METHODOLOGICAL FOUNDATIONS FOR THE CONSTRUCTION AND USE OF CALIBRATION MODELS OF A FLOW QUALITY ANALYZER AT A GASOLINE COMPOUNDING STATION

  • Almaz Sh. Zianurov Perm National Research Polytechnic University
  • Alexander E. Primak Perm National Research Polytechnic University
  • Alexander G. Shumikhin Perm National Research Polytechnic University
Keywords: compounding of gasoline, quality, flow analyzer, IR spectrum, calibration model of the flow analyzer, neural networks

Abstract

This article presents the methodological foundations for constructing calibration models of the connection of the IR spectra of a flow analyzer and gasoline quality indicators. To solve the problem of developing models, a chemometric approach was used in combination with the principal component method, as well as neural network technology. The results of the analysis of the developed models for the relationship between the quality indicators for Premium-95 and Regular-92 gasoline according to the research and motor methods and their components based on the results of IR spectrophotometry are presented in comparison with laboratory analysis data. The IR absorption spectra of various types of automobile gasoline vary within the wavelength range from 800 to 1700 nm. Previous studies have established that the wavelength range with the best information content for determining the values of the quality indicators for gasoline is between 1100 and 1650 nm. Data processing by the principal component method is performed using graphs of accounts and influences. The invoice graph allows you to evaluate the grouping of samples by their properties, and the influence graph evaluates the degree of influence and completeness of the description of each sample by the resulting model. The data on the above-mentioned gasoline were included in one sample due to the proximity of the component composition. The analysis of the influence graph showed that the necessary and enough main components is equal to three for the model according to the motor method and four for the model according to the research method. The article provides data for Premium-95 gasoline to demonstrate the results of model testing. The volume of the training sample with spectra varied from 153 to 450 spectra. The accuracy of the developed models was evaluated using such indicators (metrics) as the average forecast error, the maximum forecast error, and the correlation coefficient. The results of the development and research of calibration models for assessing the quality indicators of gasoline and their components according to the measurements of the flow analyzer also showed that models acceptable in terms of forecasting accuracy are built with no less efficiency and accuracy using a neural network approach, in comparison with models based on the method of main components currently used in one of the gasoline production facilities. Neural network-based models are flexible and allow for indirect consideration of the values affecting the infrared spectra in the models. According to the available experimental data, acceptable neural network models were obtained for those quality indicators for which the principal component method proved ineffective.

For citation:

Zianurov A.Sh., Primak A.E., Shumikhin A.G. Methodological foundations for the construction and use of calibration models of a flow quality analyzer at a gasoline compounding station. ChemChemTech [Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol.]. 2024. V. 67. N 9. P. 118-125. DOI: 10.6060/ivkkt.20246709.6952.

References

Becker Yu. Spectroscopy. M.: Tekhnosfera. 2009. 528 p. (in Russian).

Anosov A.A., Efitov G.L., Zusman S.D. // Avtomatizatsiya Prom-sti. 2012. N 7. P. 41-47 (in Russian).

Kaixun He, Maiying Zhong, Zhi Li, Jingjing Liu. // Control Eng. Practice. 2020. V. 101. 104478. DOI: 10.1016/j.conengprac. 2020. 104478.

Fazal Mabooda, Syed Abdullah Gilani, Mohammed Albroumi, Saif Alameri, Mahmood M.O. Al Nabhania, Farah Jabeen, Javid Hussain, Ahmed Al-Harrasi, Ricard Boque, Saima Farooq, Ahmad M. Hamaed, Zakira Naureen, Alamgir Khan, Zahid Hussain // Fuel. 2017. V. 194. P. 388-396. DOI: 10.1016/j.fuel.2017.02.041.

Ge Z. // Ind.. Eng. Chem. Res. 2018. V. 57 (38). P. 12646-12661. DOI: 10.1021/acs.iecr.8b02913.

He K., Zhong M., Fang J., Li Y. // IEEE Trans.. Ind. Informatics. 2020. V. 16 (4). P. 2799-2808. DOI: 10.1109/ TII.2019.2945001.

Wang X., Yang K., Kalivas J.H. // Optik. 2020. V. 200. 163325. DOI: 10.1016/j.ijleo.2019.163325.

