|
|
|
Interior noise sound quality prediction based on least square support vector machinesample entropy |
| Institute of Noise and Vibration, Jiangsu University, Zhenjiang, Jiangsu 212013, China |
|
|
|
Abstract Using a hybrid electronic vehicle as research object, the sound quality of the interior noise was investigated under the unsteady condition. The subjective and objective evaluations for the collected noise signal annoy index were conducted, and the objective parameters of psychoacoustics highly correlated with the subjective evaluation results were found through correlation analysis. The complete ensemble empirical mode decomposition was performed on the original noise samples to obtain 16 intrinsic mode functions for each signal. The correlation analysis between each IMF and original signal was performed to obtain 7-12 IMFs with high correlation to the original signals, and the sample entropy values of these components were calculated as objective characteristic parameters of sound quality. Based on the least squares support vector machine algorithm, the sound quality evaluation prediction model with psychoacoustic parameters and sample entropy feature parameters as model input was established respectively. The results show that the model with sample entropy feature parameters has high prediction accuracy and is suitable for sound quality prediction in hybrid vehicles under unsteady conditions.
|
|
Received: 18 December 2018
|
|
|
|
|
|
|
2020, Vol. 41 
