Acoustic quality evaluation of electric vehicles based on PSO-RF

Aiming at the shortcomings of machine learning model research in predicting the noise annoyance perception of electric vehicles, this paper proposes an integrated learning evaluation method using particle swarm optimization random forest model to predict the noise annoyance perception of electric vehicles. The subjective and objective data of the sound quality of electric vehicles were obtained by collecting sound samples in the experiment. The random forest model was optimized by particle swarm optimization algorithm, and the interior noise annoyance perception model was established, which was compared with the random forest model and BP neural network model. The results show that PSO random forest model has higher prediction accuracy, better fitting effect with subjective evaluation value, stronger generalization ability, and can replace subjective evaluation test.