基于脑功能网络的内燃机汽车声品质评价模型  

The Brain Functional Network-Based Model for Evaluating Sound Quality of an Internal Combustion Engine Vehicle

作　　者：谢丽萍 卢炽华[1,2] 刘志恩 朱亚伟[1,2] 徐韬 Xie Liping;Lu Chihua;Liu Zhien;Zhu Yawei;Xu Tao(Hubei Key Laboratory of Advanced Technology of Automotive Components,Wuhan University of Technology,Wuhan 430070,China;Hubei Collaborative Innovation Center for Automotive Components Technology,Wuhan 430070,China)

机构地区：[1]武汉理工大学现代汽车零部件技术湖北省重点实验室,湖北武汉430070 [2]汽车零部件技术湖北省协同创新中心,湖北武汉430070

出　　处：《内燃机学报》2022年第4期378-383,共6页Transactions of Csice

基　　金：国家自然科学基金资助项目(5217051173).

摘　　要：鉴于脑电信号(EGG)能够反映人心理变化的事实,提出了一种基于脑功能网络的内燃机汽车声品质评价方法.通过图论的脑网络分析构建δ、θ、α、β和γ共5个频带下电极通道间的相位锁值、相位延迟和包络系数的邻接矩阵,计算对应的聚类系数和路径特征长度,提取了30个脑电特征信号;利用皮尔逊相关分析法,聚类与汽车运动感声品质强相关的最佳脑电特征信号;最后,构建神经网络模型映射最佳脑电特征信号与运动感主观评价结果间的非线性关系.结果表明:客观脑电生理声学指标能够反映在运动感汽车声音刺激下评价者的主观感受.In view of the fact that electroencephalography(EEG)signals can respond to human psychological changes,a method for evaluating the powerful sound quality of an internal combustion engine vehicle based on brain functional networks was proposed.The adjacency matrix of phase lock values,phase delays and envelope coefficients between electrode channels in the five frequency bandsδ,θ,α,βandγwere constructed by the brain network graph theory method,the corresponding clustering coefficients and path feature lengths were calculated,and 30 EEG feature signals were extracted.Based on Pearson correlation analysis,the correlation between the EEG feature signals and the subjective evaluation results was analyzed to determine the best EEG feature signals.Finally,a neural network model was constructed to map the non-linear relationship between the best EEG feature signals and the subjective evaluation results.The results show that objective EEG physiological acoustic indices can reflect the subjective feelings of the evaluators under the stimulation of powerful car sound.

关 键 词：内燃机汽车 声品质 脑功能网络 生理声学 神经网络 

分 类 号：TK421.6[动力工程及工程热物理—动力机械及工程]