1 November 2020 in Energy Reports
30 October 2023 in IEEE Transactions on Industry Applications
Parallel operation of power converters in islanded DC microgrids exhibits significant trade-off in voltage regulation and current sharing with conventional droop control. The converters exhibit inaccuracies in proportionate sharing of current when subject to heavy and transient loading while sharing a common bus. Moreover, the inaccuracies further persist due to unmodeled dynamics, parametric uncertainties, disturbance in the system and communication reliability. Therefore, the resilient parallel operation of power converters in DC microgrids requires a robust and fast control strategy that can mitigate the effect of disturbances and maintain regulated bus voltage with proportional current sharing amongst the power converters. Consequently, this work proposes a novel ANN driven droop control for a DC microgrid to enhance the transient response and mitigate disturbance in finite time. Two controllers based on adaptive droop strategy are proposed; the primary controller is a generalized Hebb's learning law-based PI integrated controller that can adjust the gains in real time for finite-time disturbance compensation in the networks and the secondary control regulates the bus voltage using fractional order sliding mode control. The effectiveness of the proposed method is evaluated by simulation and experiment and compared with the conventional and distributed droop control methods, proving its robust and adaptive performance for resilient DC microgrid applications.
9 December 2024 in IEEE Transactions on Sustainable Energy
Maximum power extraction and transfer from wind energy conversion systems (WECS) to the power grid depends on a high-performance control system. This paper proposes a robust tube-based linear parameter-varying (LPV) model predictive controller (MPC) for rotor speed and stator's active and reactive power control of a Doubly-Fed Induction Generator (DFIG) based WECS. The turbine dynamics and the DFIG is modeled as a single LPV system, which enables the model transformation into an equivalent linear time-invariant (LTI) system to avoid online updates of the prediction matrix. Based on the LTI representation, a tube-based LPV MPC (TLPVMPC) is developed, consisting of a tracking nominal MPC with tightened constraint sets and a disturbance controller. In the proposed method, the disturbance upper bound is estimated by Kalman filtering, which provides less conservative performance. The proposed controller is compared to sliding mode control (SMC), LPVMPC and nonlinear MPC (NMPC) methods. Simulations are conducted under model uncertainties and partial faults in the DFIG control voltages. The results show the robust performance of the proposed controller in power extraction and reduction of mechanical stress build-up compared to the other control methods.
3 October 2024 in IEEE Transactions on Power Systems
The detection and localization of the grid oscillations stemming from subsynchronous control interactions (SSCI) presents notable challenges. These include the limited reporting rate of Phasor Measurement Units, spectral leakage, and picket fence effects, which inadvertently misreport the amplitude and frequency of modes constituting these complex oscillations. To accurately detect the oscillations and localize the participating sources, a systematic approach employing synchro-waveforms is proposed. Firstly, the modes of oscillations are extracted using a time-frequency transformation, where, a realigned instantaneous frequency-based synchro transform is developed to address noise and time-varying constraints. This is further improved by employing an adaptive window which improves the energy concentration to retain the reconstruction of modes for strongly time-varying signals with significant noise. A decoupled admittance model is then extracted from the modes obtained at varying operating conditions. Consequently, a global admittance matrix is constructed identifying the active/reactive source of complex oscillation in the network. The efficacy and robustness of the proposed method against the state-of-the-art are proven by employing numerical simulations on sample networks and real-world data obtained from networks experiencing SSCI oscillations.