峰值提取结合变分模态分解的复杂样品光谱去噪方法研究
Spectral Denoising Based on Peak Extraction Combined with Variational Mode Decomposition for Complex Samples
卢素敏 1郝悦 1石梓彤 1初园园 1张妍 2卞希慧3
作者信息
- 1. 天津工业大学化学工程与技术学院,天津 300387
- 2. 南开大学化学学院,分析科学研究中心,天津 300071
- 3. 天津工业大学化学工程与技术学院,天津 300387;山东大学,国家药品监督管理局药物制剂技术研究与评价重点实验室,济南 250012
- 折叠
摘要
为解决变分模态分解(Variational mode decomposition,VMD)用于包含尖锐峰值的光谱去噪时产生的峰损失的问题,本研究提出了峰值提取结合变分模态分解(Peak extraction variational mode decomposition,PE-VMD)的复杂样品光谱信号去噪方法.首先,采用VMD对光谱信号进行去噪;然后,计算光谱信号的一阶导数以确定峰值中心,计算光谱信号的二阶导数以提取高信噪比的峰;最后,将VMD去噪后丢失信息的峰截取去除,剩余光谱与提取的尖锐峰值依次连接,得到最终的去噪光谱.将本方法用于模拟信号和双金属催化剂(MnCo-ISAs/CN)的X-射线衍射(X-ray diffraction,XRD)谱去噪,并与Savitzky-Golay(SG)平滑、经验模态分解(Empirical mode decomposition,EMD)和VMD方法进行比较,采用去噪前后的光谱图和信噪比评价去噪效果.结果表明,PE-VMD去噪具有最大的信噪比,并且有效保留了光谱信号的有用信息.因此,对于包含尖锐峰值的光谱,PE-VMD具有更优异的去噪能力.
Abstract
To address the issue of peak loss when applying variational mode decomposition(VMD)to denoise spectra with sharp peaks,in this study,a method for spectral signal denoising of complex samples called peak extraction variational mode decomposition(PE-VMD)was introduced.Firstly,the spectral signal was subjected to a process of denoising utilising the VMD algorithm.Next,the first order derivatives of the spectral signals were calculated to determine the peak center.Subsequently,the second order derivatives of the spectral signal was calculated to extract the sharp peaks with high signal-to-noise ratio(SNR).Finally,the peak intercepted that lose information after VMD denoising were removed,and the remaining spectrum was sequentially connected with the extracted sharp peaks to obtain the final denoised spectrum.The effectiveness of the method was evaluated by one of simulated signals and X-ray diffraction(XRD)spectrum of MnCo-ISAs/CN catalysts.Furthermore,the method was compared with other denoising techniques,including Savitzky-Golay(SG)smoothing,empirical mode decomposition(EMD)and VMD.The efficacy of the denoising process was then assessed by analyzing the spectrograms and signal-to-noise ratio before and after denoising.The results demonstrated that PE-VMD denoising achieved the greatest SNR and effectively preserved the essential information of the spectral signals.Consequently,PE-VMD exhibited superior denoising capability for spectra with sharp peaks.
关键词
光谱去噪/变分模态分解/峰值提取/X射线衍射Key words
Spectral denoising/Variational mode decomposition/Peak extraction/X-ray diffraction引用本文复制引用
基金项目
药物制剂技术研究与评价国家药品监督管理局重点实验室开放课题项目(2022TREDP04)
药物制剂技术研究与评价国家药品监督管理局重点实验室开放课题项目(2023TREDP01)
出版年
2024
国家科技期刊平台
NSTL
万方数据