Abstract: Aiming at the problem that the feature points of drivability evaluation indicators in vehicle start process are inaccurately identified due to the strong specificity of engine speed waveform and external interference, a set of methods from signal preprocessing to feature point identification are proposed in this paper. The feature points are determined according to drivability evaluation indicators and the time domain features of rotational speed curve. Morphological filtering combined with empirical modal decomposition is adopted to conduct filtering processing on rotational speed signals, while the combination of D-S evidence theory and syntactic pattern is applied to feature point identification. The results of experiment show that the method proposed can effectively identify the feature points of evaluation indicators on engine speed curve in starting condition, providing an objective basis for obtaining drivability evaluation indicators later on

Key words: drivability evaluation indicators, feature point identification, morphological filtering, empirical mode decomposition, syntactic pattern, D-S evidence theory