Voltage Regulation of DC-DC Buck Converters Feeding CPLs via Deep Reinforcement Learning

Cui, Chenggang; Yan, Nan; Huangfu, Baixiang; Yang, Tianxiao; Zhang, Chuanlin

doi:10.1109/tcsii.2021.3107535

Cited by 41 publications

(21 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The detailed simulation parameters of the DC-DC buck converter is consistent with the real-life system, as shown in Table I. The parameters of the hyper-parameters for the design of the DQN controller are depicted in TABLE II [25].…”

Section: Sim-to-real Proceduresmentioning

confidence: 90%

“…The control objective of the DC microgrid is to regulate the voltage of the load bus at a nominal value by the DC-DC buck converter. In this paper, based on a previous DRL control approach in [25], we are aiming to propose a simto-real transferring procedure via a novel DRM strategy. In this regard, both the transient-time and steady-state control performance could be guaranteed.…”

Section: A Problem Formulationmentioning

confidence: 99%

“…In a previous work [25], the authors have proposed a DQN algorithm based on superior learning to adjust the duty ratio for the DC-DC buck converter. The control diagram of the proposed DRL controller is presented in Fig.…”

Section: A Drl Controller Designmentioning

confidence: 99%

See 2 more Smart Citations

Transferring Reinforcement Learning for DC-DC Buck Converter Control via Duty Ratio Mapping: From Simulation to Implementation

Cui¹,

Yang²,

Dai³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Reinforcement learning (RL) control approach with application into power electronics systems has become an emerging topic whilst the sim-to-real issue remains a challenging problem as very few results can be referred to in the literature. Indeed, due to the inevitable mismatch between simulation models and real-life systems, offline trained RL control strategies may sustain unexpected hurdles in practical implementation during transferring procedure. As the main contribution of this paper, a transferring methodology via a delicately designed duty ratio mapping (DRM) is proposed for a DC-DC buck converter. Then, a detailed sim-to-real process is presented to enable the implementation of a model-free deep reinforcement learning (DRL) controller. The feasibility and effectiveness of the proposed methodology are demonstrated by comparative experimental studies. Index Terms-Deep reinforcement learning, DC-DC buck converter, practical implementation, duty ratio mapping NOMENCLATURE β 1 , β 2 Reward coefficients β 3 Penalty coefficient ǫ 1 , ǫ 2 Subgoals of the expected error de(t) dt Time derivative of the tracking error dvo(t) dt

show abstract

Section: Sim-to-real Proceduresmentioning

confidence: 90%

Section: A Problem Formulationmentioning

confidence: 99%

See 1 more Smart Citation

Transferring Reinforcement Learning for DC-DC Buck Converter Control via Duty Ratio Mapping: From Simulation to Implementation

Cui¹,

Yang²,

Dai³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Hence, complicated control schemes are needed to cope with non‐modeled system/actuator dynamics and non‐measurable disturbances; for example, fuzzy logic control, 8,9,10 feedback and feed‐forward proportional integral derivative (PID) control, 11,12,13 and adaptive control 14,15,16 . Additional control schemes include active disturbance rejection control, 17,18,19 internal model control 20,21 and intelligent control 22,23 . However, the complexity of these solutions significantly increases the computational burden and makes them unsuitable for practical applications.…”

Section: Introductionmentioning

confidence: 99%

“…14,15,16 Additional control schemes include active disturbance rejection control, 17,18,19 internal model control 20,21 and intelligent control. 22,23 However, the complexity of these solutions significantly increases the computational burden and makes them unsuitable for practical applications. Sliding mode controllers are also able to provide the desired voltage regulation, but these controllers rely on extra current and voltage sensors 14,24 and the presence of high-frequency noise generated by fast switching components further complicates the task.…”

Section: Introductionmentioning

confidence: 99%

A robust energy processing method for fuel cell systems supplying a time‐varying uncertain load

Ramírez

2022

Intl J of Energy Research

View full text Add to dashboard Cite

Summary Robustness and performance are key properties that can contribute to the widespread dissemination of fuel cell systems (FCSs) as power sources, and a precise output voltage can serve as a reliable criterion to evaluate these properties. In this paper, we propose to utilize intentional time delays as part of controllers to achieve proper output voltage regulation in an FCS. In the past few years, there has been a growing interest toward designing controllers with intentional delays to render improved responses and better noise attenuation capabilities in various classes of systems and this investigation extends the use of delay‐based controllers to the application field of FCSs. A natural progression of this research, which to our knowledge has not been addressed, pertains to how such controllers perform in the presence of uncertainties, and whether or not these controllers can accommodate these uncertainties. In this research work, we present a solution to this open problem on a class of proportional integral retarded (PIR) controllers when controlling an FCS supplying a time‐varying uncertain load. Through stability theory, the safe operation of the PIR controlled FCS is systematically guaranteed in spite of uncertain loads and input voltages. Specifically, based on Lyapunov methods and L2‐gain analysis, we derive simple stability conditions to achieve robustness and H∞ performance. We show that utilizing only output voltage measurements the PIR controller can maintain the predictive features of a pure differentiator, but without its inherent noise amplification problems. Hence, low‐pass filtering of signals or reconstruction of the states can be obviated. Compared with published control schemes for FCSs, the PIR controller is structurally simpler and practically realizable, and it is able to maintain robustness against uncertain loads and H∞ performance under exogenous signals without relying on disturbance estimators. Numerical simulations confirm the superiority of the PIR controller over benchmark proportional integral and proportional integral derivative controllers in terms of robustness, performance and noise attenuation capabilities.

show abstract

Deep reinforcement learning-based proportional–integral control for dual-active-bridge converter

You

Yang

Chu

et al. 2023

Neural Comput & Applic

View full text Add to dashboard Cite

Voltage Regulation of DC-DC Buck Converters Feeding CPLs via Deep Reinforcement Learning

Cited by 41 publications

References 20 publications

Transferring Reinforcement Learning for DC-DC Buck Converter Control via Duty Ratio Mapping: From Simulation to Implementation

Transferring Reinforcement Learning for DC-DC Buck Converter Control via Duty Ratio Mapping: From Simulation to Implementation

A robust energy processing method for fuel cell systems supplying a time‐varying uncertain load

Deep reinforcement learning-based proportional–integral control for dual-active-bridge converter

Contact Info

Product

Resources

About