Adaptive pulse width modulation design for power converters based on affine switched systems

Simone Baldi, Antonis Papachristodoulou, Elias B. KosmatopoulosNonlinear 2018 · DOI

TL;DR

This paper introduces a novel adaptive control strategy for designing pulse width modulation (PWM) signals in power converters. By employing a switched system model and an adaptive estimation approach, the authors achieve robust control that accounts for uncertainties in converter parameters, outperforming traditional methods in challenging scenarios.

What Problem Does This Paper Address?

Power converters are ubiquitous in modern electronics, and their efficient operation relies heavily on precise control of switching behavior using Pulse Width Modulation (PWM). Traditional control methods often simplify the converter's dynamics by using an "averaged model," which assumes high switching frequencies. However, this simplification can lead to inaccuracies, especially when dealing with non-ideal components or lower switching frequencies. Furthermore, real-world power converters are subject to parametric uncertainties (e.g., variations in component values, parasitic effects) that can significantly degrade performance and even lead to instability. Existing robust control methods often struggle with large uncertainties, while adaptive control techniques have primarily focused on averaged models, leaving a gap in adaptive control for the more accurate switched models of power converters. This paper aims to bridge this gap by developing an adaptive control strategy that directly addresses the complexities of switched models and parametric uncertainties.

Methods

The core of this paper lies in an adaptive control strategy that combines several key techniques:

  • Switched System Modeling: Instead of relying on averaged models, the authors utilize a switched affine system model, which more accurately captures the dynamic behavior of power converters, especially at lower switching frequencies. This model explicitly accounts for different operating modes (e.g., switch on/off).
  • Reverse Mode-Dependent Dwell Time: The PWM strategy is designed using a "reverse mode-dependent dwell time" concept. This allows for explicit definition of the duty cycle while ensuring a minimum time spent in each switching mode, which is crucial for stability and practical implementation.
  • Adaptive Parameter Estimation: To handle unknown or uncertain circuit parameters, the paper employs a parameter estimator based on a series-parallel parametric model and a gradient update law. This allows the controller to learn and adapt to the actual system parameters online.
  • Linear Matrix Inequalities (LMIs): Stability conditions and control design are formulated using LMIs. These are solved iteratively in an adaptive fashion, balancing "exploitation" (using current estimates) and "exploration" (using a persistently exciting input voltage to improve estimates).
  • Elementary Time Unit Approach: To ensure convexity of the LMIs with respect to the converter matrices, an elementary time unit approach is used, which constructs a time-varying Lyapunov function.
  • Exploitation-Exploration Mechanism: The adaptive control loop incorporates a mechanism that uses a persistently exciting input voltage. This ensures that the parameter estimates converge to the true system parameters over time.

Datasets

Not discussed in the paper. The paper focuses on theoretical development and simulation-based validation using a boost converter with specified parameters.

Key Insights and Ideas

  • Switched models are crucial for accurate power converter control: The paper convincingly demonstrates that for realistic scenarios (lower switching frequencies, non-ideal components), switched models provide a more accurate representation of power converter dynamics than averaged models.
  • Adaptive control is essential for handling parametric uncertainty: The authors highlight that real-world components are never perfectly known, and an adaptive approach is necessary to maintain performance and stability in the face of these uncertainties.
  • Reverse mode-dependent dwell time offers explicit duty cycle control: This novel PWM strategy provides a structured way to define the duty cycle while ensuring sufficient time in each mode, a key factor for stability.
  • Online parameter estimation is feasible and effective: The paper shows that it's possible to estimate unknown converter parameters in real-time and use these estimates to adapt the control strategy, leading to robust performance.
  • LMIs provide a powerful framework for stability analysis and control design: The use of LMIs allows for systematic design and verification of stability for complex switched systems.

Key Results

The paper presents a theoretical framework and simulation results demonstrating the effectiveness of the proposed adaptive control strategy.

  • Global Uniformly Ultimately Boundedness (GUUB): The proposed adaptive controller guarantees GUUB of the system state around the desired operating point, meaning that the system's state will eventually converge to a small, bounded region around the desired state, regardless of the initial conditions.
  • Robustness to Parametric Uncertainty: Simulations show that the adaptive controller can effectively handle significant parametric uncertainties, including parasitic elements, and maintain stability where a non-adaptive robust controller (KLMI-LQR) fails.
  • Fast Parameter Convergence: The parameter estimator converges to the actual parameters in less than 1 ms, demonstrating the efficiency of the adaptive approach.
  • Improved Stability Region: Unlike linear controllers with limited stability regions, the adaptive controller provides improved stability, especially in the presence of large parametric uncertainties or changes in operating points (e.g., load variations).

Limitations and Future Work

  • Computational Complexity: The computational cost of solving the LMIs increases with the number of subsystems (L). While the paper uses L=2 for practical reasons, this could be a limitation for systems with more complex switching dynamics.
  • Tuning of Design Parameters: Similar to many adaptive control designs, the performance and stability guarantees depend on the choice of certain design parameters (e.g., α, β₁, β₂, η). Finding optimal values for these parameters can require fine-tuning.
  • High-Frequency LMI Solving: The paper notes that solving LMIs periodically at very high frequencies (hundreds of kHz) is still challenging for embedded systems. The simulations assume LMIs are solved at a lower frequency than the switching frequency.
  • Future Work: The authors suggest that further research could explore extending this approach to more complex converter topologies and investigating the trade-offs between computational complexity and conservatism in the LMIs. They also mention that the learning transient, while influenced by initial conditions, is a factor to consider.

Summary updated

BibTeX
@inproceedings{baldi2018adaptive,
  author = {Simone Baldi and Antonis Papachristodoulou and Elias B. Kosmatopoulos},
  title = {{Adaptive pulse width modulation design for power converters based on affine switched systems}},
  booktitle = {Nonlinear},
  year = {2018},
  doi = {10.1016/j.nahs.2018.07.002},
}
Figure