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

作　　者：卢素敏[1] 郝悦 石梓彤 初园园[1] 张妍[2] 卞希慧[1,3] LU Su-Min;HAO Yue;SHI Zi-Tong;CHU Yuan-Yuan;ZHANG Yan;BIAN Xi-Hui(School of Chemical Engineering and Technology,Tiangong University,Tianjin 300387,China;Research Center for Analytical Sciences,Collegeof Chemistry,Nankai University,Tianjin 300071,China;Key Laboratoryfor Technology Research and Evaluation of Drug Products,National MedicalProducts Administration,Shandong University,Jinan 250012,China)

机构地区：[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]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

出　　处：《分析化学》2024年第9期1277-1286,共10页Chinese Journal of Analytical Chemistry

基　　金：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(Nos.2022TREDP04,2023TREDP01)资助。

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。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 toa 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 

分 类 号：O657.3[理学—分析化学]