论文部分内容阅读
在转子轴承系统振动信号处理中,针对平稳信号的传统傅里叶变换精度较低、快变启动过程的非平稳信号频谱分析方法较复杂的问题,本文仿真构造了两类响应信号。通过对比给定信号参数与信号识别参数的误差研究了几种谱分析方法或过程的简便性和准确性。对转子系统振动平稳信号离散频谱分析时存在的误差进行了定量分析,利用比例插值法对误差进行校正,开发了高精度谱分析测试软件;分析了转子轴承系统快变过程非平稳振动信号的特征,探索了一种将t时空域非平稳信号转变为tn时空间域平稳信号的办法或过程,然后结合比例插值校正法对其进行频谱分析,再返回到t时空域获得某时刻的谱特征参数;构造了转子系统振动仿真信号检验了上述过程的准确性。研究结果表明:比例插值法提取的谱特征数据近乎与仿真信号设定值相等;针对本文构造的快变过程非平稳仿真信号,利用本文给出的谱分析过程产生的频率误差最大值为0.47%,幅值误差最大值为0.2%。本文的仿真研究为提出和考证新的谱分析方法提供了手段。
In the vibration signal processing of the rotor bearing system, two kinds of response signals are constructed by simulation in this paper. The traditional Fourier transform of the stationary signal is of low precision and the non-stationary signal spectrum analysis method of the fast starting process is relatively complicated. The simplicity and accuracy of several spectral analysis methods or processes were studied by comparing the errors of the given signal parameters with the signal identification parameters. The error existed in discrete frequency spectrum analysis of rotor system vibration-stable signal was quantitatively analyzed, and the error was corrected by proportional interpolation method. High-precision spectral analysis software was developed. The characteristics of non-stationary vibration signal during the rapid change of rotor bearing system were analyzed. , Explored a method or process of transforming the t-space-time non-stationary signal into a space-domain stationary signal at time tn, and then analyzed the spectrum by using the proportional interpolation method and returned to t at the time domain to obtain spectral characteristic parameters The rotor system vibration simulation signal is constructed to verify the accuracy of the above process. The results show that the spectral characteristic data extracted by the proportional interpolation is almost equal to the set value of the simulation signal. For the fast non-stationary simulation signal constructed in this paper, the maximum frequency error generated by the spectral analysis given in this paper is 0.47% , The maximum amplitude error of 0.2%. The simulation in this paper provides a means to propose and research new spectral analysis methods.