Spectral denoising based on Hilbert–Huang transform combined with F-test

Enhancing spatial resolution in Fourier transform infrared spectral image via machine learning algorithms

Owing to the intrinsic signal noise in the characterization of chemical structures through Fourier transform infrared (FT-IR) spectroscopy, the determination of the signal-to-noise ratio (SNR) depends on the level of the concentration of the chemical structures. In situations characterized by limited concentrations of chemical structures, the traditional approach involves mitigating the resulting low SNR by superimposing repetitive measurements. In this study, we achieved comparable high-quality results to data scanned 64 times and superimposed by employing machine learning algorithms such as the principal component analysis and non-negative matrix factorization, which perform the dimensionality reduction, on FT-IR spectral image data that was only scanned once. Furthermore, the spatial resolution of the mapping images correlated to each chemical structure was enhanced by applying both the machine learning algorithms and the Gaussian fitting simultaneously. Significantly, our investigation demonstrated that the spatial resolution of the mapping images acquired through relative intensity is further improved by employing dimensionality reduction techniques. Collectively, our findings imply that by optimizing research data through noise reduction enhancing spatial resolution using the machine learning algorithms, research processes can be more efficient, for instance by reducing redundant physical measurements.

The application of generalized S-transform in the denoising of surface plasmon resonance (SPR) spectrum

In order to obtain accurate resonance peaks from surface plasmon resonance (SPR) spectral curves, a reasonable denoising method is of great significance for SPR sensing systems. Therefore, the generalized S-transform is combined with the Bald Eagle Search algorithm (BES) in this study, and a denoising method based on the generalized S-transform optimized by BES is proposed and applied to the denoising processing of the SPR spectrum. First, a fiber SPR sensing system is used to obtain the original noised spectrum; then, the generalized S-transform is performed to obtain the corresponding S-domain spectrum. Next, the denoising threshold λn is optimized by the BES algorithm, which is used to denoise and reconstruct the SPR reflection spectrum. Finally, two fitness functions are evaluated until the optimal denoising threshold λn and denoising effect are obtained. The relevant validation experiments are completed, and the experimental results show that the proposed method has the best denoising performance when p is between 0.5 and 1, and λ is between 1.5 and 2.5. Meanwhile, compared to the other denoising methods, the BES-S method can maintain a relatively stable denoising effect on the SPR spectrum with high or low levels of noise; the average values of root mean square error (RMSE) and signal-to-noise ratio (SNR) are 0.27 and 23.61, respectively. Ranking first in terms of comprehensive denoising performance, it can also maintain the original shape of the SPR spectrum and better reflect its characteristic peak while filtering out noise. This method can overcome the problem of arbitrary selection of basic functions and thresholds in conventional denoising methods, and it can improve the detection accuracy of SPR sensors and provide a new idea for SPR spectrum denoising, which also lays the foundation for the application of substance composition detection based on SPR sensor.

Intelligent Fault Diagnosis Method Based on VMD-Hilbert Spectrum and ShuffleNet-V2: Application to the Gears in a Mine Scraper Conveyor Gearbox

This paper introduces a fault diagnosis method for mine scraper conveyor gearbox gears using motor current signature analysis (MCSA). This approach solves problems related to gear fault characteristics that are affected by coal flow load and power frequency, which are difficult to extract efficiently. A fault diagnosis method is proposed based on variational mode decomposition (VMD)-Hilbert spectrum and ShuffleNet-V2. Firstly, the gear current signal is decomposed into a series of intrinsic mode functions (IMF) by using VMD, and the sensitive parameters of VMD are optimized by using a genetic algorithm (GA). The Sensitive IMF algorithm judges the modal function sensitive to fault information after VMD processing. By analyzing the local Hilbert instantaneous energy spectrum for fault-sensitive IMF, an accurate expression of signal energy changing with time is obtained to generate the local Hilbert immediate energy spectrum dataset of different fault gears. Finally, ShuffleNet-V2 is used to identify the gear fault state. The experimental results show that the accuracy of the ShuffleNet-V2 neural network is 91.66% after 778 s.

Comparison of various chemometric methods on visible and near-infrared spectral analysis for wood density prediction among different tree species and geographical origins

Visible and near-infrared (Vis-NIR) spectroscopy has been widely applied in many fields for the qualitative and quantitative analysis. Chemometric techniques including pre-processing, variable selection, and multivariate calibration models play an important role to better extract useful information from spectral data. In this study, a new de-noising method (lifting wavelet transform, LWT), four variable selection methods, as well as two non-linear machine learning models were simultaneously analyzed to compare the impact of chemometric approaches on wood density determination among various tree species and geographical locations. In addition, fruit fly optimization algorithm (FOA) and response surface methodology (RSM) were employed to optimize the parameters of generalized regression neural network (GRNN) and particle swarm optimization-support vector machine (PSO-SVM), respectively. As for various chemometric methods, the optimal chemometric method was different for the same tree species collected from different locations. FOA-GRNN model combined with LWT and CARS deliver the best performance for Chinese white poplar of Heilongjiang province. In contrast, PLS model showed a good performance for Chinese white poplar collected from Jilin province based on raw spectra. However, for other tree species, RSM-PSO-SVM models can improve the performance of wood density prediction compared to traditional linear and FOA-GRNN models. Especially for Acer mono Maxim, when compared to linear models, the coefficient of determination of prediction set ( R p 2 ) and relative prediction deviation (RPD) were increased by 47.70% and 44.48%, respectively. And the dimensionality of Vis-NIR spectral data was decreased from 2048 to 20. Therefore, the appropriate chemometric technique should be selected before building calibration models.

Quantitative analysis of polycyclic aromatic hydrocarbons (PAHs) in water by surface-enhanced Raman spectroscopy (SERS) combined with Random Forest

Polycyclic aromatic hydrocarbons (PAHs) have strong carcinogenicity, teratogenicity, mutagenicity and other adverse effects on human beings. They are one of the most dangerous pollutants, which have attracted great attention in the past decades. In this work, aiming at the actual problems that water environment is polluted and human health is threatened by PAHs, surface enhanced Raman spectroscopy (SERS) combined with Random Forest (RF) calibration models were used to quantitative analysis of phenanthrene and fluoranthene in water. Firstly, the SERS data was collected after samples mixed with Ag NPs, after 31 PAHs samples were prepared. Secondly, it was discussed how spectral preprocessing integration strategies affect on the prediction performance of the RF calibration models. And then, the effect of mutual information (MI) variable selection method on the performance of RF calibration models was explored. Finally, the RF calibration models were established for phenanthrene and fluoranthene. For the prediction set, a lowest mean relative error (MRE) and a largest determination coefficient (R²) were obtained. For quantitative analysis of phenanthrene, the final prediction performance results show that R²_p is 0.9780, and MRE_p is 0.0369 based on the D1st-WT-RF calibration model. For fluoranthene, WT-D1st-MI-RF is a better calibration model, and corresponding to R²_p and MRE_p are 0.9770 and 0.0694, respectively. Hence, a rapid and accurate quantitative method of PAHs is established for the real-time detection of water environmental pollution, which is intended to provide new ideas and methods for the quantitative analysis of PAHs in water.