Journals / Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi / 2023 / Cilt: 12 - Sayı: 1

Reinforcement learning based position control of an electro-hydraulic system

Elektrohidrolik bir sistemin pekiştirmeli öğrenme tabanlı denetleyici ile konum denetiminin gerçekleştirilmesi

Pages
280–288
Publication Date
2023-01-15
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The reinforcement learning method reached 25.51% more successful results than the PID controller in terms of tracking the position reference signal with different characteristics.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Electrohydraulic systems have become an inseparable part of the industry due to the advantages they provide.
📊
Key Findings & Data:On the other hand, the nonlinear characteristics of the hydraulic systems and the parametric uncertainties make their control troublesome.
💡
Conclusion & Impact:The reinforcement learning method reached 25.51% more successful results than the PID controller in terms of tracking the position reference signal with different characteristics.
🔬Design:Research Article
📍Scope:Comprehensive Sample
⚙️Variables:Elektrohidrolik sistem · Modelleme · Konum denetimi

Abstract

Electrohydraulic systems have become an inseparable part of the industry due to the advantages they provide. On the other hand, the nonlinear characteristics of the hydraulic systems and the parametric uncertainties make their control troublesome. In this study, the position of an asymmetrical hydraulic piston driven by a proportional valve was controlled by reinforcement learning. Although there are many reinforcement learning algorithms, the deep deterministic policy gradient, which stands out with its effectiveness in continuous space, has been preferred. The hyperparameters were found by preliminary studies and the training of the controller for the designed position reference signal was carried out numerically. The obtained results are compared with the PID controller. The reinforcement learning method reached 25.51% more successful results than the PID controller in terms of tracking the position reference signal with different characteristics.

Özet

Elektrohidrolik sistemler sağladıkları avantajlar sebebiyle endüstrinin vazgeçilmezi olmuştur. Buna karşın hidrolik sistemlerin doğrusal olmayan karakteristik özellikleri ve çok sayıda parametre belirsizliği barındırması bu sistemlerin denetimini zorlaştıran etmenler olarak öne çıkmaktadır. Bu çalışmada ise oransal valf ile sürülen asimetrik bir hidrolik pistonun konumu pekiştirmeli öğrenme ile denetlenmiştir. Pek çok pekiştirmeli öğrenme algoritması olmasına rağmen sürekli uzayda etkinliği ile öne çıkan derin deterministik politika gradyanı yöntemi tercih edilmiştir. İlgili hiper parametreler öncül çalışmalarla belirlenerek çoklu konum referans sinyali için denetleyicinin eğitimi benzetim ortamında gerçekleştirilmiştir. Elde edilen sonuçları kıyaslamak için aynı çalışma PID denetleyici ile de gerçekleştirilmiştir. Çalışmada kullanılan pekiştirmeli öğrenme yöntemi farklı karakteristiklere sahip konum referans sinyalinin takibinde PID denetleyiciden daha %25.51 oranında daha başarılı sonuçlar ortaya koymuştur.

🔖

Cite This Article

COŞKUN, M. Y., İTİK, M. (2023). Reinforcement learning based position control of an electro-hydraulic system. Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi, Cilt: 12 - Sayı: 1, 280–288. https://ajindex.com/articles/877070

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (8)

  1. J. Činkelj, R. Kamnik, P. Čepon, M. Mihelj and M. Munih, Closed-loop control of hydraulic telescopic handler. Automation in Construction, 19, 954–963, 2010. https://doi.org/10.1016/j.autcon.2010.07.012
  2. M. Borghi, B. Zardin, F. Pintore and F. Belluzzi, Energy savings in the hydraulic circuit of agricultural tractors. Energy Procedia, Elsevier B.V. 45, 352–361, 2014. https://doi.org/10.1016/j.egypro.2014.01.038
  3. T. Boaventura, J. Buchli, C. Semini and D.G. Caldwell, Model-Based hydraulic impedance control for dynamic robots. IEEE Transactions on Robotics, IEEE. 31, 1324–1336, 2015. https://doi.org/10.1109/ TRO.2015.2482061
  4. A. Bayrak, E. Tatlicioglu and E. Zergeroglu, Backstepping control of electro-hydraulic arm. 2018 6th International Conference on Control Engineering and Information Technology, CEIT 2018, IEEE. pp. 1-4, 2018. https://doi.org/10.1109/CEIT.2018.8751833
  5. D. Janosevic, R. Mitrev, B. Andjelkovic, and P. Petrov, Quantitative measures for assessment of the hydraulic excavator digging efficiency. Journal of Zhejiang University: Science A, 13, 926–942, 2012. https://doi.org/10.1631/jzus.A1100318 M. Heikkilä and M. Linjama, Displacement control of a mobile crane using a digital hydraulic power management system. Mechatronics, 23, 452–461, 2013. https://doi.org/10.1016/j.mechatronics.2013.03. 009 M. Jelali and A. Kroll, Hydraulic Servo-systems. Springer London, London. 2003. https://doi.org/ 10.1007/978-1-4471-0099-7
  6. M. Fallahi, M. Zareinejad, K. Baghestan, A. Tivay, S.M. Rezaei and A. Abdullah, Precise position control of an electro-hydraulic servo system via robust linear approximation. ISA Transactions, Elsevier Ltd. 80, 503–512, 2018. https://doi.org/10.1016/j.isatra. 2018.06.002 C. Onat, M. Daskin and A. Turan, Gain scheduling linear model of an electro-hydraulic actuator. An International Journal of Optimization and Control: Theories & Applications (IJOCTA), 7, 301–306, 2017. https://doi.org/10.11121/ijocta.01.2017.00487
  7. E. Kolsi-Gdoura, M. Feki and N. Derbel, Sliding mode-based robust position control of an electrohydraulic system. 10th International Multi-Conferences on Systems, Signals & Devices 2013 (SSD13), IEEE. pp. 1–5 2013. https://doi.org/10.1109/SSD.2013.6564127
  8. E. Kolsi Gdoura, M. Feki and N. Derbel, Sliding mode control of a hydraulic servo system position using adaptive sliding surface and adaptive gain. International Journal of Modelling, Identification and Control, 23, 248–259, 2015. https://doi.org/10.1504/ IJMIC.2015.069946 B. Sh