Analysis and Mitigation of Wideband Oscillations in PV-Dominated Weak Grids: A Comprehensive Review

Abstract

The rapid global expansion of photovoltaic (PV) generation has increased the prevalence of PV-dominated weak-grid systems, where wideband oscillations (WBOs) pose a significant challenge to secure and reliable operation. Unlike conventional electromechanical oscillations, WBOs originate from inverter control loops and multi-inverter interactions, spanning sub-Hz to kHz ranges. This review provides a PV-focused and mitigation-oriented analysis that addresses this gap. First, it clarifies the mechanisms of WBOs by mapping oscillatory drivers such as phase-locked loop dynamics, constant power control, converter–grid impedance resonance, and high-frequency switching effects to their corresponding frequency bands, alongside their engineering implications. Second, analysis methods are systematically evaluated, including eigenvalue and impedance-based models, electromagnetic transient simulations, and measurement- and data-driven techniques, with a comparative assessment of their strengths, limitations, and practical applications. Third, mitigation strategies are classified across converter-, plant-, and system-levels, ranging from adaptive control and virtual impedance to coordinated PV-battery energy storage systems (BESS) operation and grid-forming inverters. The review concludes by identifying future directions in grid-forming operation, artificial intelligence (AI)-driven adaptive stability, hybrid PV-BESS-hydrogen integration, and the establishment of standardized compliance frameworks. By integrating mechanisms, methods, and mitigation strategies, this work provides a comprehensive roadmap for addressing oscillatory stability in PV-dominated weak grids.

1. Introduction

The global pursuit of carbon neutrality has accelerated the large-scale deployment of renewable energy, with photovoltaic (PV) power becoming one of the dominant generation sources in modern grids. According to the International Energy Agency, global solar PV generation grew markedly in 2023, with an additional 320 TWh representing a 25% increase, bringing the total output to more than 1600 TWh [1]. This rise was the largest absolute increase among all renewable energy sources during the year. Forecasts indicate that solar PV alone will contribute around 80% of the global renewable capacity growth by 2030, primarily supported by the development of large-scale solar power plants and the wider adoption of rooftop systems by households and businesses [1,2]. While this transformation significantly reduces carbon emissions and fossil fuel dependence, it also creates profound technical challenges. Unlike synchronous machines, PV systems are fully power-electronic interfaced. Their inverter-based nature introduces fast dynamics, frequency-coupled interactions, and reduced system inertia, all of which complicate stability in grids with limited short-circuit strength [3]. Such conditions, commonly referred to as weak grids, are increasingly prevalent in regions with high renewable penetration, where long transmission distances and sparse conventional generation weaken the system’s ability to damp disturbances [3,4].

A major concern under weak-grid conditions is the emergence of wideband oscillations (WBOs). These oscillations differ from traditional electromechanical modes of synchronous generators by spanning a broad spectrum, from very low frequencies associated with reactive power exchange, to subsynchronous and supersynchronous modes driven by phase-locked loops (PLLs), up to kHz-range dynamics related to inverter filters and switching delays [5]. The presence of multiple PV inverters within a plant or a regional grid further complicates this phenomenon, as resonance can be excited through shared grid impedance, giving rise to collective oscillatory modes not observable in single-inverter studies [6]. The rapid global expansion of PV generation has further increased the prevalence of PV-dominated weak grids, where WBOs have become a significant challenge to secure and reliable operation. As these converter-driven dynamics extend across spatial and frequency scales, identifying and mitigating WBOs becomes increasingly difficult [7,8]. Real-world events have highlighted the severity of this problem. For instance, the North American Electric Reliability Corporation (NERC) reported that thousands of PV inverters abnormally tripped during the 2016 Blue Cut Fire event, revealing severe vulnerabilities of PV systems under grid disturbances [9]. The Australian Energy Market Operator (AEMO) has also documented instability incidents in weak grids, where renewable inverters exhibited oscillatory behavior and large-scale disconnections [10]. In a case in China, field measurements have captured 0.1 Hz oscillations in large PV plants, attributed to reactive power interactions in weak-grid conditions [11]. In another reported event, phasor measurements from a renewable-rich provincial grid captured sustained oscillations linked to multi-inverter coupling and weak short-circuit conditions [12]. These cases confirm that PV-related oscillatory instabilities are not only theoretical but also practical threats that compromise system security, power quality, and economic viability. To emphasize the global dimension of this issue, Figure 1 presents a world map summarizing reported WBO events across North America, Europe, Asia, and Australia, covering frequency ranges from sub-Hz to several kHz [3,13].

