LSTM networks for anaerobic digester control
/ 0
- Details
- Category: Information Technologies, Systems Analysis and Administration
- Last Updated on 10 November 2019
- Published on 26 October 2019
- Hits: 2717
Authors:
V.Yesilevskyi, Cand. Sc. (Tech.), Assoc. Prof., orcid.org/0000-0002-5935-1505, National University of Radioelectronics, Kharkiv, Ukraine, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
S.Dyadun, Cand. Sc. (Tech.), Assoc. Prof., orcid.org/0000-0002-1910-8594, Beketov National University of Urban Economy, Kharkiv, Ukraine; Karazin National University, Kharkiv, Ukraine, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
V.Kuznetsov, Cand. Sc. (Tech.), Prof., orcid.org/0000-0003-4638-5260, Starooskolsky Technological Institute A. A. Ugarova (branch) of “National Research Technological University “MISiS”, Stary Oskol, Belgorod Region, Russian Federation, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
Abstract:
Purpose. To study the possibility of use of artificial intelligence systems based on neural networks for the improvement of the efficiency of the biogas production in the anaerobic digester by optimization of the nonlinear control system.
Methodology. Mathematical modeling and computer simulation for building an adaptive control system based on neural LSTM network and reinforcement learning. The study of the learning convergence of the neural network based on a non-linear mathematical model of the technological process.
Findings. The performed computational experiment showed that the learning convergence could be achieved after 600 episodes in a model with two control channels and four measurement channels. It also proved the efficiency of the proposed adaptive control system for the program model of the technological process. The results confirmed the ultimate possibility to use this approach for biogas production control in the anaerobic digester in the real world settings.
Originality. It is shown that the task of controlling the biogas production in the anaerobic digester can be solved based on the reinforcement learning approach implemented for a continuous nonlinear dynamic system. The system can be implemented as two neural networks in the actor-critic architecture. To build the module of critic and the module of the controller of the adaptive control system, it was proposed to use the neural network architecture based on a special type of recursive neural network called LSTM neural networks, which were not previously used to control the production of biogas in the anaerobic digester.
Practical value. The proposed method for constructing a control system for the anaerobic digesters will lead to an increase in the efficiency of biogas production as an alternative renewable energy source. Expansion of the scope of use of anaerobic digesters will also have a significant impact on solving the problem of utilization of industrial and household human waste.
References.
1. Grando, R. L., de Souza Antune, A. M., Da Fonseca, F. V., Sánchez, A., Barrena, R., & Font, X. (2017). Technology overview of biogas production in anaerobic digestion plants: a European evaluation of research and development. Renewable and Sustainable Energy Reviews, 80, 44-53.
2. Malina, J. (2017). Design of Anaerobic Processes for Treatment of Industrial and Muncipal Waste, Volume VII. Routledge.
3. Rincón, A., Villa, J., Angulo, F., & Olivar, G. (2014). A Dynamic Analysis for an Anaerobic Digester: Stability and Bifurcation Branches. In Mathematical Problems in Engineering, 2014.
4. Jimenez, J., Latrille, E., Harmand, J., Robles, A., Ferrer, J., Gaida, D., … & Mendez-Acosta, H. (2015). Instrumentation and control of anaerobic digestion processes: a review and some research challenges. Reviews in Environmental Science and Bio/Technology, 14(4), 615-648.
5. Oppong, G., & Montague, G. (2013). Towards Model Predictive Control on Anaerobic Digestion Process. IFAC Proceedings Volumes, 46(32), 684-689. DOI: 10.3182/20131218-3-IN-2045.00063.
6. Lei Xue, Li, D., & Xi, Y. (2015). Nonlinear model predictive control of anaerobic digestion process based on reduced ADM1. In 10 th Asian Control Conference (ASCC), 2015. 31 May–3 June 2015. DOI: 10.1109/ASCC.2015.7244539.
7. Mnih, V., Kavukcuoglu, K., Silver, D., Andrei A. Rusu, Veness, J., Marc G. Bellemare, … & Hassabis, D. (2015). Human-level control through deep reinforcement learning. Nature, 518(7540), 529-533.
8. Timothy P. Lillicrap, Jonathan J. Hunt, Pritzel, A., Heess, N., Erez, T., Tassa, Y., … & Daan Wierstra (2015). Continuous control with deep reinforcement learning. Retrieved from https://arxiv.org/abs/1509.02971.
9. Spielberg, S., Gopaluni, R. B., & Loewen, P. D. (2017). Deep reinforcement learning approaches for process control. In 6 th International Symposium on Advanced Control of Industrial Processes (AdCONIP) (pp. 201-206). DOI: 10.1109/ADCONIP.2017.7983780. 2017.
10. Russell, S. J., & Norvig, P. (2016). Artificial intelligence: a modern approach. Malaysia: Pearson Education Limited.
11. David, S., Lever, G., Heess, N., Degris, T., Wierstra, D., & Riedmiller, M. (2014). Deterministic policy gradient algorithms. In Proceedings of the 31st International Conference on Machine Learning, Beijing, China, 2014. JMLR: W&CP, volume 32.
12. Ghavamzadeh, M., Mannor, S., Pineau, J., & Tamar, A. (2015). Bayesian reinforcement learning: A survey. Foundations and Trends® in Machine Learning, 8(5-6), 359-483.
13. Van Hasselt, H., Guez, A., & Silver, D. (2016). Deep Reinforcement Learning with Double Q-Learning. AAAI, 2, p. 5.
14. Mnih, V., Badia, A.P., Mirza, M., Graves, A., Lillicrap, T., Harley, T., ... & Kavukcuoglu, K. (2016, June). Asynchronous methods for deep reinforcement learning. In International conference on machine learning (pp. 1928-1937).
15. Lillicrap, T. P., Hunt, J. J., Pritzel, A., Heess, N., Erez, T., Tassa, Y., ... & Wierstra, D. (2015). Continuous control with deep reinforcement learning. arXiv preprint arXiv:1509.02971.
Archive