,so as to achieve the best match between trucks and goods,and gradually realize the improvement of the efficiency of the highway logistics system.This paper is based on the data of individual truck travel trajectories.First,the time and space distribution characteristics of roadway transportation capacity are excavated,and it is found that the time distribution of truck travel intensity presents a flat peak.The spreading range of trucks has a certain time regularity,and the spatial distribution range is obviously limited,the movement mode presents three forms,namely,two-point,hub-and-spoke,and ring.Based on the regularity of truck travel,a travel prediction model is constructed to predict the travel location of individual transportation capacity.The truck location prediction model is constructed based on three machine learning classification algorithms.The results show that the decision tree classification algorithm has the highest prediction accuracy.For truck drivers with a radius of less than 180 km,the accuracy of truck travel location prediction can reach 94% under the urban scale and 82% under the district scale.Based on the perspective of the truck travel activity chain,the truck travel activity and loading state are identified,and the travel characteristics of the truck under different loading conditions are described.Finally,it excavates and analyzes the selection and transition probability of the loading state of the truck,and realizes the prediction of the next loading state of the truck.The results of this study help reveal the temporal and spatial distribution of roadway transportation capacity,and the predictability of truck travel location and loading status.It is helpful for the government to formulate relevant policies to rationally optimize the distribution of truck capacity,and is conducive to the real-time deployment of trucks in the road freight system to actively optimize the efficiency of truck-to-cargo matching,reduce truck empty load rate,and reduce road freight costs.