To address these challenges, effective mitigation strategies tailored to PV-dominated systems are imperative. Considerable research has been devoted to this topic, with solutions that can be classified across three hierarchical levels. At the converter level, studies demonstrate that techniques such as virtual resistance fitting, impedance-based control tuning, adaptive droop strategies, and refined pulse-width modulation (PWM) schemes can effectively enhance local stability [14,15,16,17]. At the plant level, coordinated control of multiple inverters and hybrid integration with battery energy storage systems (BESS) provide additional damping and improve oscillation resilience [18]. At the system level, wide-area damping controllers (WADCs) and the deployment of grid-forming (GFM) inverters coupled with energy storage have emerged as critical enablers for enhancing stability in PV-rich weak grids [19,20]. Each approach, however, comes with limitations: converter-level methods may increase design and parameterization complexity, plant-level coordination requires reliable communication and supervisory frameworks, and system-level measures often involve high cost and sophisticated infrastructure. These constraints are particularly significant in large-scale PV-dominated applications, where oscillatory characteristics differ fundamentally from other renewable technologies.

In particular, the lack of rotational inertia, sensitivity to irradiance fluctuations, and widespread deployment of PV in distribution networks amplify the complexity of WBOs [20,21]. Weak-grid conditions further narrow stability margins, intensify negative resistance effects, and increase susceptibility to resonance with grid impedance [22]. These features justify the need for a PV-specific review rather than a generalized renewable perspective. Several prior efforts have addressed related aspects but remain incomplete. For example, work [23] reviews stability issues of PV inverters in weak grids but focuses primarily on modeling and assessment rather than mitigation. Study [24] examined wideband oscillations in microgrids, but their scope was not PV-specific. Reference [13] provides a comprehensive review of wideband oscillation monitoring, yet it is concentrated on monitoring rather than mitigation. At a broader level, reference [5] summarizes WBO phenomena and suppression across renewable-integrated systems, but the treatment remains technology-general rather than PV-specific under weak-grid constraints. These existing efforts confirm that WBOs have become a critical concern in converter-dominated grids. However, existing reviews reveal several clear gaps. Most of them concentrate on stability mechanisms, identification or monitoring aspects, yet seldom extend to practical mitigation and control strategies. In addition, many studies discuss general renewable or microgrid contexts without identifying the PV-specific drivers, control interactions, and engineering implications. Furthermore, with the rapid evolution of “double-high” power systems, characterized by high renewable penetration and high power-electronics penetration, existing studies on oscillations have not yet kept pace with the increasing complexity of grid-electronic interactions. The growing share of converter-interfaced generation has intensified weak-grid conditions, resulting in reduced system damping and a higher susceptibility to wideband oscillations. Such oscillations have already caused significant operational and economic losses to the power industry and end users. In view of the rising frequency of wideband oscillation incidents worldwide, there is an urgent need for a unified understanding of their mechanisms and a systematic synthesis of mitigation methods applicable to PV-dominated weak grids.

