损耗因子测试方法研究

发布时间：2018-05-17 23:23

本文选题：模态损耗因子 + 频带损耗因子　；参考：《西北工业大学》2016年博士论文

【摘要】：损耗因子对结构设计和动力学特性预测具有重要影响,一般通过试验方法获取。本研究重点开展线性系统模态损耗因子、频带损耗因子、刚度非线性系统损耗因子的试验识别方法研究,旨在提高损耗因子试验测试精度。本文首先基于高斯-牛顿迭代提出了一种识别模态损耗因子的拟合法,该方法尽可能多地利用了固有频率附近的信息计算模态阻尼,有效消除了试验测量误差对损耗因子识别结果的影响。根据拟合函数类型,拟合法可分为直接拟合法、平方拟合法和倒数拟合法。本文对比分析了频率间隔、固有频率附近频点数以及试验重复次数等因素对采用线性插值改进的半功率带宽法和三种拟合法识别精度的影响,结果表明:三种拟合法的抗噪性均优于改进半功率带宽法,直接拟合法和倒数拟合法需要的试验重复次数少,精度高,是线性系统模态损耗因子识别的首选方法,可以有效改进信噪比在10dB~20dB的情况下采用改进的半功率带宽识别损耗因子误差较大的问题。在分析初始衰减率法识别频带损耗因子误差的基础上,论文提出了过程损耗因子的概念,用以描述衰减过程的频带损耗因子。分析表明:固有频率较小的模态具有较大的模态损耗因子,是初始衰减率法测试频带损耗因子存在较大系统误差的主要原因。为此,本文发展了循环相减法应用于频带损耗因子识别,提出了时域分离法测试频带损耗因子,该方法具有计算过程简便、应用范围广并且测试精度高的特点。仿真和试验结果均表明,时域分离法的误差可控制在10%以内,可用于准确识别频带损耗因子;可以有效解决当频带中含有多阶模态时,衰减振动曲线出现“分段”现象而导致的衰减法识别频带损耗因子难度较大的问题。最后,论文在刚度非线性系统阻尼特性分析的基础上,基于粒子群优化提出了识别系统损耗因子的积分法。研究表明:该方法具有很强的抗噪性,可以利用含噪信号获取损耗因子,在信噪比低至10dB的情况下,依然可将平均误差控制在10%以内,并且信噪比越高,误差越小。本文提出的识别含多阶模态频带损耗因子的时域分离法,弥补了衰减法和输入功率法的不足,克服了初始衰减率法存在系统误差的问题,方便工程应用。本文提出的积分法测试刚度非线性系统损耗因子,利用积分的方法计算过程量,减小了试验测试误差对结果的影响,并且利用粒子群优化求解损耗因子,结果稳定,分析表明该方法具有抗噪性强,精度高等特点。
[Abstract]:Loss factor has an important influence on structural design and prediction of dynamic characteristics, which is generally obtained by experimental method. This study focuses on the experimental identification methods of modal loss factor, frequency band loss factor and loss factor of stiffness nonlinear system, in order to improve the accuracy of the loss factor test. In this paper, based on the Gao Si Newton iteration, a fitting method for identifying the modal loss factor is proposed. The method uses as much information near the natural frequency as possible to calculate the modal damping. The effect of measurement error on the result of loss factor identification is effectively eliminated. According to the type of fitting function, fitting method can be divided into direct fitting method, square fitting method and reciprocal fitting method. In this paper, the effects of frequency interval, frequency points near the natural frequency and the number of repeated experiments on the recognition accuracy of the improved half-power bandwidth method and the three fitting methods by linear interpolation are compared and analyzed. The results show that the noise resistance of the three methods is better than that of the improved half power bandwidth method. The direct fitting method and the reciprocal fitting method need less repeated tests and have high precision, so they are the first choice for modal loss factor identification of linear systems. The SNR can be effectively improved by using the improved half-power bandwidth to identify the large loss factor error in the case of 10dB~20dB. Based on the analysis of the initial attenuation rate method to identify the frequency band loss factor error, the concept of process loss factor is proposed to describe the frequency band loss factor of the attenuation process. The analysis shows that the mode with small natural frequency has a large modal loss factor, which is the main reason for the large system error in the measurement of the band loss factor by the initial attenuation rate method. In this paper, the cyclic phase subtraction method is developed for frequency band loss factor identification, and a time domain separation method is proposed to measure the frequency band loss factor. The method is characterized by simple calculation process, wide application range and high measurement accuracy. The simulation and experimental results show that the error of the time domain separation method can be controlled within 10%, which can be used to identify the loss factor of the frequency band accurately, and can effectively solve the problem when the frequency band contains multi-order modes. It is difficult to identify the frequency band loss factor by the attenuation method due to the "segmental" phenomenon in the attenuating vibration curve. Finally, based on the analysis of the damping characteristics of the nonlinear stiffness system, an integral method for identifying the loss factor of the system is proposed based on particle swarm optimization (PSO). The results show that the method has strong noise resistance, and the loss factor can be obtained by using the noisy signal. When the SNR is as low as 10dB, the average error can still be controlled within 10%, and the higher the SNR, the smaller the error. In this paper, the time-domain separation method with multi-modal frequency band loss factor is proposed, which makes up for the deficiency of the attenuation method and the input power method, and overcomes the problem of systematic error in the initial attenuation rate method, which is convenient for engineering application. The integral method proposed in this paper is used to measure the loss factor of nonlinear system with stiffness. The process quantity is calculated by integral method, and the influence of test error on the result is reduced, and the loss factor is solved by particle swarm optimization, and the result is stable. The analysis shows that the method has the characteristics of strong noise resistance and high accuracy.
【学位授予单位】：西北工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TB535

【相似文献】