摩托车发动机的声品质评价研究

发布时间：2018-04-05 00:20

本文选题：声品质　切入点：烦恼度　出处：《重庆大学》2016年硕士论文

【摘要】：随着经济的发展,科技的进步,人们对摩托车的要求不仅仅在于具有良好的操纵性,更希望具有良好的乘坐舒适性。目前,摩托车的NVH性能成了人们选购摩托车的重要指标。发动机声品质的好坏直接影响着摩托车的NVH性能。而声品质的评价受个人因素影响较大,且需要大量的人力、物力、财力。本文根据声品质主客观参数进行分析建立主观烦恼度模型,以致能在不用大量耗费物力人力的情况下预测评价产品声品质的好坏。本文首先搜集国内外声品质研究的主要工作成果,以及描述了声品质研究的理论基础:心理声学参数及其计算模型与声品质评价的主要方法。然后,根据本文实验内容及任务要求,利用labview平台设计声信号测试分析模块,其中软件功能主要包括:声样本的采集、声样本的数据保存与读取、声样本的频谱分析、声样本的互动滤波分析及声样本的声品质分析模块。其次经过采集样本数据的回放、试听,最终选取30个发动机噪声声样本进行声品质主客观评价分析。采用分组成对比较法进行样本的主观评价。为了保证主观评价实验结果的可靠性,实验中对数据进行了误差检验、计权一致性系数确定,剔除结果偏差较大的数据,最终所有评价人员的计权一致性系数均在80.7407%以上。另外分别计算出发动机声样本的心理声学参数(A声级、响度、尖锐度、粗糙度、抖动度以及烦恼度),与声样本主观实验的结果进行单相关分析、偏相关分析,确定参量间的内在联系。确定A计权声压级、尖锐度与主观烦恼度具有很大的相关性,说明尖锐度和A声级明显影响人耳对发动机噪声烦恼度的主观评价。再次根据发动机噪声主观烦恼度得分值来指导发动机的声学改进研究,选取不同分值的几个典型的声样本,进行频谱分析、心理声学参数分析以及互动滤波方法确定影响发动机声品质评价的频段。确定证明发动机的声品质主观评价对发动机的NVH性能有指导意义。最后分别基于多元线性回归和BP神经网络建立发动机声品质主观烦恼度评价模型,用于预测评价产品的声品质。研究表明:基于BP神经网络建立的主观烦恼度评价模型预测的相对误差的最大值明显低于回归模型的预测值的最大相对误差;BP神经网络模型的性能指数也明显优于回归模型的性能指数。证明运用BP神经网络建立的声品质主观烦恼度模型明显优于基于多元线性回归建立的声品质主观烦恼度模型。
[Abstract]:With the development of economy and the progress of science and technology, the requirement of motorcycle is not only good maneuverability, but also good ride comfort.At present, the NVH performance of motorcycles has become an important index for people to choose and purchase motorcycles.The sound quality of engine directly affects the NVH performance of motorcycle.The evaluation of sound quality is influenced by individual factors, and requires a lot of manpower, material and financial resources.Based on the subjective and objective parameters of sound quality, a subjective annoyance model is established in this paper, so that the sound quality of products can be predicted and evaluated without expending a great deal of material resources and manpower.In this paper, we first collect the main research results of sound quality at home and abroad, and describe the theoretical basis of sound quality research: psychoacoustic parameters, their calculation model and the main methods of sound quality evaluation.Then, according to the experiment content and task requirement of this paper, the acoustic signal test and analysis module is designed by using labview platform. The software function mainly includes: the collection of sound sample, the data saving and reading of sound sample, the spectrum analysis of sound sample.Interactive filter analysis of sound samples and sound quality analysis module of sound samples.Secondly, 30 samples of engine noise sound were selected for subjective and objective evaluation of sound quality by playback and audition of collected sample data.The subjective evaluation of the sample was carried out by the method of comparison.In order to ensure the reliability of the experimental results of subjective evaluation, the error test of the data was carried out in the experiment, and the weight consistency coefficient was determined. The weight consistency coefficient of all the evaluators was more than 80.7407%.In addition, the psychoacoustic parameters of engine sound samples are calculated respectively, such as sound level, loudness, sharpness, roughness, jitter degree and annoyance degree, which are analyzed by single correlation analysis and partial correlation analysis with the results of subjective experiment of sound samples.Determine the intrinsic relationship between the parameters.To determine A weighted sound pressure level, the acuity has a great correlation with subjective annoyance, which indicates that sharpness and A sound level obviously affect the subjective evaluation of engine noise annoyance by human ear.Thirdly, according to the engine noise subjective annoyance score value to guide the engine acoustic improvement research, select several typical sound samples with different scores, carry on the spectrum analysis,The psychoacoustic parameter analysis and the interactive filtering method are used to determine the frequency range that affects the evaluation of engine sound quality.It is confirmed that the subjective evaluation of engine sound quality is of guiding significance to the NVH performance of engine.Finally, based on multiple linear regression and BP neural network, the subjective annoyance evaluation model of engine acoustic quality is established to predict and evaluate the sound quality of the product.The results show that the maximum value of relative error predicted by subjective annoyance evaluation model based on BP neural network is obviously lower than that of regression model. The performance index of BP neural network model is also better than that of BP neural network model.Based on the performance index of regression model.It is proved that the subjective worry degree model of sound quality based on BP neural network is obviously superior to the subjective worry degree model of sound quality based on multivariate linear regression.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U483

【相似文献】