To fill these gaps, this review provides a PV-focused and mitigation-oriented synthesis that bridges oscillation analysis with engineering application. The main contribution lies in four aspects. First, it develops a frequency-based classification of multi-scale wideband oscillations in PV-dominated weak grids, linking their mechanisms to PV-specific characteristics and field-observed engineering consequences. Second, existing analysis methods are critically categorized and reviewed to highlight their respective strengths, limitations, and domains of applicability, covering eigenvalue and impedance modeling, electromagnetic transient (EMT) simulations, and measurement- and data-driven techniques. Third, mitigation strategies are systematically organized into converter-level, plant-level, and system-level measures, with comparative assessment of their targeted scenarios, effectiveness, and limitations. Finally, the review identifies emerging research frontiers, including the transition to GFM inverters, artificial intelligence (AI)-driven adaptive stability management, PV-BESS-hydrogen hybrid integration, and the establishment of standardized testing and compliance frameworks.

2. Principles and Characteristics of Wideband Oscillations in PV-Dominated Grids

2.1. Definition and Scope

WBOs refer to dynamic instabilities in converter-dominated power systems that extend well beyond the classical 0.1–2 Hz electromechanical range, typically spanning from a few Hz to several kHz. Unlike classical oscillations, which originate from synchronous machine rotor dynamics, WBOs arise from the fast control loops of power electronic converters and their interactions with the frequency-dependent grid impedance. These instabilities are further amplified under weak-grid conditions, where reduced short-circuit strength limits the system’s inherent damping capability [3,25].

In PV-dominated systems, WBOs are particularly critical because PV inverters lack mechanical inertia, depend entirely on electronic control for power regulation, and are increasingly deployed at distribution-level networks. These characteristics make them highly sensitive to grid impedance variations and multi-inverter coupling effects [23,26]. The term wideband emphasizes that oscillations in such systems cannot be confined to a single frequency range:

• Sub-Hz to ~1 Hz: dominated by PLL dynamics and inter-area oscillatory interactions [11,14,23].
• 1–10 Hz and tens of Hz: associated with negative resistance introduced by constant power control (CPC) and resonance among multiple inverters [27].
• Hundreds of Hz to several kHz: driven by converter control loops, PWM delays, and LCL filter resonance, often leading to high-frequency instabilities [28,29].

These oscillations have been documented not only in PV systems but also in other converter-based infrastructures such as wind farms and high-voltage direct current (HVDC) links under weak-grid conditions [3,30,31,32]. Such field experiences confirm that WBOs represent a practical operational challenge rather than a purely theoretical concern. A precise definition and scope are thus essential as the foundation for mechanism-oriented classification, stability assessment, and the design of effective mitigation strategies discussed in the following sections.

2.2. Mechanism and Classification of Wideband Oscillations

The mechanisms of WBOs in PV-dominated power systems arise from multi-layered interactions between power electronic converters, plant-level controllers, and weak-grid conditions. Oscillations span a broad frequency spectrum, from sub-Hz to several kHz, and can be systematically categorized according to their dominant mechanisms [33]. A frequency-based classification provides a structured framework for understanding these instabilities, which is particularly relevant in PV-dominated weak grids where the absence of inertia, the constant-power nature of inverters, and the high density of power electronic devices significantly increase vulnerability.

At the sub-Hz to around 1 Hz range, oscillations in this range are typically linked to the slow dynamics of PLLs, outer power control loops, and inter-area interactions at the system level. Most PV inverters are grid-following (GFL) and rely on PLLs for phase synchronization. Under weak-grid conditions, voltage distortion, phase noise, and reduced short-circuit ratios narrow PLL stability margins [34,35], which can trigger very low-frequency oscillations manifesting as slow drifts in voltage or frequency, and in severe cases, large-scale loss of synchronism. Field evidence includes the 0.1 Hz oscillations in Chinese PV stations, attributed to reactive power and weak-grid interactions [11]. Analytical models further confirm that coupling between PLL dynamics and grid impedance can push the system into instability under moderate disturbances [36].

In the 1 to 10 Hz band, a dominant mechanism is CPC. Attempts to maintain active power despite irradiance fluctuations introduce a negative incremental impedance characteristic that couples the direct current (DC) link with the alternating current (AC) grid and can excite low-Hz oscillations [37,38]. The lack of mechanical inertia in PV inverters exacerbates this risk. When many inverters operate in parallel, shared grid impedance can amplify the effect and produce collective oscillatory modes not present in single-inverter studies [39]. Both modeling and field measurements indicate that such oscillations commonly appear in the 1–5 Hz range, with persistence strongly influenced by irradiance variability and weak-grid strength [5,23].

