In offshore wind farms connected via diode-rectifier-based high-voltage dc, conventional distributed control strategies have significant coupling among active and reactive power, leading to large oscillations in wind turbine (WT) converter power, voltage, and grid frequency during transients. In this work, we propose a decoupled distributed control strategy based on a novel adaptive virtual impedance (VI) method. In detail, we newly propose a simple estimator to enable each WT controller to locally estimate the global average reactive power, without the need for communication and line impedance knowledge. By comparing the estimated global average value with the actual reactive power, we achieve dynamically adjustable VI, thereby completely decoupling the power control. In addition, a proportional feedforward control is integrated into the estimator, ensuring precise reactive power-sharing among WT converters. We also develop a small-signal model of the proposed method to theoretically confirm the decoupling characteristics, and to facilitate system stability analysis and key control parameters design. Experimental results validate the superior performance of power and frequency regulation during both steady and transient states, in comparison with the existing decoupling control method.
4 November 2021 in IEEE Transactions on Power Delivery
Direct model predictive control (DMPC) has become an effective control alternative for the grid-tied modular multilevel converter. However, classical DMPC of the grid-tied MMC relies on accurate system model. System parameter variations will deteriorate the control performances of power, grid current, and circulating current. Besides, the inherent fewer switching actions of DMPC will also lead to poor circulating current performance. To solve these issues, this work proposes a robust DMPC solution for the grid-tied MMC. Novelties lie in two aspects. First, a virtual-flux (VF) based online inductance estimator is proposed, which utilizes the VF estimation error caused by inductance deviation. It achieves fast and accurate estimation of AC equivalent inductance, and reduces the dependency on system parameters. Thereafter, we propose a hybrid predictive control frame with a simple proportional-resonant (PR) controller, which considerably suppresses the circulating current. Finally, experimental results verify the effectiveness of the proposed solution.
12 November 2024 in IEEE Transactions on Industrial Electronics
In offshore wind farms connected via diode-rectifier-based high-voltage dc systems, conventional distributed control strategies of wind turbine (WT) converters suffer from significant coupling between active and reactive power and frequency, leading to large oscillations in WT power and offshore frequency during transients. Upon theoretically analyzing this issue using a small-signal model, this work proposes a novel distributed control strategy to address it. Initially, we propose a specially designed consensus-based global average reactive power observer for each local WT controller, where the consensus algorithm is modified from its basic form to support an adjustable reactive power-sharing ratio. By comparing the observed and actual values of reactive power, a dynamically adjustable virtual impedance control is achieved, completely decoupling active and reactive power control. In addition, global reactive power values are also used in frequency droop control as local references, decoupling reactive power and frequency control while achieving proportional reactive power-sharing. Experimental results confirm that the proposed solution eliminates transient oscillations of reactive power and frequency compared to existing conventional and decoupled control methods.
5 June 2024 in IEEE Transactions on Power Electronics
Predictive current control techniques rely on the accuracy of flux angle detection, which seriously impacts the control performances for induction motor drives. Aiming at this, many researches have been proposed to improve the flux angle detection accuracy, which usually complicates the system design and algorithm realization. In this work, we propose an effective and simple predictive power control strategy for induction motors that omits flux angle detection and simplifies the system design. Similar to classical predictive torque control, the proposal does not require flux angle information. Moreover, it requires no weighting factor design process. In addition, a stable flux observer in the stationary coordinate system is proposed and the conditions to satisfy Lyapunov stability are detailed. The effectiveness of this work is verified by experimental data. Compared with the classic predictive control strategies, the proposed method shows superior steady and dynamic performances with a simplified control structure and realization.
6 November 2024 in IEEE Transactions on Industrial Electronics
Transient dc offset and the relevant high current impact are crucial for dual active bridge (DAB) converters. Many previous works require additional resources or complicated implementations. Furthermore, the tradeoff between dynamic performance and the mitigation of dc offset is challenging. A fast dynamic response and a mild dc offset are usually contradictory. To address the current state of research, this article proposes a mitigation method based on model predictive control. Herein, the adjusting subshift angles are calculated based on predictive value and current value. The implementation is easy, and fast dynamics are achieved. Moreover, to address the unexpected current impact caused by the sharp changed shift angles of MPC, an optimization for transient state is proposed. The maximum output power matching references is released as much as possible, provided that the resulting current impact remains below the specified safety threshold. These methods give DABs a fast response speed with no dc offset and release the maximum power while limiting the current impact. Finally, the experimental comparisons with other schemes verify the effectiveness and superiority of the proposed method.
1 October 2026 in IEEE Transactions on Industrial Electronics