DDPG-Based Intelligent Control Strategy for Industrial Dual-Tank System
Main Article Content
Abstract
The performance and reliability of industrial control systems are impacted by external influences such as fluctuating operating environments and disruptive interferences, presenting notable challenges. Consequently, the exploration and adoption of smart control algorithms with capabilities for autonomous learning, self-tuning, and self-adjustment have emerged as a vital and significant research focus. This research investigates an intelligent control technique that employs the Deep Deterministic Policy Gradient (DDPG) algorithm for process control systems with a dual-tank system selected as the case study, in which the flow rate is manipulated to regulate the system temperature. The performance of the observer component in the critic network is improved by integrating a densely connected layer, which enhances its capacity to represent and handle data, thereby improving the identification of essential characteristics for water-level management. Additionally, the neural network’s node settings are fine-tuned, and the ReLU activation function is implemented to support ongoing monitoring and adaptation to the external tank environment while preventing gradient vanishing. The research firstly trains DDPG with various initial conditions and then validates the performance for the temperature control problem by simulation. Additionally, the performance of DDPG is compared to the conventional Proportional-Integral-Derivative (PID) controller in terms of rise time, settling time, overshoot, and steady-state error.
Keywords
DDPG, dual-tank system, intelligent control technique, reinforcement learning.
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
References
[2] E. Pivarciova and M. E. Qazizada, Applications of process control system with (SCADA) and PID controller, 12th International Conference ELEKTRO 2018, 2018 ELEKTRO Conference Proceedings, pp. 1–6, 2018. ttps://doi.org/10.1109/ELEKTRO.2018.8398312
[3] Y. Zhao, Y. Zhang, Y. Gong, and X. Zou, Design of double tank liquid level control system based on fuzzy PID, Journal of Physics: Conference Series, vol. 2650, no. 1, 2023. https://doi.org/10.1088/1742-6596/2650/1/012041
[4] H. Gouta, S. H. Said, and F. M’Sahli, Observerbased predictive liquid level controller for a double tank process, Proceedings of 2015 7th International Conference on Modelling, Identification and Control, ICMIC 2015, pp. 1–6, 2016. https://doi.org/10.1109/ICMIC.2015.7409371
[5] H. Gouta, S. H. Said, and F. M’Sahli, Predictive and backstepping control of double tank process: A comparative study, IETE Technical Review (Institution of Electronics and Telecommunication Engineers, India), vol. 33, no. 2, pp. 137–147, 2016. https://doi.org/10.1080/02564602.2015.1052580
[6] C. Panjapornpon, P. Chinchalongporn, S. Bardeeniz, R. Makkayatorn, and W. Wongpunnawat, Reinforcement learning control with deep deterministic policy gradient algorithm for multivariable pH process, Processes, vol. 10, no. 12, 2022. https://doi.org/10.3390/pr10122514
[7] H. Shi, L. Zhang, D. Pan, and G. Wang, Deep reinforcement learning-based process control in biodiesel production, Processes, vol. 12, no. 12, 2024. https://doi.org/10.3390/pr12122885
[8] A. Kadu and A. Khandekar, Deep reinforcement learning-based approach for control of two input–two output process control system, International Journal on Smart Sensing and Intelligent Systems, vol. 18, no. 1, 2025. https://doi.org/10.2478/ijssis-2025-0029
[9] N. Dey, R. Mandal, and M. M. Subashini, Design and implementation of a water level controller using fuzzy logic, International Journal of Engineering and Technology (IJET), vol. 5, no. 3, pp. 2277–2285, 2013.
[10] D. E.Seborg, T. F.Edgar, and D. A.Mellichamp, Process Dynamics and Control, Second ed., Wiley, 2004.
[11] D. Silver, Deterministic policy gradient algorithms, 31st International Conference on Machine Learning, ICML 2014, vol. 1, pp. 605–619, 2014.
[12] D. Silver, G. Lever, N. Heess, T. Degris, D.Wierstra, and M. Riedmiller, Deterministic policy gradient algorithms, in Proceedings of the 31st International Conference on Machine Learning, Beijing, China, 22–24 June 2014, pp. 387–395.