The 10 Hz to a few hundred Hz range (covering mid-frequency oscillations) is mainly governed by converter–grid impedance resonance and interactions between current controllers and PLL dynamics. Inverters rely on high-bandwidth inner current loops to ensure accurate grid current tracking. However, under weak-grid conditions, these loops interact with PLL delays and LCL filter resonances, creating mismatched impedance profiles between the inverter and the grid [15,40]. dq-domain impedance methods have shown that these mismatches create resonance peaks in the tens of hertz, which may not be revealed by simplified state-space models [4,41]. For PV inverters with LCL filters, the risk is amplified because filter resonance typically lies within this band, making damping essential. In addition, frequency-domain accuracy assessment methods for wideband models [13] confirm that mid-frequency instabilities must be carefully considered in EMT studies.

The high-frequency band (hundreds of Hz to kHz) is dominated by LCL filter resonance, PWM delay, and digital sampling effects. These factors introduce additional phase lags, reduce stability margins, and can excite oscillations at harmonic frequencies [26]. Compared with lower-frequency modes, these oscillations are less visible to supervisory-level monitoring but impose significant device stress, including increased heating of power semiconductor switches and accelerated aging of DC-link capacitors. Work [3] categorizes these oscillations as particularly detrimental for power quality, often leading to excessive harmonic distortion and hardware stress. Study [17] proposes a semi-discontinuous PWM method for three-level PV inverters to mitigate oscillations in this range. More recently, work [16] combines virtual impedance with machine learning (ML) techniques to suppress high-frequency resonance in microgrid-connected PV systems. Frequency-domain accuracy assessments further support the need to capture such high-frequency dynamics in model validation. Grid-level experiences, such as the France–Spain HVDC interconnection, where oscillations at ~1.7 kHz required targeted damping measures, further illustrate the practical importance of this issue [42].

Finally, multi-inverter coupling and weak-grid characteristics create additional layers of complexity across all frequency bands [6]. In large PV plants, with hundreds of inverters connected to a common bus, even individually stable units may collectively exhibit emergent oscillations due to spatial distribution, cable lengths, and heterogeneous controllers. These plant-level effects, combined with weak-grid features such as long transmission lines and the absence of synchronous inertia, amplify resonance propagation and may couple PV dynamics with other resources, including wind farms and HVDC systems [3,43]. In summary, WBOs in PV-dominated systems are associated with multi-scale dynamics: PLL-related synchronization issues in the sub-Hz range, negative incremental impedance characteristics introduced by CPC in the low-Hz range, converter–grid impedance interactions and filter resonances in the tens to a few hundred Hz range, and PWM as well as digital control delays in the kHz range. Plant-level and system-level coupling further amplifies these oscillations under weak-grid conditions. The proposed frequency-based classification provides a systematic basis for identifying and analyzing WBOs in PV plants. A comparative overview is presented in

Table 1. Classification of wideband oscillations.

Frequency Range	Typical Source/Mechanism	Main Characteristics	Impacts
Sub-Hz to ~1 Hz [3,11]	PLL slow dynamics, outer-loop interactions	PLL sensitivity to weak-grid voltage distortion and phase coupling	Slow voltage/frequency drifts, loss of synchronism, large-scale inverter disconnections
1–10 Hz [23,27,36,44]	Negative incremental impedance from CPC, multi-inverter coupling	Irradiance variability induces DC-link imbalance, amplifying CPC effects	Sustained low-frequency oscillations, protection misoperations
10 Hz to a few hundred Hz [30,43,45]	Converter–grid impedance mismatches, coupling between PLL and current controller	Resonance frequency is highly dependent on LCL filter design and control gains	Resonance peaks, mid-frequency voltage distortion, inverter disconnections
Several hundred Hz to kHz [3,16,17]	LCL resonance, PWM, and computation delays, digital sampling effects	Interaction of switching harmonics with weak-grid impedance	Excessive harmonic distortion, semiconductor device stress, accelerated capacitor aging

, linking oscillation bands with their dominant sources, main characteristics, and engineering impacts.

