解决复杂结构振动和噪声问题的方法包括有限元法、边界元法、模态分析法、波动分析法、阻抗导纳功率流法、结构声强法和统计能量分析法,其中统计能量分析法是目前公认的解决复杂结构高频振动问题最有效的方法。经过几十年的研究和探索,统计能量分析方法不但在理论上有了创新和突破,而且已经成功地应用到了船舶[1, 2, 3]和车辆[4]等领域来解决复杂结构的振动和噪声问题。
采用瞬态统计能量分析法进行动力学系统环境预示需要确定结构2个基本参数:内损耗因子和耦合损耗因子。这2个参数的准确估算对结构响应的预示和数值仿真结果的精度有重要的影响。其中内损耗因子是衡量系统阻尼特性、决定振动能量耗散的重要参数。耦合损耗因子是统计能量分析中表征耦合系统间能量交换的重要参数。目前除了一些简单结构的内损耗因子和耦合损耗因子可以从理论上推得,比较复杂的结构的内损耗因子和耦合损耗因子只能通过实验测量来得到[5, 6, 7, 8]。然而内损耗因子和耦合损耗因子在很多情况下是 10-2~10-4 数量级的小数,在实际工作中要对这些小数进行运算和测量很困难,故这些参数的测量结果经常存在不同程度的误差。
本文基于2个耦合子结构所组成的系统,分析了瞬态统计能量分析法中参数误差对子结构能量的影响,给出了参数误差与所导致能量误差的关系函数,对改进动力学系统数值模型及提高数值模拟的精度有很大的帮助。
1 瞬态统计能量分析中参数误差对子结构能量影响 1.1 内损耗因子误差对子结构能量的影响两子结构耦合的统计能量分析模型如图 1 所示,瞬间能量平衡方程可写为:
$ {{{E}'}_{1}}\left( t \right)+{{\eta }_{1}}\omega {{E}_{1}}\left( t \right)+{{\eta }_{12}}\omega {{E}_{1}}\left( t \right)-{{\eta }_{21}}\omega {{E}_{2}}\left( t \right)={{P}_{1}}\text{,} $ | (1) |
$ {{{E}'}_{2}}\left( t \right)+{{\eta }_{2}}\omega {{E}_{2}}\left( t \right)+{{\eta }_{21}}\omega {{E}_{2}}\left( t \right)-{{\eta }_{12}}\omega {{E}_{1}}\left( t \right)={{P}_{2}}\text{。} $ | (2) |
设子结构1的内损耗因子存在$ \Delta {\eta _1} $的误差$ (\Delta {\eta _1} > 0) $,则能量平衡方程为:
$ {{{E}'}_{1}}\left( t \right)\!+\!\left( {{\eta }_{1}}\!+\!\Delta {{\eta }_{1}} \right)\omega {{E}_{1}}\left( t \right)\!+\!{{\eta }_{12}}\omega {{E}_{1}}\left( t \right)-{{\eta }_{21}}\omega {{E}_{2}}\left( t \right)\!=\!0\text{,} $ | (3) |
$ {{{E}'}_{2}}\left( t \right)+{{\eta }_{2}}\omega {{E}_{2}}\left( t \right)+{{\eta }_{21}}\omega {{E}_{2}}\left( t \right)-{{\eta }_{12}}\omega {{E}_{1}}\left( t \right)=0\text{。} $ | (4) |
采用拉普拉斯变换方法可得两子结构随时间变化的能量表达式为:
$ \begin{aligned} & {{E}_{1}}\left( t \right)\!=\!\frac{{{E}_{1}}\left( 0 \right)\left[\left( {{\eta }_{2}}\!+\!{{\eta }_{21}}\!-\!{{\eta }_{1}}\!-\!{{\eta }_{12}}\!-\!\Delta {{\eta }_{1}} \right)\!+\!\sqrt{A} \right]}{2\sqrt{A}}{{e}^{{{S}_{1}}t}} \!+\!\\ & \quad\quad\,\,\frac{{{E}_{1}}\left( 0 \right)\left[\left( {{\eta }_{1}}\!+\!{{\eta }_{12}}\!+\!\Delta {{\eta }_{1}}-{{\eta }_{2}}\!-\!{{\eta }_{21}} \right)\!+\!\sqrt{A} \right]}{2\sqrt{A}}{{e}^{{{S}_{2}}t}} \text{。} \end{aligned} $ | (5) |
$ {{E}_{2}}\left( t \right)=\frac{{{E}_{1}}\left( 0 \right){{\eta }_{12}}}{\sqrt{A}}\left[{{e}^{{{S}_{1}}t}}-{{e}^{{{S}_{2}}t}} \right] $ | (6) |
其中:
$ \begin{aligned} {S_1},{S_2} = \frac{1}{2}\Bigg( - \left( {{\eta _1} + {\eta _{12}} + \Delta {\eta _1} + {\eta _2} + {\eta _{21}}} \right)\omega \pm \\ \omega \sqrt {{{\left[{\left( {{\eta _1} \!+\! {\eta _{12}} \!+\! \Delta {\eta _1}} \right) \!- \!\left( {{\eta _2} \!+\! {\eta _{21}}} \right)} \right]}^2}\! + \!4{\eta _{21}}{\eta _{12}}} \Bigg)\end{aligned} $ | (7) |
$ A={{\left[\left( {{\eta }_{1}}+{{\eta }_{12}}+\Delta {{\eta }_{1}} \right)-\left( {{\eta }_{2}}+{{\eta }_{21}} \right) \right]}^{2}}+4{{\eta }_{21}}{{\eta }_{12}}\text{。} $ | (8) |
①当两子结构满足$ {\eta _1} = {\eta _2} $,$ {\eta _{12}} = {\eta _{21}} $ 时,将式(5)整理后得:
$ {{E}_{1}}\left( t \right)\!=\!\frac{{{E}_{1}}\left( 0 \right)}{2}\left[{{e}^{{{S}_{1}}t}}\!+\!{{e}^{{{S}_{2}}t}} \right]+\!\frac{{{E}_{1}}\left( 0 \right)\Delta {{\eta }_{1}}}{2\sqrt{A}}{{e}^{{{S}_{2}}t}}\!-\!\frac{{{E}_{1}}\left( 0 \right)\Delta {{\eta }_{1}}}{2\sqrt{A}}{{e}^{{{S}_{1}}t}}\text{。} $ | (9) |
因为 ${e^{{S_1}t}} $ 和 $ {e^{{S_2}t}} $ 均比内损耗因子无误差时减小且 ${e^{{S_1}t}} > {e^{{S_2}t}} > 0 $ ,所以E1(t) 要比内损耗因子无误差时减小。
②通过计算 $ \frac{{d{E_2}(t)}}{{dt}} = \displaystyle\frac{{{E_1}(0){\eta _{12}}}}{{\sqrt A }}\left[{{S_1}{e^{{S_1}t}} - {S_2}{e^{{S_2}t}}} \right] = 0 $ 可以求得t=t0。因为 $\frac{{\sqrt A }}{{{\eta _1} + {\eta _{12}} + {\eta _1} + {\eta _2} + {\eta _{21}}}}{\rm{ < }}1$ 所以可以由泰勒级数展开得:
$ \begin{aligned} {t_0} & = \frac{1}{{\sqrt A \omega }}{\ln ^{\frac{{{S_2}}}{{{S_1}}}}} = \frac{1}{{\sqrt A \omega }} \times \\ & \left[{{{\ln }^{1 + \frac{{\sqrt A }}{{{\eta _{_1}} + {\eta _{_{12}}} + \Delta {\eta _{_1}} + {\eta _{_2}} + {\eta _{_{21}}}}}}} - {{\ln }^{1 - \frac{{\sqrt A }}{{{\eta _1} + {\eta _{_{12}}} + \Delta {\eta _{_1}} + {\eta _{_2}} + {\eta _{_{21}}}}}}}} \right]= \\ & \frac{2}{{\omega \left( {{\eta _1} + {\eta _{12}} + \Delta {\eta _1} + {\eta _2} + {\eta _{21}}} \right)}}\text{。} \end{aligned} $ | (10) |
将式(10)代入式(6)中可得子结构2的峰值能量为
$ \begin{aligned} {{\hat E}_2}\left( t \right) & = \frac{{{E_1}\left( 0 \right){\eta _{12}}}}{{\sqrt A }}\left[{{e^{{S_1}{t_0}}} - {e^{{S_2}{t_0}}}} \right]=\\[4pt] & \frac{{{E_1}\left( 0 \right){\eta _{12}}}}{{\sqrt A }}{e^{ - 1}}\left[{{e^{\frac{{\sqrt A }}{2}\omega {t_0}}} - {e^{ - \frac{{\sqrt A }}{2}\omega {t_0}}}} \right]n = \\[4pt] & \frac{{{E_1}\left( 0 \right){\eta _{12}}}}{{\sqrt A }}{e^{ - 1}}\left[{1 + \frac{{\sqrt A }}{2}\omega {t_0} - \left( {1 - \frac{{\sqrt A }}{2}\omega {t_0}} \right)} \right]=\\[4pt] & {E_1}\left( 0 \right){\eta _{12}}{e^{ - 1}}\frac{2}{{{\eta _1} + {\eta _{12}} + \Delta {\eta _1} + {\eta _2} + {\eta _{21}}}}\text{。} \end{aligned} $ | (11) |
