基于随机共振的滚动轴承微弱特征检测技术应用研究

发布时间：2018-11-10 13:09

【摘要】：双稳随机共振是一种非线性现象，它只有在噪声、双稳系统和输入信号三者之间达到一定的协同关系时才会发生。具体针对某一确定的大参数含噪信号，需要对随机共振的多个参数进行联合调节，才能产生随机共振。但是，目前关于系统参数、信号频率以及噪声强度三者之间的耦合性的研究很少，随机共振的联合调参缺乏指导，只能依赖于人为经验。因此，本文将围绕随机共振多参数联合调节实现困难的问题，对自适应随机共振展开研究。针对随机共振系统参数选取依赖于人为经验和变步长随机共振计算步长选取困难的缺陷，提出一种采用粒子群算法的自适应变步长随机共振方法，该方法以随机共振输出信噪比作为粒子群算法的适应度函数，通过对系统参数和计算步长的同步优化，实现变步长随机共振最优输出的自适应求解。利用仿真数据和工程实际数据对该方法进行实验验证，分析结果表明该方法简单易行，适用范围广，收敛速度快，能有效的检测出强噪声背景下的高频微弱信号，具有良好的工程应用前景。把采用粒子群算法的自适应变步长随机共振融入到级联双稳随机共振中，提出了级联双稳随机共振自适应降噪方法，通过对各级随机共振参数的自适应求解，实现了级联随机共振各级系统的自适应输出，最终达到对大参数条件下的工程信号进行自适应降噪的目的。将该方法应用于仿真数据和工程实测数据，分析结果表明该方法能快速、有效的消除大参数信号里的高频噪声干扰，并突出低频有用信号成分，降噪效果显著。将级联双稳随机共振自适应降噪方法与经验模式分解相结合，提出了基于级联双稳随机共振自适应降噪的经验模式分解方法，该方法可以消除经验模式分解的边界效应并有效提高分解效率。在实验研究中，分别对原始输入信号和级联双稳各级输出信号进行了经验模式分解，通过对这些分解结果进行比较分析，，可以发现该方法能在提高信号信噪比的同时，减少IMF分量的数量，并提高经验模式分解的质量。
[Abstract]:Bistable stochastic resonance (bistable stochastic resonance) is a nonlinear phenomenon that occurs only when there is a certain synergy among noise, bistable system and input signal. For a certain large parameter noisy signal, it is necessary to jointly adjust the multiple parameters of stochastic resonance in order to produce stochastic resonance. However, there are few studies on the coupling among system parameters, signal frequency and noise intensity, and the joint parameter tuning of stochastic resonance is unguided, which can only depend on human experience. Therefore, the study of adaptive stochastic resonance is carried out in this paper, focusing on the difficult problem of multi-parameter joint regulation of stochastic resonance. In order to solve the problem that the parameter selection of stochastic resonance system depends on artificial experience and the difficulty of selecting step size for variable step size stochastic resonance calculation, an adaptive variable step stochastic resonance method based on particle swarm optimization algorithm is proposed. The SNR of the stochastic resonance output is taken as the fitness function of the particle swarm optimization algorithm. The adaptive solution of the variable step size stochastic resonance optimal output is realized by synchronously optimizing the system parameters and the computational step size. The simulation data and engineering data are used to verify the method. The analysis results show that the method is simple, wide and convergent, and can effectively detect the high frequency weak signal in the background of strong noise. It has a good prospect of engineering application. The adaptive variable step size stochastic resonance using particle swarm optimization algorithm is integrated into cascade bistable stochastic resonance, and an adaptive noise reduction method of cascade bistable stochastic resonance is proposed. The adaptive output of cascaded stochastic resonance system is realized, and the purpose of adaptive noise reduction for engineering signals under large parameters is achieved. The method is applied to the simulation data and the engineering measured data. The analysis results show that the method can eliminate the high frequency noise interference in the large parameter signal quickly and effectively, and highlight the low frequency useful signal components, and the noise reduction effect is remarkable. Combining the cascaded bistable stochastic resonance adaptive noise reduction method with empirical mode decomposition, an empirical mode decomposition method based on cascaded bistable stochastic resonance adaptive noise reduction is proposed. This method can eliminate the boundary effect of empirical mode decomposition and improve the efficiency of decomposition. In the experimental study, the original input signal and cascade bistable output signal are decomposed by empirical mode decomposition. By comparing and analyzing these decomposition results, it can be found that this method can improve the signal to noise ratio at the same time. Reduce the number of IMF components and improve the quality of empirical mode decomposition.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH133.33;TP18

【参考文献】