2.3. Engineering Implications

WBOs in PV-dominated weak-grid systems are not only theoretical concerns but also pose significant engineering challenges. Their impacts span across system stability, power quality, equipment reliability, economic performance, protection coordination, and compliance with standards. Understanding these consequences is essential for motivating the development of robust analysis methods and effective mitigation strategies. The 2016 Blue Cut wildfire in California is a typical example [9]. As shown in Figure 2 [9], Figure 2a depicts the frequency response of the Western Interconnection during the 2016 Blue Cut Fire event, where a 500 kV line fault occurred at 11:45 a.m. This caused the system frequency to drop to 59.867 Hz and recover gradually within several minutes [9]. This large disturbance led to an abrupt curtailment of approximately 1200 MW of utility-scale PV generation in the Southern California Edison (SCE) service area. Figure 2b presents the corresponding solar PV output in SCE footprint on August 16, 2016, as reported by NERC. The four numbered segments (①–④) mark successive PV curtailment and recovery events that occurred throughout the day, each associated with separate 500 kV faults recorded in the NERC report. Event No. 1 (11:45 a.m.) produced the most severe power loss of 1178 MW, while the following three events (2–4) represented additional localized trips and partial reconnections. This correlation between frequency excursions (Figure 2a) and sequential PV curtailments (Figure 2b) demonstrates the coupled dynamic response of inverter-based generation to system-wide disturbances in weak-grid conditions. Moreover,

Table 2. Engineering implications of WBOs in PV-dominated systems.

Impact Domain	Specific Effects	Representative Case/Evidence
System security and stability	Frequency drift, synchronism loss, cascading tripping	2016 NERC Blue Cut Fire (USA): 1200 MW PV tripped [9]; West Murray PV plants (Australia): weak-grid oscillations at ~7 Hz (2015–2019) and 15–20 Hz (2020) caused large-scale PV disconnection and system instability [46].
Power quality	Voltage flicker, harmonic distortion, resonance propagation	Real-world PV plants in Virginia (USA) reported 22 Hz, 38 Hz, and 82 Hz oscillations [27,47]; Ontario PV plants (Canada) experienced 20–80 Hz oscillations linked to weak-grid conditions [5,48].
Equipment reliability	Overheating, capacitor stress, transformer malfunction	High-frequency oscillations in the kilohertz range cause severe stress on IGBTs and accelerate capacitor aging [49]; Parallel PV inverters generate circulating currents in the 2–20 kHz band, significantly increasing DC-link capacitor stress [50,51]
Operational and economic performance	Energy curtailment, reduced availability, operational constraints	2016 NERC Blue Cut Fire (USA): ~1200 MW of PV output was abruptly curtailed following a 500 kV fault, resulting in significant renewable generation loss and operational constraints [9].
Protection and coordination	False relay trips, asynchronous inverter disconnection, plant-wide tripping	Wideband oscillations (5 Hz, 95 Hz) in Zhangbei PV-DC integration caused protective relay actions and large-scale inverter tripping [3]. 2016 NERC Blue Cut Fire (USA): Widespread PV inverters misoperated during a 500 kV fault, causing uncoordinated disconnection and incomplete recovery after fault clearance [9].
Standards and compliance	Lack of oscillation-specific grid codes and test procedures	IEEE PES Task Force proposed frequency-domain methods for wideband model validation in EMT studies [31]. IEEE Std 2800-2022 establishes minimum performance requirements for inverter-based resources [52], including PV, covering small-signal stability and oscillation compliance.

provides a more comprehensive summary of engineering implications with representative real-world and experimental cases.

3. Analysis Methods for Wideband Oscillations

The analysis of WBOs in PV-dominated weak-grid systems is essential for both understanding the mechanisms outlined in Section 2 and developing effective mitigation strategies. Since WBOs exhibit multi-frequency, nonlinear, and system-coupled dynamics, diverse approaches have been adopted to characterize them. Model-based techniques, such as small-signal eigenvalue analysis and impedance-based methods, provide theoretical insight into oscillatory mechanisms but may be limited under strong nonlinearity and parameter uncertainty [5,53]. Time-domain simulations [14,19] are another key tool: phasor/root mean square (RMS)-level models allow efficient long-term stability studies, while EMT simulations provide detailed dynamics and have successfully reproduced field-observed oscillations in weak grids, though they are computationally intensive. Complementing these, measurement-based methods, including fast Fourier transform (FFT), Prony analysis, wavelets, and data-driven identification, enable real-time detection and monitoring of oscillatory modes in practice [13,54]. Each approach carries distinct strengths and limitations, suggesting that hybrid strategies are increasingly necessary. The following sections review these methods systematically, establishing the analytical foundation for the mitigation strategies discussed in Section 4.

3.1. Model-Based Analysis

Model-based approaches remain the most widely applied framework for analyzing WBOs in PV-dominated weak-grid systems. Small-signal modeling and eigenvalue analysis are classical tools that linearize converter dynamics around an operating point and assess stability through eigenvalue trajectories [24,26]. These methods provide valuable theoretical insights into oscillatory modes and parametric sensitivities. For example, reference [55] builds a detailed small-signal model for an integrated PV–hydro system and show, via eigenvalues and participation factors, how ultra-low-frequency modes (<0.1 Hz) are dominated by hydro governor–turbine dynamics rather than PV dynamics. Study [56] constructs a small-signal model for renewable-rich systems with line commutated converter-based high voltage direct current (LCC-HVDC) and validates WBO analysis against detailed time-domain simulations. Work [57] develops a whole-system small-signal model for grid-connected PV plants and uses eigenvalue/participation-factor analysis to identify subsynchronous modes sensitive to voltage outer-loop parameters. Reference [36] proposes reduced-order analytical models for PV farms that accurately reproduce low-frequency oscillations and align with EMT-based admittance and eigenvalue results. However, the linear nature of small-signal analysis limits its accuracy under strong nonlinearities, such as large irradiance fluctuations and multi-inverter interactions, which are common in utility-scale PV plants [14].

To overcome these limitations, impedance-based methods have become increasingly prominent. By modeling inverters as frequency-dependent impedances in the dq domain, these methods evaluate stability through impedance ratios and Nyquist criteria [3,58]. Impedance analysis is particularly effective for capturing mid- to high-frequency oscillations, such as 10–100 Hz resonance between current loops and weak-grid impedance or hundreds to kHz instabilities linked to LCL filters [14]. Recent studies have extended impedance methods to black-box identification, using frequency-scanning techniques to characterize inverter dynamics without access to internal models. For instance, reference [59] introduces a spectral impedance-based probabilistic framework, visualizing stability margins across frequency bands and quantifying uncertainty in WBO risk. Reference [60] investigates wideband oscillation mechanisms in grid-tied PV inverters and summarizes effective suppression strategies, including PLL tuning, outer-loop gain adjustment, and active damping design. Zhao et al. establish a dq-frame impedance model for two-stage PV inverters and demonstrate that increased PLL bandwidth or reduced voltage-loop gain introduces negative-resistance behavior [61], potentially destabilizing weak-grid systems. Reference [62] presents a black-box impedance identification method that fits Laplace-domain models to frequency responses, retaining transient accuracy while reducing computation cost. More importantly, impedance-based testing has also been incorporated into industrial practice: AEMO requires impedance measurements and considers grid impedance when determining power system connection compliance for new and existing inverter-based resources and power electronic devices in the National Electricity Market (NEM) [6]; and IEEE task forces have proposed standardized procedures [52]. In summary, model-based analysis provides a clear mechanistic understanding, with small-signal models best suited for theoretical studies and impedance methods offering practical applicability in weak-grid conditions. Yet, both approaches require accurate parameters and may not fully capture nonlinear or large-signal behavior, motivating complementary use with simulation and measurement-based techniques.

3.2. Time-Domain and EMT Simulations

Time-domain simulations [13,14], particularly EMT models, are indispensable tools for analyzing WBOs in PV-dominated weak-grid systems. Unlike linearized small-signal approaches, EMT simulations preserve the nonlinear and switching dynamics of inverters [58], allowing oscillations to be reproduced with high fidelity across sub-Hz to kHz ranges. These simulations help validate analytical predictions and enable systematic testing of mitigation strategies under diverse operating conditions.

Work [19] develops a time-delayed feedback control-based damping controller for solar PV-integrated power systems and validates its robustness against operating and delay uncertainties via nonlinear time-domain simulations and real-time experiments. Detailed EMT models can capture multi-inverter interactions, controller saturation, and hardware-specific features such as PWM delay and digital sampling, which are often neglected in reduced-order models [36]. They are particularly valuable for investigating resonance clustering in large-scale PV stations, where hundreds of inverters interact through shared impedance. Work [58] compares EMT-type scanning techniques and shows that accuracy and efficiency strongly depend on the chosen reference frame and perturbation setup. Reference [63] uses black-box EMT models of a real PV farm to analyze subsynchronous oscillations and demonstrates stability trends directly from dq-admittance responses. However, EMT simulations face challenges of scalability and computational burden [64]. System-level studies involving thousands of nodes may require high-performance computing or model-order reduction techniques. Hybrid approaches, such as combining EMT with phasor-domain simulations, have been proposed to balance accuracy and efficiency in weak-grid studies.

In practice, EMT simulations are increasingly required by utilities and operators during PV commissioning and compliance assessments. For example, work [65] performs a high-fidelity EMT simulation of both the power grid and a PV plant (from field data) to replicate the 2018 “Angeles Forest” disturbance. As illustrated in Figure 3, the EMT-based workflow integrates real-world event data with high-fidelity PV-plant and transmission-grid modeling to reproduce field disturbances. The process begins with the extraction of network information from the Western Electricity Coordinating Council (WECC) 2022 Heavy Summer case and its conversion to an EMT-compatible format through E-Tran software, as reported in work [65] (Figure 3a). The study region, shown in Figure 3b, corresponds to the Angeles Forest event area, which provides the fault scenario for validation. The voltage and current responses obtained from the EMT simulation at the PV plant’s point of interconnection (Figure 3c) reveal transient under-voltage behavior consistent with the NERC field report. Finally, the EMT simulation results (Figure 3d) reproduce the measured oscillatory waveforms at both the near and remote ends of the faulted line, confirming the model’s capability to capture weak-grid interactions and low-frequency oscillations in converter-dominated systems [65]. While powerful, EMT tools should be viewed as complementary to analytical and measurement-based methods, serving as the bridge between theoretical modeling and real-world performance evaluation.

3.3. Measurement- and Data-Driven Methods

Measurement-based methods play a critical role in analyzing WBOs, particularly in PV-dominated weak-grid systems where theoretical models may be incomplete and EMT simulations computationally intensive. These approaches rely on data from phasor measurement units (PMUs), digital fault recorders (DFRs), or inverter-embedded sensors to extract oscillatory features in real time [14,66]. Classical signal-processing techniques such as FFT, Prony analysis, and wavelet transforms have been widely used to identify oscillation frequency, damping ratio, and mode shape from field data [5,13]. For example, reference [67] applies FFT combined with wavelet transforms to enhance harmonic and low-frequency oscillation detection in power systems. Study [68] uses the Prony method to extract oscillation frequency and damping ratios in grid-connected PV plants with virtual inertia.