最终可见$ {{\hat{E}}_{2}}\left( t \right) $ 比内损耗因子无误差时减小。
当不考虑内损耗因子的误差$ \Delta {{\eta }_{1}} $ 时,由式(1)和式(2)可得子结构2的峰值能量为
$ {{\hat{e}}_{2}}\left( t \right)={{E}_{1}}\left( 0 \right){{\eta }_{12}}{{e}^{-1}}\frac{2}{{{\eta }_{1}}+{{\eta }_{12}}+{{\eta }_{2}}+{{\eta }_{21}}}\text{。} $ | (12) |
那么子结构2峰值能量的相对误差为
$ \begin{aligned} {\sigma _2} & = \frac{{\Delta {{\hat e}_2}\left( t \right)}}{{{{\hat e}_2}\left( t \right)}} = \frac{{{{\hat e}_2}\left( t \right) - {{\hat E}_2}\left( t \right)}}{{{{\hat e}_2}\left( t \right)}}=\\ & \frac{{{\tau _1}}}{{{\tau _1} + \left( {\displaystyle\frac{{{\eta _1} + {\eta _2} + {\eta _{12}} + {\eta _{21}}}}{{{\eta _1}}}} \right)}}\text{。} \end{aligned} $ | (13) |
设子结构1的耦合损耗因子存在$ \Delta {\eta _{12}} $的误差$\Delta {\eta _{12}}{\rm{ > }}0$ ,同分析类似,可得到2个子结构随时间变化的总能量的表达式为:
$ \begin{aligned} {{E}_{1}}\left( t \right) \!=\! \frac{{{E}_{1}}\left( 0 \right)\left[\left( {{\eta }_{2}}\!+\!{{\eta }_{21}}\!-\!{{\eta }_{1}}-{{\eta }_{12}}\!-\!\Delta {{\eta }_{12}} \right)\!+\!\sqrt{B} \right]}{2\sqrt{B}}{{e}^{{{S}_{1}}t}}\! +\!\\ \quad\quad \frac{{{E}_{1}}\left( 0 \right)\left[\left( {{\eta }_{1}}\!+\!{{\eta }_{12}}\!+\!\Delta {{\eta }_{12}}\!-\!{{\eta }_{2}}\!-\!{{\eta }_{21}} \right)+\sqrt{B} \right]}{2\sqrt{B}}{{e}^{{{S}_{2}}t}}\text{。} \end{aligned} $ | (14) |
$ {{E}_{2}}\left( t \right)=\frac{{{E}_{1}}\left( 0 \right)\left( {{\eta }_{12}}+\Delta {{\eta }_{12}} \right)}{\sqrt{B}}\left[{{e}^{{{S}_{1}}t}}-{{e}^{{{S}_{2}}t}} \right]\text{。} $ | (15) |
$ \begin{aligned} & {S_1},{S_2} = \frac{1}{2}\Bigg( { - \left( {{\eta _1} + {\eta _{12}} + \Delta {\eta _{12}} + {\eta _2} + {\eta _{21}}} \right)\omega \pm } \\ & {\omega \sqrt {{{\left[{\left( {{\eta _1} \!\!+\!\! {\eta _{12}} \!\!+\!\! \Delta {\eta _{12}}} \right) \!\!-\!\! \left( {{\eta _2} \!\!+\!\! {\eta _{21}}} \right)} \right]}^2} \!\!+\!\! 4{\eta _{21}}\left( {{\eta _{12}} \!\!+\!\! \Delta {\eta _{12}}} \right)} } \Bigg)\text{。} \end{aligned} $ | (16) |
$ B\!\!=\!\!{{\left[\left( {{\eta }_{1}}\!\!+\!\!{{\eta }_{12}}\!\!+\!\!\Delta {{\eta }_{12}} \right)\!\!-\!\!\left( {{\eta }_{2}}\!+\!{{\eta }_{21}} \right) \right]}^{2}}\!+\!4{{\eta }_{21}}\left( {{\eta }_{12}}\!+\!\Delta {{\eta }_{12}} \right)\text{。} $ | (17) |
①当两子结构满足 $ {\eta _1} = {\eta _2} $ ,${\eta _{12}} = {\eta _{21}}$ 时,将式(14)整理后得
$ \begin{aligned} {E_1}\left( t \right) & = \frac{{{E_1}\left( 0 \right)}}{2}\left[{{ e^{{S_1}t}} + {e^{{S_2}t}}} \right] + \\ & \frac{{{E_1}\left( 0 \right)\Delta {\eta _{12}}}}{{2\sqrt B }}{e^{{S_2}t}} - \frac{{{E_1}\left( 0 \right)\Delta {\eta _{12}}}}{{2\sqrt B }}{e^{{S_1}t}}\text{。} \end{aligned} $ | (18) |
