Research on anti-disturbance adaptive control strategy of power electronic converters based on deep reinforcement learning

DC-DC boost converters are widely used in renewable-energy interfaces, energy-storage systems, electric vehicles, microgrids, and regulated DC supplies, but their voltage response can deteriorate under load changes, input-voltage fluctuations, and circuit-parameter uncertainty. This study develops an offline-trained state-dependent anti-disturbance controller based on twin delayed deep deterministic policy gradient (TD3) for an averaged continuous-conduction-mode boost-converter model. The actor maps normalized electrical observations to a bounded duty-cycle command, and the reward contains three independent normalized terms for tracking error, error variation, and duty-cycle smoothness. The numerical environment is integrated with fourth-order Runge–Kutta substeps and is evaluated against proportional-integral (PI), anti-windup PI (PI-AW), sliding-mode control (SMC), and an independently trained deep deterministic policy gradient (DDPG) controller. In the representative median-seed run, TD3 limits peak overshoot to 2.41%, 4.17%, and 2.02% in the load, input-voltage, and parameter-mismatch tests, respectively, while all reported trajectories remain below the numerical state guards. SMC attains lower RMSE in the load and parameter-mismatch tests, whereas TD3 provides a lower peak excursion and competitive tracking. Matched six-seed TD3/DDPG experiments show overlapping uncertainty and no statistically significant paired difference at the 0.05 level. These results support a balanced finite-horizon disturbance-rejection benefit of the state-dependent TD3 policy under the stated simulation assumptions, rather than universal superiority or a hardware-level stability guarantee.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-11-12
DOI
https://doi.org/10.1038/s41598-026-77868-w
Primary Topic
Advanced DC-DC Converters
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Research on anti-disturbance adaptive control strategy of power electronic converters based on deep reinforcement learning

Juanjuan Wang, Zifeng Han
Scientific Reports
Advanced DC-DC Converters
article

Research on anti-disturbance adaptive control strategy of power electronic converters based on deep reinforcement learning

Juanjuan Wang, Zifeng Han
article en

Abstract

DC-DC boost converters are widely used in renewable-energy interfaces, energy-storage systems, electric vehicles, microgrids, and regulated DC supplies, but their voltage response can deteriorate under load changes, input-voltage fluctuations, and circuit-parameter uncertainty. This study develops an offline-trained state-dependent anti-disturbance controller based on twin delayed deep deterministic policy gradient (TD3) for an averaged continuous-conduction-mode boost-converter model. The actor maps normalized electrical observations to a bounded duty-cycle command, and the reward contains three independent normalized terms for tracking error, error variation, and duty-cycle smoothness. The numerical environment is integrated with fourth-order Runge–Kutta substeps and is evaluated against proportional-integral (PI), anti-windup PI (PI-AW), sliding-mode control (SMC), and an independently trained deep deterministic policy gradient (DDPG) controller. In the representative median-seed run, TD3 limits peak overshoot to 2.41%, 4.17%, and 2.02% in the load, input-voltage, and parameter-mismatch tests, respectively, while all reported trajectories remain below the numerical state guards. SMC attains lower RMSE in the load and parameter-mismatch tests, whereas TD3 provides a lower peak excursion and competitive tracking. Matched six-seed TD3/DDPG experiments show overlapping uncertainty and no statistically significant paired difference at the 0.05 level. These results support a balanced finite-horizon disturbance-rejection benefit of the state-dependent TD3 policy under the stated simulation assumptions, rather than universal superiority or a hardware-level stability guarantee.

Scientific Reports
Dalian Jiaotong University (CN)
Natural Science Foundation of Liaoning Province
Advanced DC-DC Converters
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon