Control Method for Improving Voltage Response and Ripple of Isolated DC-DC Converter

Ming Li, Kairang Wang, Zhou Zhao, Hongyong Liu, Wenyu Zhang, Jian Kou, Jianhui MengFrontiers in Energy Research 2022 · DOI

TL;DR

Researchers have developed a hybrid control strategy for isolated DC-DC converters that combines a Radial Basis Function Neural Network (RBFNN) with a novel delay-link ripple suppression method. This approach slashes voltage recovery times from over a second to just 1–2 milliseconds and reduces output voltage ripple by 75%, making it ideal for high-precision industrial applications like automotive painting.

What Problem Does This Paper Address?

Isolated DC-DC converters—specifically the phase-shifted full-bridge variety—are workhorses in heavy industry due to their high power density and electrical safety. However, they face two major "quality of life" issues:

  1. Slow Dynamic Response: When a load suddenly changes (e.g., an industrial tool kicks on), traditional Proportional-Integral (PI) controllers are often too slow to react. This causes significant voltage "sags" or "swells" that can ruin sensitive processes like electrophoretic painting or crystal growth.
  2. Input Ripple: These converters are often fed by unregulated three-phase rectifiers, which introduce a persistent "six-pulse" voltage ripple. Standard controllers struggle to suppress this periodic oscillation, leading to poor output power quality.

Methods

The authors propose a two-pronged "intelligent + deterministic" control architecture:

  • RBFNN for Load Mutation: Instead of replacing the PI controller, the authors added an RBFNN as a high-speed assistant. The neural network monitors output voltage deviation and load current. Because RBFNNs have a simple three-layer structure (input, hidden, output) and fast learning speeds, they can instantly calculate a "compensation amount" for the duty ratio. This "kickstarts" the PI controller, allowing the system to stabilize almost immediately after a load jump.
  • Delay-Free Ripple Suppression: To tackle the periodic ripple from the input rectifier, the team implemented a delay-link strategy. By calculating the exact period of the pulsation (approx. 3.3 ms) and applying a calculated compensation to the duty ratio with a specific time delay, they effectively "cancel out" the ripple before it reaches the output.
  • Hardware Implementation: The system was tested on an experimental platform using a TMS320F28335 DSP controller, proving that the algorithm is efficient enough to run on standard industrial hardware.

Datasets

The study utilized a custom dataset generated from their experimental test rig:

  • Training Data: 400 groups of sample data were collected under various operating conditions (load current, voltage deviation, and corresponding duty ratios).
  • Testing Data: 20 groups of measured data were used to verify the RBFNN's accuracy.
  • Benchmarks: The proposed method was compared directly against traditional PI control without compensation.

Key Insights and Ideas

  • The "Assistant" Approach: One of the most practical insights is using the neural network to enhance rather than replace the PI controller. This maintains the stability of a known system while adding the "reflexes" of AI.
  • Offline Training, Online Execution: To save precious clock cycles on the DSP, the RBFNN was trained offline in MATLAB. The resulting structural parameters (weights and centers) were then implemented as a lookup table, allowing for near-instantaneous online compensation.
  • Timing is Everything: The ripple suppression doesn't just change the amount of compensation; it precisely manages the delay time ($\Delta\tau$) to ensure the compensation acts exactly when the ripple peak occurs.

Key Results

The experimental results showed dramatic improvements over standard PI control:

  • Recovery Speed: Under a sudden load increase, the voltage sag duration dropped from over 1,000 ms to just 1–2 ms.
  • Voltage Stability: The amplitude of the voltage sag was reduced from 80% to 15%.
  • Ripple Reduction: The output voltage ripple amplitude was cut from 42V down to 9V, representing a 75% improvement in steady-state power quality.

Limitations and Future Work

While the results are impressive, the paper notes a few areas for further exploration:

  • Topology Specificity: The current study focuses on the phase-shifted full-bridge topology; further work is needed to see how easily this "plug-and-play" RBFNN assistant adapts to other converter types (like LLC or DAB).
  • Parameter Sensitivity: The RBFNN relies on pre-collected data. If the system components age significantly or the environment changes drastically, the "offline" training might need to be refreshed to maintain peak accuracy.
  • Complexity vs. Cost: While the DSP used is standard, the added complexity of the algorithm might require more engineering overhead than simple PI tuning for less sensitive applications.

Summary updated

Abstract

Affected by the abrupt changes of load and unregulated first-stage rectifiers, it is difficult for the output voltage of the isolated DC-DC converter with traditional proportional-integral (PI) control to meet the needs of industrial applications such as automotive electrophoretic painting. A voltage response improving the method based on a radial basis function neural network (RBFNN) is proposed in this study for the output voltage mutation of DC converter caused by load mutation. By detecting the output voltage deviation and load current value, the RBF model gives the compensation amount of the duty ratio to enhance the PI controller to speed up the voltage regulation and improve voltage response. For the six-pulse voltage ripple introduced by the unregulated rectifier in the previous stage, a delay link is adopted, where the compensation amount of the duty ratio and its delay time are reasonably designed to reduce the ripple value. The experimental test platform is built, and the proposed method is verified. The results show that the proposed method can effectively improve the output voltage response and ripple as well as improve the output voltage quality of the DC converter, which is of certain engineering application value.
BibTeX
@inproceedings{li2022control,
  author = {Ming Li and Kairang Wang and Zhou Zhao and Hongyong Liu and Wenyu Zhang and Jian Kou and Jianhui Meng},
  title = {{Control Method for Improving Voltage Response and Ripple of Isolated DC-DC Converter}},
  booktitle = {Frontiers in Energy Research},
  year = {2022},
  doi = {10.3389/fenrg.2022.858601},
}
Figure