Abstract
Multi-input multi-output (MIMO) DC-DC converters are widely used in the on-board propulsion or traction systems of transportation electrification applications for power density enhancement. Among these converter topologies, multiple active bridge (MAB) converters, which can be regarded as an extended MIMO version of Dual active bridge (DAB) converters, has been recently developed aiming to decrease the weight and/or volume consumption of active components and enhance the availability. However, the conventional control and modulation strategies like proportional-integral (PI) control with single phase shift (SPS) modulation are not suitable for its nonlinear voltage characteristics, facing the drawbacks of low power transfer quality and efficiency, slow dynamic responses and limited controllable power transfer capability. In addition, accompanied by the growth of the power density demanding, the tendency of power density improvement leads to the magnetic integration investigation of the isolated HF links, that is to integrate the series inductors and a high frequency transformer (HFT) into a single magnetic component by increasing the leakage inductance of the HFT.Therefore, this thesis firstly proposes a series of advanced control methods combined with the modulation algorithms to improve the efficiency of power transfer, quality of power transfer, dynamic performance, control accuracy, and control flexibility for the MAB converter systems. In modulation part, non-ideal effects which are commonly ignored in conventional modulations are evaluated with their impacts aimed to promote the quality of power transfer. After evaluations, generalized phase shift (GPS) modulation was utilized in the decoupling algorithm to quantitatively characterize the relationship between the phase shift and the port average current, aiming at achieving the ‘modular’ switching of the fundamental modulations. Based on these modulations, a Newton-Iteration based decoupling algorithm combined with a current decoupling control algorithm is proposed. This control strategy composes of the DC reference current generators and the current to phase shifts decoupling algorithm, which divides the MAB converter with 𝑁 ports, into 𝑁 − 1 virtual DAB sub-branches. Besides the switching of modulations, the decoupling control strategies are also switchable. In this thesis, a power decoupling based configurable control (PDC-MPC) strategy inspired by the model predictive control (MPC) was developed to improve the transition performance.
To achieve the magnetic integration while promising the accuracy of voltage control, the leakage inductance of the integrated transformer should be large and accurate enough. Hence, the first step is to estimate the leakage inductance accurately, and judge whether structural modification is required for the selected core at the transformer design stage to achieve the aimed leakage inductance. This thesis proposed a reluctance-based leakage inductance evaluation strategy for the accurate estimation with air-path leakage considered. As the common calculations of the winding-arrangement-dominated leakage inductance commonly ignores the air-path leakage fluxes. Based on the estimation of the threshold leakage inductance, an optimized magnetic integration procedure for planar transformer has been developed. Then, a novel radially symmetrical star-shape core (RSSC) high frequency multi-winding transformer (HFMT) with split windings for the generalized N-port MAB converters is proposed. The developed HFMT can adjust the leakage inductances within a wide range by the magnetic integration control parameters, which includes lengths of extension legs, side and central pillars, horizontal air gaps, turns of windings and winding distribution ratio. Port consistency is promised by the insertion of vertical flux barriers.
The feasibility and effectiveness of the proposed improved modulation, modular decoupling control strategies and controllable integrated HFMT have been validated by the simulations and experiment. Results of the validations shows good matches about the proposed issues.
| Date of Award | 15 Jul 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Chunyang Gu (Supervisor), Jing Li (Supervisor) & Alan Zhang (Supervisor) |