壳段厚度激光检测信号的变分模态分解去噪
发布时间：2018-10-17 13:03
【摘要】：针对双激光位移传感器测量大型壳段厚度过程中噪声对检测精度的影响,提出利用变分模态分解来实现对厚度信号的自适应去噪,利用相邻固有模态函数之间的离散Hellinger距离来获取最佳的模态数。该方法将变分模态分解算法引入到激光信号的自适应滤波过程中,分析并改进了变分模态分解算法的过分解、欠分解以及能量泄露的问题。然后,对改进的变分模态分解与希伯特振动分解和自适应噪声总体集合经验模态分解进行性能对比,提出了固有模态函数的相对瞬时能量概率的概念。最后,结合离散Hellinger概率分布距离理论判断固有模态之间的信噪分界点,实现了对信号的重构及滤波处理。仿真和实验结果表明,该方法对壳段厚度信号处理的信噪比为39.27dB,比自适应噪声总体集合经验模态分解方法提高了10dB,具有良好的自适应性,无需先验条件便能快速有效地识别并分离激光信号中的噪声成分。
[Abstract]:In view of the effect of noise on the detection accuracy in the process of measuring the thickness of a large shell with a double laser displacement sensor, a variational mode decomposition (VMD) is proposed to realize the adaptive de-noising of the thickness signal. The optimal modal number is obtained by using the discrete Hellinger distance between adjacent inherent modal functions. In this method, the variational mode decomposition algorithm is introduced into the adaptive filtering process of laser signal, and the overdecomposition, underdecomposition and energy leakage of the variational mode decomposition algorithm are analyzed and improved. Then, the concept of relative instantaneous energy probability of inherent mode function is proposed by comparing the performance of improved variational mode decomposition with Hilbert vibration decomposition and adaptive noise set empirical mode decomposition. Finally, combining the discrete Hellinger probability distribution distance theory to judge the signal-noise boundary point between the natural modes, the signal reconstruction and filter processing are realized. The simulation and experimental results show that the SNR of this method is 39.27 dB for the shell thickness signal processing, which is 10 dB higher than the adaptive noise set empirical mode decomposition method. The noise components in laser signals can be quickly and effectively identified and separated without prior conditions.
【作者单位】： 哈尔滨工业大学电气及自动化学院;
【基金】：国家自然科学基金资助项目(No.61108073) 上海航天科技创新项目(No.SAST2015029)
【分类号】：TN249;V475.1
，
本文编号：2276743
[Abstract]:In view of the effect of noise on the detection accuracy in the process of measuring the thickness of a large shell with a double laser displacement sensor, a variational mode decomposition (VMD) is proposed to realize the adaptive de-noising of the thickness signal. The optimal modal number is obtained by using the discrete Hellinger distance between adjacent inherent modal functions. In this method, the variational mode decomposition algorithm is introduced into the adaptive filtering process of laser signal, and the overdecomposition, underdecomposition and energy leakage of the variational mode decomposition algorithm are analyzed and improved. Then, the concept of relative instantaneous energy probability of inherent mode function is proposed by comparing the performance of improved variational mode decomposition with Hilbert vibration decomposition and adaptive noise set empirical mode decomposition. Finally, combining the discrete Hellinger probability distribution distance theory to judge the signal-noise boundary point between the natural modes, the signal reconstruction and filter processing are realized. The simulation and experimental results show that the SNR of this method is 39.27 dB for the shell thickness signal processing, which is 10 dB higher than the adaptive noise set empirical mode decomposition method. The noise components in laser signals can be quickly and effectively identified and separated without prior conditions.
【作者单位】： 哈尔滨工业大学电气及自动化学院;
【基金】：国家自然科学基金资助项目(No.61108073) 上海航天科技创新项目(No.SAST2015029)
【分类号】：TN249;V475.1
，
本文编号：2276743
本文链接：https://www.wllwen.com/kejilunwen/dianzigongchenglunwen/2276743.html
