Abstract
KEYWORDS:
vehicle parameters, velocity estimation, Kalman filter, fuzzy estimator.
ABSTRACT:
Various approaches for the estimation and observation of variables which are not measurable are described. This paper presents the design and development of two methods to obtain the vehicle velocity in the inertial co-ordinate system: a Kalman filter approach and a Fuzzy estimator. Generally, the vehicle velocity is obtained via a fusion of the data from all rotational wheel velocity and longitudinal acceleration sensors. Integrating the acceleration, a fifth estimate for the vehicle velocity is provided. The estimation must be very accurate, as a basis for the wheel slip calculation. Some systems select only the maximum rotational wheel speed as the estimate for the vehicle velocity. In the situation when all four wheels are simultaneously locked, this approach is very inaccurate. The authors propose two alternative estimation methods for the vehicle velocity: the Kalman filter and the fuzzy estimator. The Kalman filter is designed for stochastic or time varying processes and has advantages especially in cases where stochastic properties of the noise processes are known. It has been developed a structure of the sensor data pre-processing. The fuzzy system designed is divided into three parts and the both fuzzification/defuzzification processes are described. The fuzzy estimator calculates the slip values from the estimated vehicle velocity and the corrected wheel speeds. These values are more suited for the classification of driving situations. For both situations, computer simulation results are depicted.