Setembro 2023 vol. 10 num. 1 - XXX Simpósio Internacional de Engenharia Automotiva

Trabalho completo - Open Access.

Idioma principal | Segundo idioma

Uso de digital twins treinados em machine learning aplicados em teste veiculares de emissões.

Use of machine learning based digital twins on vehicle emissions tests.

TOMANIK, Eduardo ; TOMANIK, Victor ; JIMENEZ-REYES, Antonio J. ; TORMOS, Bernardo ;

Trabalho completo:

Testes veiculares de emissões geram uma grande quantidade de dados, mas os resultados são em geral avaliados apenas quanto ao seu valor acumulado segundo os limites de homologação. Neste trabalho, dois modelos de ?machine learning? (Random Forest e Artificial Neural Network) foram aplicados em testes transientes de emissões. Após serem treinados no ciclo FTP-75 partida a frio, os modelos foram capazes de prever com boa acurácia tanto o valor acumulado quanto os valores instantâneos de consumo de combustível e as emissões de CO2 em outros ciclos, inclusive o NEDC para os veículos em questão. Um primeiro, ainda preliminar, exercício num veículo hibrido mostrou o potencial do modelo, mas mais estudos são ainda necessários.

Trabalho completo:

Emission tests generate a huge amount of measurement data but are usually evaluated only regarding the accumulated value according to the emission homologation limits. In this work, a Random Forest machine leaning code was used to create a vehicle digital twin able to predict an output of interest, e.g., instantaneous fuel consumption. The digital twin, after being trained on measurements from the FTP-75 cold start, was able to predict with good accuracy, not only the accumulated values but also the instantaneous values of fuel consumption and CO2 emissions. A first exercise on a hybrid vehicle showed some potential, but more work is needed to reproduce some of the parameters of interest.

Download (PDF)

Palavras-chave: ,

Palavras-chave: ,

DOI: 10.5151/simea2023-PAP90

Referências bibliográficas
  • [1] " Machine Learning - A Probabilistic Perspective by
  • [2] Kevin P. Murphy. The MIT press, 201 ISBN
  • [3] 9780262018029
  • [4] 2. Sarker, I.H. Machine Learning: Algorithms, Real-
  • [5] World Applications and Research Directions. SN
  • [6] Comput. Sci. 2021, 2, 160.
  • [7] 3. Ho, T. Random decision forests. In Proceedings of the
  • [8] 3rd International Conference on Document Analysis
  • [9] and Recognition, Montreal, QC, Canada, 14–16
  • [10] August 1995; pp. 278–282.
  • [11] 4. Hien, N.L.H.; Kor, A.L. Analysis and prediction
  • [12] model of fuel consumption and carbon dioxide
  • [13] emissions of light-duty vehicles. Appl. Sci. 2022.
  • [14] 5. Katreddi, S.; Thiruvengadam, A. Trip based modeling
  • [15] of fuel consumption in modern heavy-duty vehicles
  • [16] using artificial intelligence. Energies 2021.
  • [17] 6. Gong, J.; Shang, J.; Li, L.; Zhang, C.; He, J.; Ma, J. A
  • [18] Comparative Study on Fuel Consumption Prediction
  • [19] Methods of Heavy-Duty Diesel Trucks Considering 21
  • [20] Influencing Factors. Energies 2021.
  • [21] 7. He, Y.; Rutland, C. Application of artificial neural
  • [22] networks in engine modelling. Int. J. Engine Res.
  • [23] 2005.
  • [24] 8. Ziółkowski, J.; Oszczypała, M.; Małachowski, J.;
  • [25] Szkutnik-Rogoż, J. Use of Artificial Neural Networks
  • [26] to Predict Fuel Consumption on the Basis of Technical
  • [27] Parameters of Vehicles. Energies 2021.
  • [28] 9. Perrotta, F.; Parry, T.; Neves, L. Application of
  • [29] machine learning for fuel consumption modelling of
  • [30] trucks. In Proceedings of the 2017 IEEE International
  • [31] Conference on Big Data, Boston, MA, USA, 11–14
  • [32] December 2017; pp. 3810–3815.
  • [33] 10. Tomanik, E., Tomanik, V., Morais, P. Use of
  • [34] tribological and AI models on vehicle emission tests to
  • [35] predict fuel savings through lower oil viscosity, SAE
  • [36] Technical Paper (2021-36-0038) 2021.
  • [37] 11. Tomanik, E.; Jimenez-Reyes, A.J.; Tomanik, V.;
  • [38] Tormos, B. Machine-Learning-Based Digital Twins for
  • [39] Transient Vehicle Cycles and Their Potential for
  • [40] Predicting Fuel Consumption. Vehicles 2023.
  • [41] 12. Downloadable Dynamometer Database—Argonne
  • [42] National Laboratory. Available online:
  • [43] https://www.anl.gov/es/downloadable-dynamometerdatabase
  • [44] (accessed on 22 March 2023).
  • [45] 13. Overview of the ANL Chassis Dynamometer Test
  • [46] Facilities and Methodology. Available online:
  • [47] https://anl.app.box.com/s/5tlld40tjhhhtoj2tg0n4y3fkw
  • [48] dbs4m3 (accessed on 22 March 2023).
  • [49] 14. Krysmon, S.; Claßen, J.; Pischinger, S.; Trendafilov,
  • [50] G.; Düzgün, M.; Dorscheidt, F. RDE Calibration—
  • [51] Evaluating Fundamentals of Clustering Approaches to
  • [52] Support the Calibration Process. Vehicles 2023
  • [53] 15. https://www.anl.gov/taps/d3-2019-acura-mdx-sporthybrid
  • [54] (accessed on 22 March 2023)."
Como citar:

TOMANIK, Eduardo; TOMANIK, Victor; JIMENEZ-REYES, Antonio J.; TORMOS, Bernardo; "Uso de digital twins treinados em machine learning aplicados em teste veiculares de emissões.", p. 447-456 . In: Anais do XXX Simpósio Internacional de Engenharia Automotiva . São Paulo: Blucher, 2023.
ISSN 2357-7592, DOI 10.5151/simea2023-PAP90

últimos 30 dias | último ano | desde a publicação


downloads


visualizações


indexações