Stationary non-Gaussian random vibration control: A review

This paper presents a review of the various methods for the stationary non-Gaussian random vibration control. Random vibration tests can be divided, according to the number of exciters, in single-shaker tests and multiple-shaker tests. In the stationary non-Gaussian random vibration test, the time and frequency domain characteristics of the responses should be controlled independently and simultaneously. Skewness and kurtosis are usually selected as the non-Gaussian time control references (targets) while power spectral density is the frequency domain control procedure before it recalls the concepts of non-Gaussianity. Then, the generation of a one frame stationary non-Gaussian random signal for both the single and multiple shakers are reviewed. The commonly used methods for the single non-Gaussian random signal generation in the random vibration test are memoryless nonlinear transformation, phase modification and Filtered Poisson process. For the multiple-shaker case, the sequential phase modification and memoryless nonlinear transformation are used to generate one frame coupled multi-channel non-Gaussian random signal. In order to obtain a stationary and consecutive dynamic input, the time domain randomization procedure is introduced with high computational efficiency and its influences on the skewness and kurtosis are analyzed. Finally, two existing problems in the non-Gaussian random vibration control are addressed.     

能量匹配条件下的CMA盲均衡算法

对CMA盲均衡技术在海上靶场遥测信号接收中的应用进行了仿真研究,同时在CMA盲均衡算法的基础上提出了一种能量匹配约束条件,对在能量匹配约束条件下的CMA算法进行了理论分析和算法推导。根据峰度准则,输入输出信号能量相等是盲均衡实现的充要条件的前提,依据这一前提,对CMA算法代价函数进行修正,有效提高了算法的收敛速度和均衡性能。     

高光谱图像融合最佳波段选择方法

针对高光谱图像高数据维给图像处理带来的困难和影响,本文构造了高光谱图像融合的最佳波段选择新模型-联合偏度-峰度指数(Joint Skewness-Kurtosis figure,JSKF)模型,利用JSKF指数进行自适应子空间的分解和波段选择,降低高光谱数据维数;并将选择出的最佳波段组合进行了融合,实验结果表明,该方法所选择的波段信息差异较大、互补特征明显,融合后图像包含的信息量丰富,效果优于传统的自适应波段选择方法和主成分分析累计贡献率方法.     

基于WPD-KVI-Hilbert变换相结合的滚动轴承早期故障特征精准识别

为了能够有效地从轴承早期故障激励的高频振动信号中提取出故障特征信息,基于最优小波包基选取方法和峭度值最大筛选原则,提出了一种改进的小波包分解（WPD）、峭度值指标（KVI）与Hilbert变换相结合的滚动轴承早期故障特征识别方法。计算选取最优小波包基,确定分解层数;采用WPD方法对轴承故障振动信号进行分解,获得若干个Node分量;基于峭度值指标最大原则筛选出有效的Node分量进行信号重构;对重构信号进行包络解调分析,提取出故障特征频率对轴承故障进行诊断。采用建立的方法对凯斯西储大学滚珠轴承外圈、内圈故障实验数据和自行开展的滚棒轴承外圈、滚动体故障实验数据进行了分析与诊断。研究结果表明：该方法能够有效提高故障信号高频分辨率、保留周期性冲击成分,并能准确有效提取出滚珠和滚棒轴承故障特征频率的1~7倍频及其与轴转频调制的系列边频带频率,实现对滚动轴承故障特征的精准识别与故障诊断。