因为 $ {e^{{S_2}t}} $ 比耦合损耗因子无误差时减小而 $ {e^{{S_1}t}} $ 无变化,同时 ${e^{{S_1}t}} > {e^{{S_2}t}} > 0 $ ,所以E1(t)比耦合损耗因子无误差时减小。
②通过计算可得到子结构2的峰值能量为
$ \begin{aligned} {{\hat E}_2}\left( t \right)& \!\!= \!\!\frac{{{E_1}\left( 0 \right)\left( {{\eta _{12}} \!\!+ \!\!\Delta {\eta _{12}}} \right)}}{{\sqrt B }}\bigg[{{ e^{\displaystyle\frac{{ \!\!- \!\!\left( {{\eta _1}\!\! +\!\! {\eta _{12}}\!\! +\!\! \Delta {\eta _{12}}\!\! +\!\! {\eta _2}\!\! +\!\! {\eta _{21}}} \right) \!\!+\!\! \sqrt B }}{2}\omega {t_0}}}-} \\[8pt] & { {{e^{\displaystyle\frac{{ - \left( {{\eta _1} \!\!+\!\! {\eta _{12}} \!\!+\!\! \Delta {\eta _{12}} \!\!+\!\! {\eta _2} + {\eta _{21}}} \right) - \sqrt B }}{2}\omega {t_0}}}} } \bigg] = \\[8pt] & \frac{{{E_1}\left( 0 \right)\left( {{\eta _{12}} + \Delta {\eta _{12}}} \right)}}{{\sqrt B }}{e^{ - 1}}\left[{{e^{\frac{{\sqrt B }}{2}\omega {t_0}}} - {e^{ - \displaystyle\frac{{\sqrt B }}{2}\omega {t_0}}}} \right] = \\[8pt] & {E_1}\left( 0 \right)\left( {{\eta _{12}} \!\!+\!\! \Delta {\eta _{12}}} \right){e^{ \!- \!1}}\frac{2}{{{\eta _1}\!\! + \!\!{\eta _{12}}\!\! + \!\!\Delta {\eta _{12}}\!\! +\!\! {\eta _2} \!\!+ \!\!{\eta _{21}}}}\text{。}\\ \end{aligned} $ | (19) |
可见$ {\widehat E_2}(t)$ 比耦合损耗因子无误差时增加,
当不考虑耦合损耗因子误差$ \Delta {\eta _{12}} $ 时,由式(1)和式(2)得子结构2的峰值能量为
$ {{\hat{e}}_{2}}\left( t \right)={{E}_{1}}\left( 0 \right)\left( {{\eta }_{12}} \right){{e}^{-1}}\frac{2}{{{\eta }_{1}}+{{\eta }_{12}}+{{\eta }_{2}}+{{\eta }_{21}}}\text{。} $ | (20) |
子结构2峰值能量的相对误差为
$ \begin{aligned} {\sigma _2} & = \frac{{\Delta {{\hat e}_2}\left( t \right)}}{{{{\hat e}_2}\left( t \right)}} = \frac{{{{\hat e}_2}\left( t \right) - {{\hat E}_2}\left( t \right)}}{{{{\hat e}_2}\left( t \right)}}=\\ & \frac{{ - {\tau _{12}}\left( {{\eta _1} + {\eta _2} + {\eta _{21}}} \right)}}{{\left( {{\eta _1} + {\eta _2} + {\eta _{12}} + {\eta _{21}} + {\eta _{12}}{\tau _{12}}} \right)}}\text{。} \end{aligned} $ | (21) |
如图 2 所示,系统由两块相互垂直并铆接的钢板组成,并且两块钢板的其他边界条件是自由的。钢板的密度为 7 800 kg/m3。板1的长度为 0.42 m,宽度为 0.36 m,厚度为 0.01 m。板2的长度为 0.42 m,宽度为 0.36 m,厚度为 0.004 5 m。以上2块钢板的内损耗因子和耦合损耗因子已经在文献[9]中由实验测量得到,表 1 展示了两板在中心频率f=2 000 Hz 时测量的内损耗因子和耦合损耗因子。
现对板1施加一个短时的脉冲激励,使板1具有初始能量E1(0)=1 kJ,通过式(13)和式(21)可以得到板2的峰值能量相对误差和板1参数相对误差的关系曲线,如图 3 和图 4 所示。
