Gas cells are widely used in THz and IR absorption spectroscopy. However, reflections of an optical wave on cell input and output windows cause a gas sample measured spectrum distortion due to the Fabry-Perot effect. The hardware approach to decrease the Fabry-Perot effect in the THz spectral range is suggested. The approach is based on using a standard gas cell and the additional movable optically transparent window. To illustrate usefulness of the suggested approach, a simulation of registration of the sulfur dioxide absorption band near 660 GHz in the gas mixture consisted of 0.1 % SO2 and 99.9 % N2 was conducted. The absorption band profile fluctuations due to the Fabry-Perot effect were shown to be reduced by almost 4 times with the suggested approach implementation in relation to the default gas cell. This approach can be combined with other hardware and software variants of the Fabry-Perot effect suppression.

A preliminary experimental absorption spectrum denoising improves the further analysis. The majority of popular filters causes not only a nose decreasing but also a spectrum shape distortion. An approach to a gasmixture IR absorption spectrum denoising using a multilayered perceptron deep neural network (MLP DNN) with autoencoder architecture was suggested. This deep neural network was trained and tested on gas mixtures related closely to the ground atmosphere. Absorption spectra of the latter were calculated in the 0.9- 11 mu m spectral range using data from the HITRAN database, then a random noise was added. The results of the MLP DNN filter application were compared with the standard Gaussian filter. In common, MLP DNN filter provided more effective noise decreasing and less initial spectrum shape distortion compared to the Gaussian filter.

An approach to multicomponent gas mixtures IR absorption spectra decomposition using a deep neu-ral network was developed. The process of refinement and optimization of the absorption spectra data model to improve the accuracy of the inverse spectroscopic task solution is described. A criterion for the reliability of restoring the concentration of an individual component in a gas mixture based on its share of the area under the absorption spectrum curve was suggested and tested.(c) 2023 Elsevier Ltd. All rights reserved.

The laser sources used in absorption spectroscopy of gas media are a compromise between spectral turn-ability and line width. For example, optical parametric oscillators have a very wide tuning range, but also have a rather wide laser radiation linewidth. To mitigate the disadvantages of the latter, an approach to absorption spectroscopy gas-analysis spec-tral resolution improving using super-resolution (SR) reconstruction is proposed. It was implemented us -ing several machine learning models based on different artificial neural network (ANN) architectures, including an original sequential ensemble ANN approach. The problem of random noise influence on SR reconstruction quality was resolved in two ways: (i) by learning convolutional neural networks with noisy spectra, (ii) by high-frequency noise preliminary decreasing, using Gaussian or Fast Fourier Transform fil -tering.The following ANN architecture models were designed and tested: convolutional neural network (CNN) and multilayer perceptron (MLP). The former performed at a lower accuracy compared to the lat-ter.ANNs' sequential combination was implemented when each subsequent ANN used the results of pre-vious ANN data processing. This architecture pursues a paradigm of ensemble algorithms. Sequential models consisting of two or five MLP ANNs were designed and tested. In general, at low noise, the se-quential models provided better SR reconstruction quality compared to the single-stage MLP ANN. When the noise amplitude was 4% and more, the sequential models demonstrated 3-8% worse accuracy than the single-stage MLP ANN, even using filtering. Therefore, the sequential models are quite accurate and effective in combination with effective filtering in cases of moderate noise level.(c) 2022 Elsevier Ltd. All rights reserved.

The regression model was applied to solve the problem of restoration of the concentration of components in a gas mixture using infrared absorption spectra. A solution to the problem of determining the concentrations of individual components of a multicomponent gas mixture is suggested and tested.

The blood serum samples from healthy rats and rats with inoculated tumors 14 and 28 days after transplantation of cholangiocarcinoma cells were studied using Raman spectroscopy. It was shown that the amide I band intensity differs in the spectra depending on tumor stage and it correlates with the serum protein concentration. The blood glucose concentration increases in some rats, leading to protein glycation and the amide 1 band change. The band at 1670 cm-1 is most pronounced in the spectra on 28 day of the experiment and may be associated with an increase of β-structural elements in protein conformation. The principal component analysis allows evaluating the differences in Raman spectra of healthy rats and rats on day 28 after tumor transplantation.

An important role in component analysis with spectral methods has a spectral resolution of used tools. The most useful and perspective methods to improve spectral resolution is decreasing of impulse response function (IRF) and improving resolution using superresolution (SR) reconstruction methods. We have analyzed different types of neural networks (convolution neural network, multilayered perceptron) for improving the spectral resolution of initial absorption spectra. The used approach is based on an association of a high-resolution and a low-resolution spectrum. The latter was constructed from high-resolution spectra to which IRF and some random noise were added. Highresolution spectra were generated using the HITRAN database. Most optimal architectures of neural networks to improve spectral resolution were defined.