摘要：AbstractA car-following model describes the longitudinal control strategy of a driver in reaction to the movements of the front cars in the same lane. Because of inter-driver differences, drivers may demonstrate distinct maneuvers in the same excitation of the surrounding traffics. Therefore, the parameters of a car-following model need to be determined per each driver individually. Calibrating a car-following model is commonly treated as a constrained optimization problem. The model parameters, viewed as the optimized variables, are found by minimizing a predefined cost function with a nonlinear numeric solver. However, nonlinear optimization can hardly guarantee global optimality, and more importantly, different formulations of the cost function frequently yield different parameter identification results. To bypass the issues mentioned above, we propose a purely algebraic approach to identify the parameters of a car-following model. Simulation results demonstrate its effectiveness.
