Integration of convolutional neural network into a chromatographic gas analyzer for continuous screening

The article is devoted to research on the integration of a verification neural network model into a chromatographic gas analyzer for the purpose of continuous screening. The use of a convolutional neural network makes it possible to determine the component and mass composition of a gas with high accuracy, as well as to simplify the design of a chromatographic gas analyzer. Safety in closed mining operations in coal and potash mines, as well as the timeliness of detecting the opening of a gas blister and its release into the working area, are important factors when selecting equipment and upgrading existing solutions. In order to eliminate the limitation on the frequency of measurements of the chromatographic analyzer and optimize the design, it is proposed to replace the separation columns with an artificial neural network that performs gas analysis of the spectrum image. It has been determined that for efficient processing of gas spectrum images, it is advisable to use a convolutional neural network model with ResNet-34 architecture. To train the neural network model, data obtained from certified chromatographs directly at the drilling site, as well as in laboratory conditions, was used. The model was also validated on real data by sequentially taking readings from a certified chromatograph and a chromatograph with a neural network. The accuracy of the spectrum classification according to the accuracy metric was 88.80%, and the sensitivity was 82.50%, which confirms the reliability of the gas mixture separation results. The results of testing the accuracy of component-by-component separation showed that the relative error in determining the total content of hydrocarbon gases, as well as methane separately, in the multicomponent gas composition was no more than 6%.

Shtang A. A., Dedov S. I., Gripas V. K. Integration of convolutional neural network into a chromatographic gas analyzer for continuous screening. MIAB. Mining Inf. Anal. Bull. 2025;(1-1):221-233. [In Russ]. DOI: 10.25018/0236_1493_2025_11_0_221.

A.A. Shtang¹, Cand. Sci. (Eng.), Assistant Professor, Assistant Professor, e-mail: shtang@corp.nstu.ru, ORCID ID: 0000-0001-9772-1784,
S.I. Dedov¹, Cand. Sci. (Eng.), Assistant Professor, e-mail: dedov@corp.nstu.ru, ORCID ID: 0000-0003-4750-3927,
V.K. Gripas¹, Graduate Student, e-mail: gripas.2017@stud.nstu.ru, ORCID ID: 0009-0002-3451-4176,
¹ Novosibirsk State Technical University, 630073, Novosibirsk, Russia.

S.I. Dedov, e-mail: dedov@corp.nstu.ru.

1. Koturbash T., Bicz A., Bicz W. New instrument for measuring velocity of sound and quantitative characterization of binary gas mixtures composition. Measurement Automation Monitoring. 2016, vol. 65, no. 8, pp. 254—258.

2. Agarkov A. Analysis of the emergency at the mining enterprises and evaluation of the method of remote selection of samples of mine air when conducting mine-rescue work. Fire and technosphere safety: Problems and ways of improvement. 2019, no. 2(3), pp. 10—20.

3. Agarkov A. V. On accidents in coal mines and the use of a remote control system for the composition of the mine environment during mine rescue operations. Innovatsionnye tekhnologii razrabotki mestorozhdeniy poleznykh iskopaemykh. Sbornik nauchnykh trudov kafedry razrabotki mestorozhdeniy poleznykh iskopaemykh (No. 5) [Innovative technologies for the development of mineral deposits: collection of scientific papers of the Department of Mineral deposits Development (No. 5)], Donetsk, 2019, pp. 294—313. [In Russ], available at: http://ed.donntu.org/books/19/cd8966.pdf. (accessed 15.02.2024).

4. Matyushin D. D., Buryak A. K. Gas chromatographic retention index prediction using multimodal machine learning. IEEE Access. 2020, vol. 8, pp. 2169—3536. DOI: 10.1109/ACCESS.2020.3045047.

5. Noritaka U., Tatsuya T. Gas detection via machine learning. World Academy of Science, Engineering and Technology. International Journal of Computer and Information Engineering. 2020, vol. 2, pp. 61—65. DOI: 10.5281/zenodo.1075908.

6. Zixuan G., Yingjie F. GCMSFormer: A fully automatic method for the resolution of overlapping peaks in gas chromatography — Mass spectrometry. Analytical Chemistry. 2024, vol. 96, no. 15, pp. 5878—5886. DOI: 10.1021/acs.analchem.3c05772.

7. Matyushin D. D., Yu A. S. A deep convolutional neural network for the estimation of gas chromatographic retention indices. Journal of Chromatography A. 2019, vol. 1607, article 460395. DOI: 10.1016/j.chroma.2019.460395.

