【作者】 杨扬；

【导师】 刘一星； 刘镇波；

【作者基本信息】 东北林业大学 ， 木材科学与技术， 2017， 博士

【摘要】 泡桐木材作为民族乐器的首要取材,其振动性能可在很大程度上影响着民族乐器的声学品质。为探索民族乐器的选材由主观评价向客观评价过渡,本文选择有代表性的民族乐器(琵琶、月琴、阮)为研究对象,从民族乐器共鸣板素材、共鸣面板、共鸣构件和木质共鸣体振动特征的测量入手,在两位导师及其研究团队前期研究工作的基础上,进一步采用多元统计方法、经济学预测方法和智能控制算法,系统研究民族乐器生产过程中各阶段的共鸣板素材(主要影响指标:比动弹性模量E/ρ、弹性模量与剪切模量之比E/G、声辐射阻尼系数R、声阻抗ω)、共鸣面板(主要影响指标:顺纹、斜纹135°、横纹、斜纹45°方向表面波传播速度)、共鸣构件(主要影响指标.:顺纹、斜纹135°、横纹、斜纹45°方向表面波传播速度)、木质共鸣体(主要影响指标:顺纹、斜纹135°、横纹、斜纹45°方向表面波传播速度)的声振动性能参数与乐器声学品质等级之间的关系,以及上述四阶段的被测体声振动性能参数之间的关联性。在此大量分析的基础上,将制作的三种乐器各9把,按照专家主观评价的得分,由高到低分为三个等级,其中1～3级定义为高档乐器,4～6级定义为中档乐器,7～9级定义为低档乐器,选取影响民族乐器共鸣板素材振动特性中最明显的特征比动弹性模量五/ρ、弹性模量与剪切模量之比E/G、声辐射阻尼系数R、声阻抗ω作为输入量,选择共鸣板素材相对应专家评定乐器的高、中、低三个等级作为输出量,力图构建从民族乐器共鸣板素材的检测即可在一定程度上直接预测乐器声学品质的预测模型,实现直接在生产源头对民族乐器产品品质进行科学预测。深入研究民族乐器声学评价系统参数的预测方法,建立民族乐器声学品质预测模型并探索创建相应的评价方法。(1)按民族乐器生产流程的顺序,将被测体按共鸣板素材-共鸣面板、共鸣面板-共鸣构件、共鸣构件-木质共鸣体、共鸣板素材-木质共鸣体进行相邻两两配对编组,采用皮尔森简单相关分析法,探讨了民族乐器生产过程各阶段被测体每对参数振动特性两两指标之间的相关性。结果表明,民族乐器生产过程各阶段参数间的相关系数在数值上存在较大的差异,随着乐器加工过程的深入,前后两个指标间的相关系数的显著性水平逐渐降低,相关性也逐渐减弱。(2)将民族乐器按共鸣板素材-共鸣面板、共鸣面板-共鸣构件、共鸣构件-木质共鸣体、共鸣板素材-木质共鸣体进行编组,把每个生产阶段相邻被测体的振动特性指标作为一对,运用典型相关分析法,剖析了民族乐器生产过程各阶段每组之间每对参数的相关性。结果证实,民族乐器共鸣板素材与木质共鸣体的振动特性高度相关,共鸣板素材-木质共鸣体之间的典型相关系数的最大值整体处于4个阶段两两之间典型相关系数最大值的中间水平,符合乐器生产加工中的实际情况。(3)以民族乐器共鸣板素材-木质共鸣体振动特性为研究对象,选用多元统计方法,构造了共鸣板素材振动特性预测民族乐器声学品质的多元选择模型。分析得出,预测值在不同的临界值范围内,所得到的民族乐器声学品质多元选择模型训练效果相距甚大,说明存在较大的误差,说明多元选择模型不适用于共鸣板素材振动特性对于民族乐器声学品质方面的预测。(4)以民族乐器共鸣板素材-木质共鸣体振动特性为研究对象,应用BP神经网络法,建立了共鸣板素材振动特性预测民族乐器声学品质分类等级模型。研究发现,共鸣板素材振动特性预测民族乐器分类等级验证误差非常小,验证过程分类的结果可以充分说明该模型对于民族乐器声学品质分类具有较好的预测性。(5)以民族乐器共鸣板素材-木质共鸣体振动特性为研究对象,使用支持向量机法,构建了共鸣板素材振动特性预测民族乐器声学品质分类等级模型。研究表明,共鸣板素材振动特性预测民族乐器分类等级验证的误差非常小,验证过程分类的结果可以充分说明该模型对于民族乐器声学品质分类的预测性较好。综合分析得出,应用BP神经网络和支持向量机的方法,预测了民族乐器声学品质的等级,其等级预测的结果均可达到较高的预测精度,两种方法对于民族乐器声学品质的预测均具有较好的适用性,其中支持向量机法作为结构风险最小的学习方法,在以共鸣板素材振动特性预测民族乐器声学品质的精度方面要优于BP神经网络的预测效果,且支持向量机法在解决小样本预测问题时表现出更多的优势。更多还原

