In order to improve the stability and safety of intelligent connected vehicles on the road, an adaptive optimization EKF estimation algorithm is proposed based on the traditional EKF algorithm. Firstly, based on the analysis of the driving state that is difficult to estimate, the parameters that need to be observed are determined as longitudinal speed, yaw Angle speed and side deflection Angle of the center of mass. Then, based on the problem that the noise is difficult to deal with in the traditional EKF algorithm, the adaptive optimization algorithm is used to estimate and optimize the system noise and the measurement noise synchronously, so that the estimation process can better fit the actual operating conditions. Finally, based on MATLAB and Carsim co-simulation platform, an intelligent connected vehicle model is established to estimate the driving conditions of high adhesion and low adhesion road under high-speed conditions. The experimental results show that compared with the comparison algorithm, the algorithm has better effect in response speed, estimation accuracy and curve fitting, and is more secure for the stability and safety of the vehicle.