图 3 表明增大板1的内损耗因子,将引起板2峰值能量的降低,因为增加的阻尼过多消耗了板1的能量,导致传递到板2的能量减少。图 4 表明增加板1对板2的耦合损耗因子,实际增大了板1对板2能量的传递,导致板2峰值能量的增加。因此精确确定内损耗因子和耦合损耗因子尽量减少测量误差对瞬态统计能量分析中结构能量的预示精度有较大的影响。
3 结 语本文从两耦合子结构所组成系统的能量平衡方程入手,借助拉普拉斯变换研究了瞬态统计能量分析法中参数误差对子结构总能量的影响,同时给出了参数误差与所导致能量误差的关系函数。对于外载荷直接激励的子结构,内损耗因子和耦合损耗因子的误差都会导致总能量的减小。对于外载荷间接激励的子结构,内损耗因子的误差会导致峰值能量的减小,而耦合损耗因子的误差会导致峰值能量的增加。不同于稳态统计能量分析理论,在瞬态统计能量分析中,求解 3 个及以上子结构组成的系统的精确能量解是很复杂的,后续需要更多数学方法的发展和引进,本文内容对于2个以上的子结构组成的复杂系统将有一定参考意义。
[1] | HYNNá P, KLINGE P, VUOKSINEN J. Prediction of structure-borne sound transmission in large welded ship structures using statistical energy analysis[J]. Journal of Sound and Vibration, 1995, 180(4):583-607. |
[2] |
邵亮. 统计能量法在船舶舱室噪声预报中的应用[J]. 舰船科学技术, 2012, 34(5):98-100, 107. SHAO Liang. Prediction and research of ship cabin noise with statistical energy analysis[J]. Ship Science and Technology, 2012, 34(5):98-100, 107. |
[3] | YAYLADERE B C, ÇALIŞKAN M. Prediction of noise levels within a submerged vessel by statistical energy analysis[C]//Proceedings of the 20th International Congress on Sound & Vibration. Bangkok, Thailand:ICSV, 2013:66-71. |
[4] | CHEN S M, WANG D F, LEI Y F. Automotive interior noise prediction based on single sound cavity using statistical energy analysis method[J]. Noise & Vibration Worldwide, 2011, 42(11):36-43. |
[5] |
程广利, 朱石坚, 伍先俊. 统计能量分析法及其损耗因子确定方法综述[J]. 船舶工程, 2004, 26(4):10-15. CHENG Guang-li, ZHU Shi-jian, WU Xian-jun. A summary of statistical energy analysis method and its loss factor's determination[J]. Ship Engineering, 2004, 26(4):10-15. |
[6] |
叶敏, 郎作贵, 郝志勇. 复杂系统振动能量平衡方程中SEA参数的测定[J]. 内燃机学报, 1998, 16(4):469-474. YE Min, LANG Zuo-gui, HAO Zhi-yong. The determination of vibration energy balance sea system parameters in a complex structure[J]. Transactions of CSICE, 1998, 16(4):469-474. |
[7] |
盛美萍, 王敏庆, 孙进才, 等. 非保守耦合系统的等效内损耗因子[J]. 声学学报, 1997, 22(6):555-561. SHENG Mei-ping, WANG Min-qing, SUN Jin-cai, et al. Equivalent internal loss factors for non-conservatively coupled systems[J]. Acta Acustica, 1997, 22(6):555-561. |
[8] |
盛美萍, 王敏庆, 孙进才. 非保守耦合系统的耦合损耗因子[J]. 声学学报, 1999, 24(5):550-556. SHENG Mei-ping, WANG Min-qing, SUN Jin-cai. The coupling loss factors of non-conservatively coupled systems[J]. Acta Acustica, 1999, 24(5):550-556. |
[9] | MAO B Y, XIE S L, XU M L, et al. Simulated and experimental studies on identification of impact load with the transient statistical energy analysis method[J]. Mechanical Systems and Signal Processing, 2014, 46(2):307-324. |