采样时间长度对阻尼识别影响的研究

发布时间：2018-10-31 08:24

【摘要】：阻尼作为重要的动力学特性参数,表征振动系统能量耗散的快慢。能否准确识别出阻尼值,直接影响阻尼在结构损伤诊断、减震降噪等领域的应用。改进阻尼识别方法,提高阻尼识别精度,一直是国内外学者所关注的重要课题。各种阻尼识别方法中,学者们对采样时间长度研究分析不足。本文针对采样时间长度对阻尼识别影响,展开了研究工作。本文首先分析无噪声干扰情况下,采样时间长度对阻尼识别精度的影响。以半功率带宽法与内积法为例展开研究,给出阻尼比为0.006、0.03、0.1时,500采样点、5000采样点、40000采样点的仿真算例。仿真结果表明无噪声干扰时,两种阻尼识别方法均存在采样时间越长阻尼识别精度越高的现象。实际工程不可避免地存在噪声干扰,研究存在噪声干扰时,采样时间长度对阻尼识别精度的影响。分别推导响应信号频谱幅值和噪声信号频谱幅值与采样点数的关系式。理论分析指出采样时间长度不宜过长,否则会出现响应信号被噪声信号掩盖,阻尼识别误差增大的现象。本文将响应信号与噪声信号频谱相等时,所对应的采样时间长度定义为临界采样长度,通过理论推导得到临界采样长度表达式,为选取合理采样时间长度提供参考。针对有噪声干扰时采用过长采样时间会带来阻尼识别误差的问题,研究采用联立方程法使用较短数据识别信号阻尼。运用傅里叶变换初估信号固有频率,以此固有频率构造傅里叶变换基函数,通过对信号与构造函数作两次时间不同的积分运算构造方程组,求解得到固有频率误差及衰减系数,进而得到信号阻尼比。对方程组积分时间,讨论任意积分时间与特定积分时间T(28)k??(k为正整数)两种方案,分析指出选用特定积分时间能降低负频率干扰,提高阻尼识别精度。联立方程法核心运算选取积分时间较短,使噪声干扰影响不突出,具有运算量少、识别速度快、抗噪能力强的特点。研究联立方程法对多自由度响应信号阻尼识别的适用性。对非密集模态信号,将识别出的各阶次模态依次从原信号中减去,多自由度响应信号阻尼识别简化为多个单自由度响应信号阻尼识别。对密集模态信号引入迭代算法,消除各阶次模态的相互干扰,准确识别各阶次模态阻尼。将联立方程法运用于齿轮箱盖阻尼识别试验,识别出齿轮箱盖响应信号在208.824 Hz、363.913 Hz、636.561 Hz、940.817 Hz处具有四阶模态,对应的四阶模态阻尼比分别为0.004798、0.004509、0.001933和0.001674。将识别出四阶模态信号从原信号中减去,剩余信号频域图中观测不到明显峰值,证明联立方程法准确有效,具有工程实用性。
[Abstract]:As an important dynamic characteristic parameter, damping represents the speed of energy dissipation of vibration system. Whether the damping value can be accurately identified will directly affect the application of damping in the fields of structural damage diagnosis, vibration absorption and noise reduction, etc. Improving the damping identification method and improving the accuracy of damping identification has always been an important subject that scholars at home and abroad pay close attention to. Among all kinds of damping identification methods, the research and analysis of sampling time length is insufficient. In this paper, the influence of sampling time length on damping identification is studied. In this paper, the influence of sampling time length on the accuracy of damping identification is analyzed. Taking the half-power bandwidth method and inner product method as an example, a simulation example of the damping ratio of 0.006 ~ 0.03,0.1, 500 sampling points, 5000 sampling points and 40000 sampling points is given. The simulation results show that the longer the sampling time, the higher the accuracy of damping identification for both damping identification methods without noise interference. The influence of sampling time length on the accuracy of damping identification is studied. The relationship between the spectrum amplitude of response signal and the amplitude of noise signal and the number of sampling points are derived respectively. The theoretical analysis indicates that the sampling time should not be too long, otherwise the response signal will be masked by the noise signal and the damping identification error will increase. In this paper, when the spectrum of response signal and noise signal is equal, the corresponding sampling time length is defined as critical sampling length, and the expression of critical sampling length is obtained by theoretical derivation, which provides a reference for selecting reasonable sampling time length. In order to solve the problem of damping identification error caused by long sampling time when there is noise interference, the simultaneous equation method is used to identify the damping of the signal with short data. Using the Fourier transform to estimate the natural frequency of the signal, the Fourier transform basis function is constructed, and the equations are constructed by integrating the signal and the construction function at different times, and the natural frequency error and attenuation coefficient are obtained. Then the signal damping ratio is obtained. For the integral time of equations, two schemes of arbitrary integral time and special integral time T _ (28) K ~? (k is a positive integer) are discussed. It is pointed out that the selection of specific integral time can reduce the negative frequency disturbance and improve the damping identification accuracy. The integration time of the core operation of simultaneous equation method is short, which makes the influence of noise interference not prominent, and has the characteristics of less computation, faster recognition speed and stronger anti-noise ability. The applicability of simultaneous equation method to the identification of damping of response signals with multiple degrees of freedom is studied. For the non-dense modal signal, the identified modes are subtracted from the original signal in turn, and the damping identification of the multi-degree-of-freedom response signal is simplified to that of multiple single-degree-of-freedom response signals. The iterative algorithm is introduced to the dense modal signal to eliminate the interference of each order mode and to identify the damping of each order mode accurately. The simultaneous equation method is applied to the gear box cover damping identification test. It is found that the response signal of the gear box cap has the fourth order mode at 208.824 Hz,363.913 Hz,636.561 Hz,940.817 Hz. The corresponding fourth-order mode damping ratios are 0.004798 / 0.004509N 0.001933 and 0.001674 respectively. The fourth order modal signal is subtracted from the original signal and no obvious peak value can be observed in the frequency domain diagram of the remaining signal. It is proved that the simultaneous equation method is accurate and effective and has engineering practicability.
【学位授予单位】：江苏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB53

【参考文献】