The drivability plays an important role for marketability and competitiveness of passenger car in meeting some customer requirements, which directly affects the driving experience and the desire of purchasing. In this paper, a framework of objective drivability evaluation with multi-source information fusion for passenger car is proposed. At first, according to vehicle powertrain system and optimization theory, certain vehicle performances, which are closely related to objective drivability are analyzed, including vehicle longitudinal acceleration, vehicle speed, engine torque, engine speed, gear position, accelerator pedal, brake signal and voltage signal. Then, combined with the evaluation criterion of signal-to-noise ratio (SNR), mean error (ME), root mean squared error (RMSE) and signal smoothness (SS), a de-noising method is developed for the drivability evaluation information. The optimal wavelet base-function and decomposition layer are analyzed, which is suitable for filtering processing of vehicle longitudinal acceleration signal. Last but not least, based on data layer, the multi-source information is implemented. Combined the driving characteristics with expert knowledge of passenger car, the classes of standard drivability test cycles are formulated, and the sliding window method is used to identify the working conditions. Finally, the effectiveness of the selected filtering algorithm and the reliability of the hardware and software platform are verified by the actual vehicle test of starting condition.
This research provide a reliable reference for obtaining valuable special diagnosis of objective drivability for passenger cars, and also can be used to optimize the intelligent level of automatic transmission ramp. As for the development of autonomous vehicle, path planning and tracking control algorithm can be offered with a theoretical basis.