Wenjuan Du

The modern power grid is undergoing a major transition as converter-interfaced renewables and grid-forming virtual synchronous generators (VSGs) are increasingly deployed at scale. Region of attraction (ROA) estimation is effective for analyzing large disturbance dynamics in such converter-dominated systems, but conventional model-based methods are often conservative and more difficult to apply when voltage control dynamics are included. In this paper, we propose a data-driven Koopman Lyapunov framework for certified ROA estimation of VSGs. In the offline training phase, a spectrally regularized Koopman operator, conditioned via feature-wise linear modulation (FiLM), is learned to realize linear evolution in the observable space across diverse operating conditions, thereby encouraging a better-conditioned and more dissipative latent spectrum. In the online ROA estimation phase, a branched Koopman neural Lyapunov function is constructed with Koopman eigenfunctions as the core and a lightweight convex neural branch. Approximation errors in the learned Koopman operator are handled through the Koopman neural Lyapunov function, supporting certified ROA estimation with respect to the learned dynamics. Case studies demonstrate less conservative ROA estimates and broad applicability across varying VSG conditions. Code is available at https://github.com/dAngelo1208/dko tsa vsgs.

Virtual synchronous generator (VSG) control may lead to active power oscillations under insufficient damping. Although various feedback control-based methods have been proposed to address this issue, they often degrade the inherent inertial response of the VSG. In this paper, fractional-order control is introduced into the VSG, providing additional degrees of freedom, which leads to the formulation of three fractional-order VSG (FOVSG) control strategies. The active power and frequency dynamic performances of these FOVSG schemes are analyzed and compared, and then proposes a generalized FOVSG (GFOVSG) control. By appropriately designing the GFOVSG, controllable active power dynamics can be achieved while significantly enhancing the system's inertial response. To evaluate the impact of GFOVSG on the small-signal stability of the grid-connected system, a stability analysis method is developed based on the fractional-order small-signal model of the system, and the small-signal stability criterion is derived. Compared to the VSG, grid-forming inverters with the GFOVSG can maintain small-signal stability under stronger grid conditions. The effectiveness of the proposed method is verified by simulation and experimental results.

Grid-connected wind turbine generators (WTGs) may induce sub-synchronous oscillations (SSOs) in a power system. Due to the difficulty to gain the detailed parameters of the WTGs in practice, data-driven AI method is considered to be a potential solution to identify the trouble-making WTGs (or SSO sources) in the power system. Unlike the SSO mechanism observed in traditional power systems, numerous studies and real-world SSO events have indicated that the SSO mechanisms in wind power grid-connected systems can be more complex and varied. However, most AI-based works ignore the fact that the SSO can arise from different mechanisms. Typically, AI models are trained and evaluated using data generated from a single type of SSO mechanism. However, in practice, AI models may need to identify the sources of SSOs caused by different or even unknown mechanisms. This raises an interesting question: Are mechanisms important for AI to identify oscillation sources? This paper preliminarily explores the answer to this question by study cases that consider two general SSO mechanisms: negative resistance and open-loop modal resonance. Further explainability analysis is carried out to investigate whether the SSO mechanisms affect the performance of the AI models. Results of study cases and explainability analysis provide researchers and engineers with deeper insights into the generalization ability of AI with respect to SSO mechanisms.

This paper investigates the transient stability of doubly-fed induction generator (DFIG)-based wind farms connected via a modular multilevel converter (MMC)-based high-voltage direct-current (HVDC) transmission system in the electromechanical timescale. During faults, the MMC tends to be switched to the current-limiting mode, which further interacts with the control systems of DFIG-based wind farms, thereby complicating the transient stability problem. In this paper, a large-signal model is constructed for transient stability analysis under this condition. The transient stability mechanism is revealed, and a transient stability index that considers the impact of a phase-locked loop in the electromechanical timescale is derived. The proposed index quantitatively assesses the instability risk arising from the coupled interaction between the DFIG-based wind farms and the MMC. In addition, the impacts of the current- limiting mode and other parameters (i.e., speed control loop gains, rotor shaft damping, inertia coefficient and fault location and severity) are analyzed. Based on the results of the parametric analysis, a suite of targeted countermeasures is proposed to enhance the transient stability of the power system. Finally, a theoretical analysis is demonstrated and verified through hardware-in-the-loop simulations using the Modeling Tech microgrid real-time simulation platform.

