摘要
为解决工况传递路径分析(OTPA)方法面临的核心问题,提升OTPA方法应用范围和工程实用性,提出一种融合盲源分离的传递路径分析技术。首先,结合舱段模型振动试验,深度剖析OTPA方法面临的振源交叉耦合和遗漏振源无法识别等问题;然后,将盲源分离(BSS)技术和OTPA方法进行集成、融合:引入经验模态分解(EMD)-特征值分解的源数估计方法定量估计振源个数,指导振源输入测点的选取;视多个振源信号为卷积混叠,引入非正交联合块对角化方法进行耦合振动源的分离;在解决分离振源排序不确定性的前提下,将分离出的振源作为OTPA方法的输入振源,建立应用限制条件少的BSS-OTPA模型;最后,开展实船海上振动声辐射试验,利用BSS-OTPA模型成功实现了船舶机械振动噪声源的量化、识别。研究成果可直接支持船舶机械设备振动源辐射声场预报和振动噪声控制。
To solve the core problem of the operational transfer path analysis(OTPA)method,and improve its application range and engineering practicability,a TPA technique combined with blind source separation(BSS)was proposed.Firstly,problems of vibration sources’cross coupling and missing vibration source being not able to be identified in OTPA method were analyzed deeply.Then,BSS and OTPA method were integrated and fused.The EMD-SVD method was introduced to quantitatively estimate the number of vibration sources,and guide selection of input measuring points of vibration sources.Signals of multiple vibration sources were regarded as convolutional aliasing,a novel non-orthogonal joint block diagonalization(JBD)method was applied to separate cross-coupling vibration sources.On the premise of solving the uncertainty of sorting of separated vibration sources,the BSS-OTPA model with few restrictions was established by taking separated vibration sources as input vibration sources of OTPA method.Finally,ship vibration acoustic radiation tests were conducted and the BSS-OTPA model was used to successfully identify and quantify ship mechanical vibration noise sources.It was shown that the study results can directly support real-time prediction of radiated acoustic field of ship mechanical equipment vibration sources,and vibration and noise control.
作者
张磊
李彬
杨自春
陈国兵
ZHANG Lei;LI Bin;YANG Zichun;CHEN Guobing(Power Engineering College,Naval University of Engineering,Wuhan 430033,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2020年第17期150-156,共7页
Journal of Vibration and Shock
基金
国家自然科学基金(51609251)。
关键词
传递路径分析(TPA)
振动源
盲源分离(BSS)
耦合
识别
transfer path analysis(TPA)
vibration source
blind source separation(BSS)
cross-coupling
identification