Open AccessSound Quality Prediction of Fuel Cell Vehicles Based on Regression Analysis
Author(s) -
Zeyu Liu,
Hai Liu,
Yanyi Zhang,
Dong Hao,
Xiuxiu Sun,
Xue Ji
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1605/1/012062
Subject(s) - weighting , noise (video) , linear regression , computer science , artificial neural network , sound quality , pairwise comparison , regression analysis , anechoic chamber , quality (philosophy) , statistics , artificial intelligence , machine learning , mathematics , speech recognition , acoustics , telecommunications , physics , image (mathematics) , philosophy , epistemology
The typical fuel cell vehicle(FCV)was taken as the research object. The noise test of the FCV was carried out in the environment of the semi-anechoic room of the vehicle. The noise samples in the vehicle were collected under the condition of constant speed, which was intercepted and processed with equal time. Noise was subjectively evaluated using a pairwise comparison method. The objective evaluation parameters suitable for evaluating the sound quality of FCV were selected. The objective evaluation parameters were calculated and solved. According to the subjective and objective evaluation results of sound quality, run multiple linear regression analysis to quantitatively solve the correlation between the subjective and objective evaluation results. MAPE results showed that the accuracy of the multiple linear regression prediction model was low, in order to further improve the accuracy of the prediction model of the sound quality of FCVs, The GA-BP neural network prediction model was used to establish the evaluation prediction model between the subjective and objective evaluation results. The weight of the influence of objective parameters on the subjective evaluation results was calculated. The MAPE calculation results showed that the neural network model was more suitable for the establishment of FCV sound quality prediction According to the model, A-weighting and roughness were the most influential to the subjective evaluation results. The two algorithms had the same results.