Neural network learning to determine the maximum filling coefficient of the model energy condensed system polymer base with the powder components

Authors: Sharov G.S.
Published in issue: #8(85)/2023
DOI: 10.18698/2541-8009-2023-8-927
Category: Aviation and Rocket-Space Engineering | Chapter: Aircrafts Development, Design and Manufacture
Keywords: neural networks, perceptron, artificial intelligence, Python, Keras, granulometric composition optimization, limiting volumetric filling with dispersed particles, model energy condensed system
Published: 30.08.2023

The granulometric composition of ammonium perchlorate was optimized by calculating the limiting degree of volumetric filling using a neural network. The code for the neural network was written using the Python 3.0 programming language with the imported Keras library. Optimal parameters of the neural network were selected. The EarlyStopping method built into the Keras library was applied in order to eliminate the problem of the neural network relearning. Results of the neural network operation were analyzed. The resulting dependence of the coefficient of the polymer base maximum filling was visualized using the triangular Gibbs diagrams. Relevance of introducing the neural networks in optimization of the modern energy system granulometric composition was established.

References

[1] Ermilov A.S. Teoreticheskie osnovy protsessov polucheniya i pererabotki polimernykh materialov [Theoretical foundations of the processes of obtaining and processing polymeric materials]. Perm, PSTU Publ., 2009, 159 p. (In Russ.).

[2] Minsky M., Papert S. Perceptrons. The MIT Press, 1969, 258 p. (Russ. ed.: Minskiy M., Peypert S. Perseptrony. Moscow, Mir Publ., 1971, 261 p.).

[3] Latkina I.I. An overview of TensorFlow capabilities for solving machine learning problems. Innovatika, 2015, no. 1, pp. 33–42. (In Russ.).

[4] Alikin V.N., Vakhrushev A.V., Golubchikov V.B. et. al. Tverdye topliva reaktivnykh dvigateley [Solid jet fuels]. Moscow, Mashinostroenie Publ., 2011, vol. 4, 380 p. (In Russ.).

[5] Nurullaev E.M., Ermilov A.S., Gurov D.S. Optimization of grading of polymer composites solid dispersed fillers. Vestnik Permskogo natsional’nogo issledovatel’skogo politekhnicheskogo universiteta. Aerokosmicheskaya tekhnika, 2013, no. 34, pp. 108–123. (In Russ.).

[6] MakGrat M. Programmirovanie na Python dlya nachinayushchikh [Python programming for beginners]. Moscow, Eksmo Publ., 2015, 192 p. (In Russ.).

[7] Tolmachev S.G. Sistemy iskusstvennogo intellekta. Neyrosetevye modeli [Artificial intelligence systems. Neural network models]. Saint Petersburg, BSTU Publ., 2011. (In Russ.).

[8] Metriki v zadachakh mashinnogo obucheniya [Metrics in machine learning problems]. URL: https://habr.com/ru/company/ods/blog/328372/ (accessed February 15, 2023).

[9] Keras: The Python Deep Learning library. URL: https://keras.io/ (accessed April 22, 2023).

[10] Platonov A.V. Mashinnoe obuchenie [Machine learning]. Moscow, Yurayt Publ., 2023, 85 p. URL: https://urait.ru/bcode/520544 (accessed April 22, 2023).