【Abstract】 Paulownia wood is main material used to make Chinese national musical instrument.Acoustical quality of instrument greatly depends on wood vibration property.To explore the transition of subjective evaluation to objective evaluation,the experiment choose the typical Chinese national musical instrument(lute,yueqin,ruan)as study objects.The vibration property of sound board material(The main influence index:specific elastic modulus,elastic modulus and shear modulus ratio,acoustic radiation damping coefficient,acoustic impedance),resonance board(The main influence index:specific elastic modulus,elastic modulus and shear modulus ratio,acoustic radiation damping coefficient,acoustic impedance),resonance component(The main influence index:specific elastic modulus,elastic modulus and shear modulus ratio,acoustic radiation damping coefficient,acoustic impedance)and wood resonant(The main influence index:specific elastic modulus,elastic modulus and shear modulus ratio,acoustic radiation damping coefficient,acoustic impedance)was measured.Based on the previous work of my two tutors and their research team,multivariate statistics method,economics prediction method and intelligent control method were utilized to study the relation of the acoustic vibration performance parameters of sound board material,resonance board,resonance component,wood resonant and instrument acoustical quality,as well as the above four stages the measured sound vibration performance parameters of the correlation between the large number of analysis on the basis of the production of the three instruments each nine,according to the expert subjective evaluation of the score,from high to low divided into three grades,in which,grade 1～3 were the high-grade musical instruments,grade 4～6 were the mid-range instruments,grade 7～9 were the low gear musical instruments.Pick of the most significant factors influencing the acoustic quality of musical instruments,the dynamic modulus of elasticity E/ρ,elasticity modulus and shear modulus ratio E/G,acoustic damping coefficient R,and acoustic impedance as input,and select the sound board material corresponding to the experts to assess the instrument of high,medium and low three grades as output.The instrument acoustical quality prediction model was trying to built based on the detection of national instrument sound board material,which can predict national instrument quality from production source.Prediction method of national instrument acoustical evaluation system parameters were investigated to build the prediction model of national instrument acoustical quality and establish an evaluation method.(1)According to the order of the production process of the national musical instruments,the tested bodies were divided into two groups according to the material of the sound board material-resonance board,resonance board-resonance component,resonance component-wood resonant,sound board material-wood resonant.Pearson simple correlation analysis was used to investigate the relation between the two parameters of parameter and parameter of each stage of the instrument production process.The results show that the correlative coefficients of national musical instrument productive process parameters are different from each other.With proceeding of the productive process,the difference between the correlative coefficients of the two indexes declines,and the correlation also recedes.(2)The vibration characteristics of sound board material-resonance board,resonance board-resonance component,resonance component-wood resonant,sound board material-wood resonant were grouped,the vibration characteristics of the adjacent test object in each production stage as a pair.Typical correlation analysis method was used to investigate the correlation between each pair of parameters in each stage of national musical instrument production process.The results show that national instrument wood resonant vibration property highly depends on sound board material.The maximum value of typical correlation coefficients of sound board material and wood resonant is the middle value of the four productive steps typical correlation coefficients,which matches the actual situation of the instrument productive process.(3)Consider sound board material-resonance board,wood resonant vibration property as study objects.Typical correlation analysis method was used to build the multiple choices model,which was applied to predict national musical instrument acoustical quality through sound board material vibration property.The results show that the training results of national musical instrument acoustical quality multiple choices model have great difference in different critical value range.The great error indicates that multiple choices model is not suitable for predicting national musical instrument acoustical quality through sound board material vibration property.(4)Consider sound board material,wood resonant vibration property as study objects.BP neural network was used to build the classification model of national musical instrument acoustical quality through sound board vibration property.The results show that the validation error of national instrument classification grade predicted by sound board material vibration property is small,and the classification model is suitable for predicting national instrument acoustical quality classification.(5)Consider the vibration property of national instrument sounding board and wood resonant as study objects,using SVM method,the national musical instrument acoustical quality classification prediction model was built.The results show that the error of classification grade predicted by the vibration property of sounding board material is small.The classification results indicate that the classification model can predict the national instrument acoustical quality.The analysis shows that the accuracy of national instrument acoustical quality predicted by BP neutral network and SVM is high and has an ideal applicability.SVM is a learning method structure risk minimization,the accuracy of national instrument acoustical quality predicted by SVM is higher than BP neutral network.And SVM method has advantages to solve small sample prediction.更多还原