Adaptive neural control of PEMFC system based on data-driven and reinforcement learning approaches

Lin-Kwong-Chon, Christophe; Damour, Cédric; Benne, Michel; Kadjo, Jean-Jacques Amangoua; Grondin-Pérez, Brigitte

doi:10.1016/j.conengprac.2021.105022

Cited by 15 publications

(5 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several scholars have proposed diverse methods to enhance the temperature control of fuel cells. Traditional control algorithms comprise proportional-integral control [7], feedback control [8], piecewise predictive negative feedback control [9], and adaptive linear quadratic regulator feedback control [10]. Nevertheless, owing to the intrinsic nonlinearity of PEMFC, these control algorithms exhibit limitations.…”

Section: Introductionmentioning

confidence: 99%

Temperature Control of Fuel Cell Based on PEI-DDPG

Lu,

Yan

2024

Energies

View full text Add to dashboard Cite

Proton exchange membrane fuel cells (PEMFCs) constitute nonlinear systems that are challenging to model accurately. Therefore, a controller with robustness and adaptability is imperative for temperature control within the PEMFC stack. This paper introduces a data-driven controller utilizing deep reinforcement learning for stack temperature control. Given the PEMFC system’s characteristics, such as nonlinearity, uncertainty, and environmental conditions, we propose a novel deep reinforcement learning algorithm—the deep deterministic policy gradient with priority experience playback and importance sampling method (PEI-DDPG). Algorithm design incorporates technologies such as priority experience playback, importance sampling, and optimized sample data storage structure, enhancing the controller’s performance. Simulation results demonstrate the proposed algorithm’s superior effectiveness in temperature control for PEMFC, leveraging the PEI-DDPG algorithm’s high adaptability and robustness. The proposed algorithm’s effectiveness is additionally validated on the RT-LAB experimental platform. The proposed PEI-DDPG algorithm reduces the average adjustment time by 8.3%, 17.13%, and 24.56% and overshoots by 2.12 times, 4.16 times, and 4.32 times compared to the TD3, GA-PID, and PID algorithms, respectively.

show abstract

Section: Introductionmentioning

confidence: 99%

Temperature Control of Fuel Cell Based on PEI-DDPG

Lu,

Yan

2024

Energies

View full text Add to dashboard Cite

show abstract

“…Wang et al [ 21 ] applied a dynamic surface three‐step method in fault‐tolerant control, a method capable of controlling cathode pressure stabilization in the event of an air supply system fault. Lin‐Kwong‐Chon et al [ 22 ] proposed a data‐driven neural controller to ensure system stability in the event of water flooding, membrane drying, and actuator faults of the fuel cell system. Carlos et al [ 23 ] proposed a data‐driven method based on fault diagnosis and fault‐tolerant control, setting multiple controllers to switch when different faults are discerned, and proved the effectiveness of the method.…”

Section: Introductionmentioning

confidence: 99%

Fault Diagnosis and Fault‐Tolerant Control Method for a Multistack Fuel Cell Thermal Management Subsystem

et al. 2023

View full text Add to dashboard Cite

Multistack fuel cell systems (MFCS) are promising developments of fuel cell technology. For MFCS, faults affect durability and stability, and appropriate fault diagnosis methods and control strategies need to be proposed for different multistack structures. Herein, the integrated MFCS thermal management subsystem (TMS), which is based on the optimized split‐stack approach, is first studied, and its possible sensor and actuator faults are analyzed. The sensor fault signal is corrected based on the unscented Kalman filter (UKF), the controller design under the hardware in the loop platform is carried out, and an active fault‐tolerant control (FTC) strategy based on sliding mode control is deployed for the TMS to enable it to recover its performance quickly and effectively in case of faults. The results show that the estimated values are closer to the true values after using the UKF on the sensor signals, avoiding the impact of sensor faults on the system performance. Under New European Driving Cycle dynamic conditions, the FTC controller also exhibits better control compared to the conventional proportional–integral controller, with smaller maximum overshoot and shorter system recovery and stabilization time in the event of a TMS fault, mitigating the impact of system fault on performance.

show abstract

“…Deep reinforcement learning combines the powerful perceptual capability of deep learning with the excellent decision-making capability of reinforcement learning, which can solve many complex and coupled environmental problems. In the field of PEMFC research, Lin et al [21] proposed a data-driven neural controller with an automatic adaptive system based on the health state to study faults such as PEMFC channel overflow and membrane desiccation. Khadhraoui et al [22] developed an energy management model for PEMFC vehicles and used reinforcement learning algorithms to optimize the efficiency of PEMFC vehicles in real time under operating conditions to significantly reduce hydrogen consumption per hundred kilometers of the vehicle while optimizing efficiency.…”

Section: Introductionmentioning

confidence: 99%

Real-Time Control of Gas Supply System for a PEMFC Cold-Start Based on the MADDPG Algorithm

2023

View full text Add to dashboard Cite

During the cold-start process of a PEMFC, the supply of air and hydrogen in the gas supply system has a great influence on the cold-start performance. The cold-start of a PEMFC is a complex nonlinear coupling process, and the traditional control strategy is not sensitive to the real-time characteristics of the system. Inspired by the strong perception and decision-making abilities of deep reinforcement learning, this paper proposes a cold-start control strategy for a gas supply system based on the MADDPG algorithm, and designs an air supply controller and a hydrogen supply controller based on this algorithm. The proposed strategy can optimize the control parameters of the gas supply system in real time according to the temperature rise rate of the stack during the cold-start process, the fluctuation of the OER, and the voltage output characteristics. After the strategy is trained offline according to the designed reward function, the detailed in-loop simulation experiment results are given and compared with the traditional control strategy for the gas supply system. From the results, it can be seen that the proposed MADDPG control strategy has a more effective coordination control effect.

show abstract

Adaptive neural control of PEMFC system based on data-driven and reinforcement learning approaches

Cited by 15 publications

References 47 publications

Temperature Control of Fuel Cell Based on PEI-DDPG

Temperature Control of Fuel Cell Based on PEI-DDPG

Fault Diagnosis and Fault‐Tolerant Control Method for a Multistack Fuel Cell Thermal Management Subsystem

Real-Time Control of Gas Supply System for a PEMFC Cold-Start Based on the MADDPG Algorithm

Contact Info

Product

Resources

About