8. Leiyu C., Shaobo L. Review of image classification algorithms based on convolutional neural networks. Remote Sensing. 2021, vol. 13, no. 22, article 4712. DOI: 10.3390/rs13224712.

9. Bengio Y., Glorot X. Understanding the difficulty of training deep feed forward neural networks. Proceedings of the International Conference on Artificial Intelligence and Statistics. 2010, vol. 8, pp. 249—256.

10. Ayub A. K., Laghari A. Machine learning in computer vision: A review. EAI Endorsed Transactions on Scalable Information Systems. 2021, vol. 8, pp. 1—11. DOI: 10.4108/eai.21-4-2021.169418.

11. Yuanpu G., Dan Z. SemiAMR: semi-supervised automatic modulation recognition with corrected pseudo-label and consistency regularization. IEEE Transactions on Cognitive Communications and Networking. 2024, vol. 10, pp. 107—121. DOI:10.1155/2023/2683780.

12. Zewen L., Fan L. A survey of convolutional neural networks: Analysis, applications, and prospects. IEEE Transactions on Neural Networks and Learning Systems. 2022, vol. 13, pp. 6999—7019. DOI: 10.1109/TNNLS.2021.3084827.

13. Anamika D., Verma G. K. Convolutional neural network: a review of models, methodologies and applications to object detection. Progress in Artificial Intelligence. 2019, vol. 9, pp. 25—112. DOI: 10.1007/s13748-019-00203-0.

14. Grace W. Convolutional neural networks as a model of the visual system: Past, present, and future. Journal of Cognitive Neuroscience. 2021, vol. 33, pp. 2017—2031. DOI: 10.1162/jocn_a_01544.

15. Yong Y. Xiaosheng S. A review of recurrent neural networks: LSTM cells and network architectures. Neural Computation. 2019, vol. 31, pp. 1235—1270. DOI: 10.1162/neco_a_01199.

16. Hansika H., Christoph B., Kasun B. Recurrent neural networks for time series forecasting: current status and future directions. International Journal of Forecasting. 2021, vol. 37, pp. 388—427. DOI: 10.1016/j.ijforecast.2020.06.008.

17. Szegedy C., Liu W. Going deeper with convolutions. IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2015, vol. 3, pp. 1—10. DOI: 10.1109/CVPR.2015.7298594.

18. Simonyan K., Zisserman A. Very deep convolutional networks for large-scale image recognition. International Conference on Learning Representations. 2014, pp. 1—12. DOI: 10.48550/arXiv.1409.1556.

19. Glorot X., Bengio Y. Understanding the difficulty of training deep feedforward neural networks. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics (PMLR). 2010, vol. 9, pp. 249—256.

20. He K., Zhang X., Ren S., Sun J. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification. IEEE International Conference on Computer Vision (ICCV). 2015, pp. 1026—34. DOI: 10.48550/arXiv.1502.01852.

21. Ioffe S., Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. International Conference on Machine Learning. 2015, pp. 1—11. DOI: 10.48550/ arXiv.1502.03167.

22. Srivastava R. K., Greff K., Schmidhuber J. Highway networks. available at: https://arxiv.org/abs/ 1502.03167 (accessed 15.02.2024).

23. Kaiming H., Xiangyu Z., Shaoqing R., Jian S. Deep residual learning for image recognition. IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016, pp. 1—12. DOI: 10.1109/ cvpr.2016.90.

24. Ramachandran P., Zoph B., Le Q. V. Searching for Activation Functions. available at: https:// arxiv.org/abs/1710.05941 (accessed 15.02.2024).

25. Federal'nye normy i pravila v oblasti promyshlennoy bezopasnosti «Pravila bezopasnosti v ugol'nykh shakhtakh». Seriya 05. Vyp. 40 [Federal standards and regulations in the field of industrial safety «Safety rules in coal mines». Episode 05. Issue 40], Moscow, 2014, 200 p. [In Russ].

26. Polozhenie ob aerogazovom kontrole v ugol'nykh shakhtakh. Seriya 05. Vyp. 23 [The position of ob aerogazovom kontrole v ugol'nykh shakhtakh. Seriya 05. Vyp. 23 Regulation on Aerogas control in coal mines. Episode 05. Issue 23], Moscow, 2012, 110 p. [In Russ].

27. Analiz avariy i gornospasatel'nykh rabot na predpriyatiyakh, obsluzhivaemykh podrazdeleniyami GVGSS za 2018 god [Analysis of accidents and mining rescue operations at enterprises serviced by GVGSS units in 2018], Donetsk, 2019, 76 p. [In Russ].