Ripple Voltage Suppression of Dual Active Bridge Converter in Electric Locomotive Using H∞ Loop Shaping Control Method

Weiyi He, Лей Ма, Shaokun ChengJournal of Physics Conference Series 2023 · DOI

Abstract The next-generation electric locomotive drive system faces the challenge of double grid frequency ripple voltage on the DC side of single-phase PWM rectifiers. This paper investigates the intermediate stage dual-active-bridge (DAB) DC-DC converter and proposes an $H_\infty$ loop shaping ripple voltage suppression control strategy. The goal is to restrain harmonic voltage while optimizing the dynamic and robust performance of the closed-loop system. A mathematical model of the DAB control system with fluctuating input voltage is established, and a reasonable compensator is designed to reconstruct the magnitude-frequency characteristics of the nominal system. The robust stability margin is then addressed as an $H_\infty$ robust control problem under coprime factor uncertainty. Simulation results in MATLAB-Simulink demonstrate that the proposed method significantly improves ripple suppression compared to traditional PI and feedforward methods, maintaining performance even under system perturbations within tolerant robust boundaries.

Introduction

In modern electric locomotives, Power Electronic Transformers (PETs) are replacing traditional heavy transformers to improve efficiency and reduce weight. A critical component of the PET is the single-phase PWM rectifier. However, single-phase systems inherently suffer from pulsating power at twice the grid frequency, which manifests as a "double grid frequency ripple" on the DC side.

This ripple is detrimental to efficiency and can destabilize the entire traction system. While traditional solutions involve large LC resonant circuits to filter this noise, these components add significant weight and cost. Alternative software-based methods, such as input voltage feedforward, are often sensitive to parameter inaccuracies. This paper proposes a robust $H_\infty$ loop shaping control strategy to actively suppress this ripple through the intermediate DAB converter stage using software logic, eliminating the need for bulky hardware filters.

System Modeling and Problem Formulation

The study focuses on the Dual Active Bridge (DAB) converter, which serves as the DC-DC isolation stage in the PET architecture. The circuit topology, shown in Figure 1, consists of two full-bridge converters linked by a high-frequency transformer.

Circuit topology of the Dual Active Bridge (DAB) converter.

To address the ripple problem, a small-signal mathematical model of the DAB converter is derived. This model specifically includes the input voltage disturbance term, creating a transfer function from the input ripple to the output voltage. This allows the controller to treat the ripple as a predictable disturbance that can be rejected through precise loop shaping.

$H_\infty$ Loop Shaping Control Strategy

The proposed control framework operates in the frequency domain and follows a structured process to ensure both performance and robustness. The core architecture is illustrated in Figure 2.

Block diagram of the $H_\infty$ loop shaping control structure.
  1. Loop Shaping via Compensators: Weights ($W_1$ and $W_2$) are designed to "shape" the open-loop magnitude-frequency curve. The objective is to ensure high gain at low frequencies for accurate tracking and specifically at the 100 Hz ripple frequency to reject the harmonic, while maintaining a steep roll-off at high frequencies to ignore sensor noise.
  2. Robust Controller Synthesis: Using the theory of coprime factor uncertainty, a suboptimal $H_\infty$ controller ($K_\infty$) is solved. This step ensures the system remains stable even if hardware parameters, such as load resistance, deviate from the nominal model. The final controller is obtained by combining $K_\infty$ with the pre-designed compensators.

Experimental Results and Discussion

The controller was validated in MATLAB-Simulink using parameters typical for an electric locomotive PET system (400V DC input/output, 10 kHz switching frequency). The proposed $H_\infty$ Loop Shaping Voltage Ripple Suppression (LSVRS) method was compared against traditional PI-based voltage loop control and input voltage feedforward methods.

Key findings include:

  • Ripple Suppression: The output voltage peak-to-peak ripple was reduced to 4.1V, a significant improvement over the 11.3V observed with traditional voltage loop control (nearly a 64% reduction).
  • Dynamic Performance: During sudden input voltage fluctuations, the proposed controller achieved faster convergence with smaller voltage spikes than the traditional PI controller.
  • Robustness: The $H_\infty$ controller maintained effective suppression even when the load resistance was decreased by 25% ($\Delta R = -2.5\Omega$). While performance degrades if perturbations exceed the calculated robust stability margin ($\gamma_{min}$), the harmonic content remains lower than that of alternative methods.

Conclusion

This paper demonstrates that $H_\infty$ loop shaping is an effective method for suppressing double grid frequency ripples in DAB converters. By treating the control loop as a virtual filter, the system achieves superior ripple rejection and robust stability without additional physical components. Future work will involve verifying these results on physical prototypes to account for real-world non-linearities and switching dead-times.

Summary updated

Abstract

Abstract The next-generation electric locomotive drive system has the problem of double grid frequency ripple voltage in the DC side of a single-phase PWM rectifier. In this paper, the intermediate stage dual-active-bridge DC-DC converter is studied, and an H ∞ loop shaping ripple voltage suppression control strategy is proposed to restrain the harmonic voltage and optimize the dynamic and robust performance of the closed-loop system. Firstly, the mathematical model of the DAB closed-loop control system with fluctuating input voltage is established. Then, based on the H ∞ loop shaping control theory, the dynamic performance, the ability of ripple voltage suppression, and the robust performance of the system are analyzed in the frequency domain. According to the results, a reasonable compensator is selected to reconstruct the magnitude frequency characteristic curve of the nominal system. After reconstruction, the solution of the robust stability margin of the corrected object is transformed into the H ∞ robust control problem under the coprime factor uncertainty. The suboptimal H ∞ robust controller is solved to obtain the best robustness boundary of the constructed system. Finally, the H ∞ loop shaping ripple voltage suppression controller is obtained by combining the H ∞ controller with the compensator. The experimental environment is set up in MATLAB-Simulink software. Compared with the traditional voltage loop control method and the input voltage feedforward control method, the dynamic performance of the proposed control strategy is slightly improved, and while within the allowable robust boundaries of the system, it has better ripple voltage suppression performance. When the system magnitude frequency curve is outside the tolerant robust stability margin, the ripple voltage suppression performance fails, but the harmonic content of output voltage is still far less than the other two control methods. The experiment results also prove the correctness of the theoretical analysis and that the proposed control method is effective and proper.
BibTeX
@article{he2023ripple,
  author = {Weiyi He and Лей Ма and Shaokun Cheng},
  title = {{Ripple Voltage Suppression of Dual Active Bridge Converter in Electric Locomotive Using H<sub>∞</sub> Loop Shaping Control Method}},
  journal = {Journal of Physics Conference Series},
  year = {2023},
  doi = {10.1088/1742-6596/2492/1/012031},
}
Figure