Enhancing Digital Twin Model for Connected Vehicles by Powertrain and Longitudinal Dynamics

Digital twin for connected vehicles is a key enabling technology to accelerate the full digitization of Intelligent Transportation System (ITS). In a well-designed wireless environment such as 6G network, an accurate digital twin model for connected vehicles will significantly strengthen the efficiency, reliability and security of ITS. For a vehicle system, there are many potential practical factors that affect the accuracy of its digital twin model, including air resistance, tire rolling resistance, carrying loads, transmission shafts, controller and engine performance, etc. To create a realistic digital representation of the vehicle system, this paper elaborately integrates the powertrain and longitudinal dynamics into the digital twin model for car-following scenarios. The Measure-of-Performances (MoPs) and Goodness-of-Fit functions (GoFs) are exploited for model parameter fitting and fitting error measurement, respectively, using real trajectory dataset. The analytical and experimental results indicate that the accuracy of digital twin model increases 13.28%, 14.51% and 12.80% than that of the conventional Full Velocity Difference Model (FVDM) under the GoFs of Mean Absolute Percentage Error (MAPE), Root Mean Square Percentage Error (RMSPE_tilde) and Theil’s Inequality Coefficient U-function, respectively.