【作者】 王俊元；

【导师】 熊诗波；

【作者基本信息】 太原理工大学 ， 机械电子工程， 2008， 博士

【摘要】 机械结构模态参数的提取是机械结构动力学分析的基础。常用的模态参数提取方法有实验模态分析法和工作模态分析法,这两种方法在工程应用中都有各自的局限性。实验模态分析(EMA)法需要对结构施加激励,这对于大型复杂机械系统是非常困难的。工作模态分析(OMA)法无需外加激励,仅依靠机械结构工作状态下的振动响应信号提取结构模态参数。这种分析方法更接近实际工作状态,然而由于其理论分析计算中假设系统所受载荷为白噪声,与大多数实际情形有或多或少的距离,导致计算结果误差较大。近年来,盲信号处理中的独立分量分析(ICA)法在盲源分离的工程应用中获得广泛应用。本文在回顾了OMA分析理论与方法的基础上,详细研究了ICA的基本算法原理,发现此二者的密切联系。在解释了ICA基向量的模态含义的基础上,提出了将ICA算法应用于机械结构动力学分析中的模态参数识别。讨论了ICA计算模态与机械振动模态的一致性,为ICA在机械结构动力学中的应用提供了理论依据。通过计算机数字仿真和实验室物理模拟,详细研究了ICA技术在机械结构动力学分析中的理论基础、算法实现以及对不同激励和噪声背景的适应性。为衡量ICA分量的分离效果,提出了独立性指数的概念。借助于三自由度弹簧质量系统模型的响应数据,进行ICA数字仿真的研究结果表明,ICA分离的模态频率与结构进行动力学分析得到的固有频率是吻合的。在对集中质量系统研究的基础上,以悬臂梁为例研究了连续系统模型。通过实施在不同激励、不同比例的混合噪声及不同长度数据量的输出响应的ICA分解,系统研究了ICA的模态参数提取能力。在实验室用锤击法获取了悬臂梁的模态参数后,又用B&K的OMA软件与自主研发的ICA软件分别对实验装置的振动响应信号进行了模态参数识别。研究结果表明ICA中的JADE、MSNR、GOSA三种算法均可提取试验模型的结构模态参数,它与OMA提取的模态参数具有良好的一致性。其中MSNR算法优于GOSA和JADE算法。与OMA法相比较,ICA算法具有较好的结构模态分离性能,特别是对短响应数据具有更好的适应性。在理论和实验研究的基础上,将所提出的基于ICA技术的模态参数识别方法成功地应用于立体仓库堆垛机动态设计中。利用ICA方法获得了堆垛机的模态参数。通过对比OMA方法获取的模态参数,ICA方法剔除了虚假模态,提高了模态参数的可信度,为在此基础上完成的堆垛机的动态设计奠定了基础。更多还原

【Abstract】 Modal parameters extraction of the mechanical structure is the basis for the mechanical structure dynamic analysis. The experimental modal method and the operational modal method are commonly used to extracting modal parameter. The structure need external excitation by using the experimental modal analysis, which is extremely difficult in operation of large and complex mechanical systems. Operational modal analysis (OMA) is without additional excitation. This is only relied on the state response signal to extract structure modal parameters of the mechanical structure and more close to actual situation. However, for assuming system suffered load of the white noise in OMA, which does not match with actual situation, the method leads to greater error results.Recently, independent component analysis (ICA) of the Blind signal processing was widely used in the project application in blind source separation (BSS).This article dissected the ICA basic algorithm principle after reviewing the OMA theory and method, discovering the close relation between them. On the basis of the explanation of the ICA base-vector mode meaning, this article proposed the mode parameter identification with the ICA algorithm in the mechanical structure dynamic analysis, discussed the consistency between the ICA computation mode and the mechanical vibration mode, which provided theoretical basis by using ICA in the mechanical structure dynamic analysis.Through computer numeral simulation and laboratory physical analogue, this article dissected analysis rationale, algorithm realization of ICA technology in mechanical dynamics, and the compatibility to different excitation and noise background. In order to weigh the separation effect, it put forward the concept of independent performance index. With the aid of the model response data of three degree of freedom spring quality system, the ICA numeral simulation results indicated that, the ICA separation mode frequency and the structure natural frequency coincidenced completely.This article had attempted to research continuous system model of the cantilever beam under various excitations, various niose levers and various data amounts of response. In the laboratory with the hammering , the reliable mode frequency and mode shape of cantilever beam was obtained. Using the OMA software of B & K and the self- developed ICA software, we had respectively carried on the mode parameter identification to acceleration response signal of the experimental facility. The results indicated that the ICA in three algorithms JADE, MSNR, GOSA can be extracted to structure mode, which had a good consistency with the modal parameters from the OMA. MSNR algorithm is superior to GOSA and JADE algorithm. Comparing with the OMA, the ICA algorithm is better in separation and adaptability, especially for short data.Based on the research of the theory and experimentation, we applied the method of identifying parameter on basis of the ICA to dynamic design of the stacking crane in automated warehouse. We also get the modal parameters of stacking crane along with the OMA software, and enhanced the reliability of the modal parameters by eliminating the false parameter. It’ s the basis of dynamic design of the stacking crane.更多还原