Estimation of Vehicle Mass and Road Grade of an Electric Vehicle Based on a Hybrid Algorithm

Abstract

In view of the parameter estimation of gross vehicle mass and road gradient during the driving of battery electric vehicles and according to the characteristics of the vehicle mass with steady variability and road gradient with time variability, considering the basis of vehicle longitudinal dynamics, a method for estimating gross vehicle mass and road gradient based on a hybrid algorithm is proposed, and the method is applied to the starting stage of battery electric vehicles. The gross vehicle mass was estimated by using the forgetting factor recursive least square algorithm with good estimation efficacy for steady variables, and the output result of mass was regarded as one of the input parameters for gradient estimation. The road gradient was estimated with adaptive Kalman filtering, and the noise influence that cannot be concluded with external noise statistical characteristics was reduced by introducing a noise estimator with forgetting factor to improve the estimation accuracy of the road gradient. The electric vehicle starting test was carried out on the selected flat ground and microslope sections, and air resistance was ignored based on the characteristics of low speed during vehicle starting. The required data was collected with the Global Navigation Satellite System (GNSS) terminal and vehicle controller area network (CAN) bus to perform offline calculations. The results show that (1) a road section with variable gradient is selected to verify the validity of the algorithm, and the gross vehicle mass estimation result of the section is converged within 10 kg of the true value eventually with small error in the gradient estimation result; (2) the estimations of gross vehicle mass and the changing of road gradient can be accurately estimated by using the hybrid algorithm; and (3) the mass estimation result of the twice-starting test has the same convergence trend and the mass error is less than 2%, and although the gradient estimation error is increased slightly, it is still less than 0.6%, which indicates the high estimation accuracy at the starting stage of a battery electric vehicle by using the hybrid algorithm.