电动车动力总成噪声品质粒子群-向量机预测模型

发布时间：2018-01-25 04:34

本文关键词： 电动车动力总成噪声品质粒子群优化支持向量机敏感频带能量比　出处：《西安交通大学学报》2016年01期 　论文类型：期刊论文

【摘要】：为了实现电动车动力总成噪声品质的预测,以某集中驱动式电动车为例,在考虑动力总成辐射噪声品质频域特性和已设立的敏感频带能量比这一客观评价参数的基础上进行了心理声学参数,即响度、尖锐度、粗糙度、抖动度、语音清晰度等与主观评价的相关性分析,由此建立了电动车动力总成噪声品质粒子群支持向量机预测模型,内容涉及采用支持向量机建立噪声品质预测模型、利用粒子群优化算法对向量基惩罚因子及核函数参数进行优化,最后验证了敏感频带能量比评价参数的有效性。研究结果表明:敏感频带能量比与主观评价相关度达到0.946,可以较好地反映主观感受;基于粒子群支持向量机的噪声品质预测模型的平均相对误差和最大相对误差分别为2.0%和6.7%,表明以敏感频带能量比作为输入特征的粒子群优化支持向量机模型,在电动车动力总成噪声品质的预测精度上优于基于遗传算法优化及网格搜索优化的预测模型。
[Abstract]:In order to predict the noise quality of electric vehicle powertrain, a centralized drive electric vehicle is taken as an example. On the basis of considering the frequency domain characteristic of radiated noise quality of power assembly and the objective evaluation parameter of sensitive band energy ratio, the psychoacoustic parameters, namely loudness, sharpness, roughness and jitter, are carried out. Based on the correlation analysis between speech articulation and subjective evaluation, a prediction model of noise quality particle swarm optimization support vector machine (PSO) for electric vehicle powertrain is established, which involves the establishment of noise quality prediction model using support vector machine (SVM). The particle swarm optimization algorithm is used to optimize the vector basis penalty factor and kernel function parameters. Finally, the validity of the evaluation parameters of the sensitive band energy ratio is verified. The results show that the correlation degree between the sensitive band energy ratio and the subjective evaluation is 0.946, which can better reflect the subjective feeling. The average relative error and maximum relative error of the noise quality prediction model based on particle swarm optimization support vector machine are 2.0% and 6.7% respectively. It is shown that the particle swarm optimization support vector machine model is based on the sensitive band energy ratio as the input feature. The prediction accuracy of the noise quality of electric vehicle powertrain is better than the prediction model based on genetic algorithm optimization and grid search optimization.
【作者单位】：同济大学新能源汽车工程中心;同济大学汽车学院;同济大学中德学院;
【基金】：国家“863计划”资助项目(20U11AA11A265) 国家自然科学基金资助项目(51205290) 中央高校基本科研业务费专项资金资助项目(1700219118)
【分类号】：U469.72
【正文快照】： 大量的声学研究发现,A计权声压级不能完全反映人对噪声的主观感受。在这种情况下,噪声品质这个现代噪声研究的全新概念应运而生,它指出人对噪声的感觉是受心理和生理因素的共同影响。噪声品质的准确预测是对产品声学优化设计的重要前提。噪声品质预测研究包括车内噪声[1-2]、

【共引文献】