Zianurov A.Sh., Shumikhin A.G. // Inzh. Vestn. Dona. 2021. N 10. P. 22-31 (in Russian).

Bachoc F., Gamboa F., Loubes J.-M., Venet N.A. // IEEE Trans. Inform. Theory. 2017. V. 64. P. 6620-6637. DOI: 10.1109/TIT.2017.2762322.

Chan L.L.T., Qing-Yang W., Junghui C. // Chemomet. Intell. Lab. Syst. 2018. V. 175. P. 87-103. DOI: 10.1016/j.chemolab.2018.01.015.

Zianurov A.Sh., Shumikhin A.G. // Mezhdunar. nauch.-prakt. konf. «Innovacionnye proizvodstvennye tekhnologii i resursosberegayushchaya energetika” 8, 9 dekabrya 2021. Omsk. 2021. P. 163-168 (in Russian).

Silaev D.V., Shestopalova N.B., Fomina Yu.A., Rusanova T.Yu. // ChemChemTech [Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol.]. 2022. V. 65. N 2. P. 50-59 (in Russian). DOI: 10.6060/ivkkt.20226502.6497.

Akhnazarova S.L., Kafarov V.V. Methods of optimization of experiment in chemical technology. M.: Vyssh. shk. 1985. 327 p. (in Russian).

Esbensen K. Analysis of multidimensional data. Selected chapters. Chernogolovka: Izd-vo IPHF RAN. 2005. 160 p. (in Russian).

Neural networks in Matlab: A practical guide. SPb.: Balt. Gos. Tekhn. Un-t. 2017. 165 p. (in Russian).

Primak A.E., Shumikhin A.G., Stashkov S.I. // Vest. PNRPU. Khim. Tekhnol. Biotekhnol. 2012. N 14. P. 34-43 (in Russian).

Zianurov A.Sh., Shumikhin A.G. // Mater. vseross. nauch.-praktich. Konf. (s mezhdunar. uch.) 28–29 aprelya 2022. Perm'. P. 186-190 (in Russian).

Shurygina L.I., Surovoy E.P. Methods of optimization of chemical experiment. Part 2: Regression analysis and statistical planning of the experiment. Kemerovo: Kemerov. Gos. univ. 2011. 67 p. (in Russian).

Koryachko V. P., Bakuleva M. A., Oreshkov V. IIntelligent systems and fuzzy logic. M.: Kurs. 2020. 352 с. (in Russian).

Rostovtsev V.S. Artificial neural networks. SPb: Lan. 2021. 216 p. (in Russian).

Kazakov A.V., Subbotin E.V. Mathematical modeling and processing of experimental results. Perm: PNIPU. 2021. 58 p. (in Russian).

Moises B.B., Plotnikova I.V., Redko L.A. Statistical methods of quality control and experimental data pro-cessing. M.: Yurayt. 2021. 119 p. (in Russian).

Lemeshko B.Yu., Lemeshko B.V., Semenova M.A. // Vest. Tomsk Gos. Univ. Upravl. Vychisl. Tekh. Inform. 2018. N 44. P. 40–49 (in Russian). DOI: 10.17223/19988605/44/5.

Chistyakova T.B., Polosin A.N., Meshalkin V.P., Razygraev A.S. // ChemChemTech [Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol.]. 2021. V. 64. N 8. P. 90-98 (in Russian). DOI: 10.6060/ivkkt.20216408.6428.

Sidnyaev N.I. Theory of experiment planning and analysis of statistical data. M.: Izd-vo Yurayt. 2020. 495 p. (in Russian).

Published
2024-07-02
How to Cite
Zianurov, A. S., Primak, A. E., & Shumikhin, A. G. (2024). METHODOLOGICAL FOUNDATIONS FOR THE CONSTRUCTION AND USE OF CALIBRATION MODELS OF A FLOW QUALITY ANALYZER AT A GASOLINE COMPOUNDING STATION. ChemChemTech, 67(9), 118-125. https://doi.org/10.6060/ivkkt.20246709.6952
Section
CHEMICAL TECHNOLOGY (inorganic and organic substances. Theoretical fundamentals)