Intelligent evaluation of the interior sound quality of electric vehicles

Abstract

In this paper, an intelligent evaluation of the interior sound quality (SQ) of electric vehicles (EVs) was designed using intelligent algorithms and artificial neural networks. First, interior noises from different EV brands and models were collected using noise tests that are based on relevant standards. Second, a subjective evaluation was applied to appraise the SQ using a laboratory-scale jury test. Meanwhile, the objective evaluation of the SQ was quantitatively analyzed using eight acoustic characteristic parameters: A-weighted sound pressure level, loudness, sharpness, roughness, fluctuation strength, articulation index, tonality, and impulsiveness. Third, simulated annealing (SA) and genetic algorithm (GA) were used to optimize the backpropagation neural network (BPNN). Based on the subjective and objective evaluation results of the SQ of the tested EVs, an intelligent evaluation model of the interior SQ of EVs was designed using SAGA-BPNN. Based on the weight analysis of SAGA-BPNN, the key objective parameters that had a great impact weight were identified. Also, the prediction error of the reconstructed model, which was built by the key objective parameters, was very small and basically consistent with the original model. This evaluation model can not only accurately evaluate the SQ of EV, but it can also serve as a convenient and effective evaluation tool for the acoustic design of EVs.