Subsynchronous oscillations (SSOs) in renewable power systems have emerged as a major challenge, jeopardizing the stability and safety of power system operations. Thus, it is essential to accurately and timely locate SSO sources. Artificial intelligence (AI)-based methods for locating SSO sources have become increasingly popular, existing AI-based methods usually fail in practical applications due to unavailable or insufficient real-world SSO data for model training, and significant distribution gaps in samples under different operational conditions. They also fail to fully utilize the temporal characteristics of oscillations and the spatial topology of the system. Moreover, these methods only focus on locating either negative-damping-SSO or forced-SSO sources. To overcome these limitations, we introduce a novel strategy termed Spatiotemporal-Adversarial-Domain-Generalization (STADG) to locate oscillation sources in both SSO scenarios of real power systems. This method allows the model to train on multi-source domains (simplified-simulation power systems) with sufficient labeled samples, and to be directly applied to an unseen test target domain (real power system) under unknow operating conditions. The proposed approach employs a graph-attention network and a long-short-term-memory network to fully leverage spatial and temporal features of SSOs. Extensive experiments on the modified IEEE-39 and WECC-179 bus systems confirm the effectiveness of the proposed approach.

The paper proposes an analytical examination of the transient synchronization stability (TSS) of a grid-connected wind farm with multiple permanent magnet synchronous generators (PMSGs). The paper establishes the large-signal model of the wind farm comprising of multiple wind turbines during the fault and derives the TSS criterion by using the transient energy function (TEF) method. The TSS criterion can be employed to estimate the TSS of the grid-connected wind farm with multiple PMSGs during the fault analytically. The paper also reveals and analytically explains how various factors may jointly impact the TSS of the wind farm during the fault. The impact of the setting values of current injections of the PMSGs and variation of power collecting lines on the TSS of the PMSG is examined in the paper. Finally, study cases of an example grid-connected wind farm with multiple PMSGs are presented to demonstrate and evaluate the effectiveness and applicability of the derived TSS criterion and hypotheses.

Maintaining power system stability necessitates optimizing power system dynamics and suppressing oscillations. To achieve this objective, significant progress has been made in proposing optimization strategies based on model-based analysis theory. However, accurately obtaining parametric models and operation conditions in a large-scale renewable power system to establish precise analysis models remains challenging. In this article, we introduce a novel strategy termed Disentangled Factor Transfer Reinforcement Learning (DFTRL) for designing supplementary damping controllers (SDCs) in static synchronous compensators (STATCOM) online to enhance the stability of practical power system. The proposed DFTRL approach allows the reinforcement learning (RL) agent, trained on a simplified power system, to be directly applied to an unseen practical power system. Through case studies, the proposed optimization strategy demonstrates the RL agent's capability to generalize effectively to the target practical power system and successfully suppress oscillations. Moreover, the agent exhibits robustness to variations in power system operating scenarios and noise present in observations.

As the proportion of renewable power generation (RPG) based on power electronic interfaces increases annually, the issue concerning the subsynchronous oscillations (SSOs) induced by the interactions between RPG and power systems is becoming increasingly prominent. A variety of mechanisms can induce SSOs, and each one necessitates a corresponding countermeasure for suppression or elimination. Therefore, precise and rapid SSO diagnosis, which enables the identification of SSO mechanisms and the tracing of SSO sources, is crucial. Consequently, a novel SSO diagnostic method based on deep learning (DL) using measurement data is proposed in this article. To overcome the challenges of obtaining sufficient and labeled data for DL training in an actual power system, a deep transfer learning (DTL) architecture is proposed to transfer knowledge from a simplified simulation power system with sufficient labeled data to an actual power system with limited unlabeled data. Furthermore, gradient-weighted class activation mapping (Grad-CAM) is utilized to analyze the interpretability of the trained DTL model for SSO diagnosis. Finally, the validity and adaptability of the proposed method are confirmed through comprehensive experiments conducted on a test example.

It is known that multiple voltage source converters (VSCs) with similar dynamics may induce subsynchronous oscillations (SSOs), particularly under weak grid connections. This study investigated the small-signal stability of multiple VSCs with different dynamics, utilizing both the model with default parameters and the model that accounts for deviations in VSC dynamics. The results indicates that the DC voltage control loops of the VSCs may cause growing SSOs even under strong grid connections. The direction and amplitude of the deviations are affected by the differences between the actual and default parameters. The impact of deviations is the largest when all the parameters are larger or smaller than the default parameters. However, the impact is reduced when some of the parameters are larger and others are smaller than the default parameters. On this basis, two reduced-order stability analysis methods were proposed to quickly and accurately assess the SSO risk in the power system. A sample power system with multiple VSCs is used to demonstrate and evaluate the conclusions and proposed methods.