Page 75 - 《应用声学》2022年第3期
P. 75

第 41 卷 第 3 期                                                                       Vol. 41, No. 3
             2022 年 5 月                          Journal of Applied Acoustics                      May, 2022

             ⋄ 研究报告 ⋄



                采用响应面回归的汽车多属性声品质预测方法                                                                    ∗




                                                 吕向飞      1,2†  陈 进     3



                                         (1 重庆大学    机械传动国家重点实验室        重庆   400044)
                                         (2 重庆城市管理职业学院智能工程学院           重庆   401331)
                                       (3 重庆电子工程职业学院智能制造与汽车学院            重庆   401331)
                摘要:单一主观评价分数无法准确描述人耳对声品质的多属性偏好特征。该文在多属性声品质试验数据的基
                础上,以愉悦度、平顺度和驾驶乐趣的主观评价分数为因变量,通过相关分析筛选出响度、尖锐度和 A 计权声
                压级 3 个主要自变量,引入响应面回归方法,分别建立因变量与自变量之间的预测模型,通过与多元线性模型
                和 BP 神经网络模型对比验证了精度。最后,建立多属性主观评价分数之间的量化映射模型。该研究可为多属
                性汽车声品质的优化控制提供参考。
                关键词:车辆工程;声品质;响应面回归;相关分析;多属性
                中图法分类号: U461.4          文献标识码: A          文章编号: 1000-310X(2022)03-0397-08
                DOI: 10.11684/j.issn.1000-310X.2022.03.009


                Automobile multi-attribute sound quality prediction using response surface

                                                  regression method


                                               LYU Xiangfei 1,2  CHEN Jin 3

                      (1 State Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing 400044, China)
                       (2 School of Intelligent Engineering, Chongqing City Management College, Chongqing 401331, China)
              (3 Intelligent Manufacturing and Automobile School, Chongqing College of Electronic Engineering, Chongqing 401331, China)

                 Abstract: A single subjective evaluation score cannot accurately describe the multi-attribute preference char-
                 acteristics of human ears for vehicle sound quality. Based on the multi-attribute sound quality test data,
                 this paper takes the subjective evaluation scores of pleasure, ride comfort and driving pleasure as dependent
                 variables, and three main independent variables of loudness, sharpness and A-weighted sound pressure level
                 are screened out through correlation analysis. Response surface regression method is employed to establish
                 the predictive model between the dependent variable and the independent variable respectively, with which
                 the multi-attribute sound quality is evaluated accurately by compared with backpropagation neural network
                 (BPNN) model and multiple linear regression model. Finally, a quantitative mapping model between multi-
                 attribute subjective evaluation scores is established. The research can provide a reference for the optimization
                 and control of multi-attribute sound quality for vehicles.
                 Keywords: Vehicle engineering; Sound quality; Response surface regression; Correlation analysis; Multi-
                 attribute


             2021-05-12 收稿; 2021-09-07 定稿
             重庆市教委科学技术研究项目 (KJ1503006)
             ∗
             作者简介: 吕向飞 (1984– ), 男, 河南南阳人, 博士, 讲师, 研究方向: 智能制造、车辆故障诊断技术。
             † 通信作者 E-mail: xiangfei113072@163.com
   70   71   72   73   74   75   76   77   78   79   80