More recently, data-driven methods have emerged as a powerful complement to classical spectral techniques. ML algorithms, including clustering, adaptive filtering, and deep neural networks, have been applied to classify oscillatory patterns and distinguish WBOs from harmonics or noise [69]. Li et al. apply PMU-based data-driven oscillation identification combining FFT [70], Prony, and dynamic mode decomposition (DMD) to a renewable station, showing that DMD improves the capture of non-stationary wideband oscillations compared to classical methods. Importantly, measurement-based analysis supports online monitoring and early-warning applications, providing practical value to system operators. Gao et al. further propose an online oscillatory stability assessment framework that integrates ANN-based wideband impedance identification with knowledge-driven models [71], enabling real-time evaluation of oscillatory risks in renewable-rich systems, including PV plants.

Nevertheless, these methods are not without challenges. Accurate identification requires high-quality, synchronized measurements, and noise sensitivity may lead to false alarms. Furthermore, data-driven approaches often function as “black boxes,” offering limited physical interpretability compared with model-based methods. To address these issues, hybrid frameworks have been proposed [3,5,24], especially combining impedance model estimation (from field or test-bed data) with model validation or tuning [58]. For example, study [72] estimates inverter-based resources impedance from varying operating points to validate models by measurement. Work [73] applies AI to approximate inverter dynamic behavior. The National Renewable Energy Laboratory (NREL) of the USA [74] uses test-bed impedance measurements to verify and calibrate impedance and admittance models. Such approaches reduce the mismatch between theory and field behavior. In summary, measurement- and data-driven methods bridge theory and practice by enabling real-time diagnosis of oscillatory phenomena. Their increasing deployment in renewable-dominated grids makes them indispensable for ensuring reliable PV plant operation, while also informing the design and validation of mitigation strategies.

While measurement- and data-driven techniques enable the identification and characterization of oscillatory modes from observed signals, their effectiveness fundamentally depends on the quality, resolution, and synchronization of field data [7,8]. In this context, the success of data-driven analysis relies on real-time monitoring to ensure the accuracy and responsiveness of oscillation detection. In practice, the real-time acquisition, transmission, and coordination of such data are enabled by advanced wide-area monitoring infrastructures. Therefore, the following section discusses the framework and technological progress of wide-area monitoring and real-time observation, which provide the essential foundation for the practical implementation of oscillation analysis and stability supervision in PV-dominated weak grids.

3.4. Wide-Area Monitoring and Real-Time Observation in PV-Dominated Weak Grids

The increasing penetration of renewable energy has transformed conventional power systems into converter-dominated and low-inertia networks [75]. Such converter-based integration reduces system stiffness and short-circuit capacity, giving rise to weak-grid conditions that are highly sensitive to power-flow fluctuations and control interactions [76]. As conventional generators, which inherently provide damping, are replaced by distributed renewable generation, the system’s ability to suppress disturbances declines. In PV-dominated weak-grid systems, voltage and frequency instabilities are aggravated by low inertia and intermittent PV output, often leading to WBOs. Wide-area monitoring systems (WAMS) provide synchronized, high-resolution phasor data from PMUs [7], enabling operators to online detect and mitigate these oscillations that traditional RMS-based systems cannot capture [8]. WAMS also enables the detection of high-frequency oscillations and allows for the implementation of advanced control strategies, such as wide-area control (WAC), to enhance system stability in renewable-rich networks [7,8,77]. In practice, converter-dominated grids also exhibit increased sensitivity to communication latency, control tuning, and measurement delays, highlighting the importance of integrated hardware-and-software monitoring solutions for real-world applications.

To address this, renewable power systems increasingly rely on PMUs and hardware-in-the-loop (HIL) configurations for continuous health monitoring and operational validation [7,78]. PMUs and phasor data concentrators (PDCs) are therefore being rapidly deployed worldwide, as they comprise the fundamental components of WAMS [8,79]. PMUs provide high-resolution, time-synchronized measurements of voltage, current, and frequency, while PDCs aggregate these data streams for real-time visualization and analysis. For PV-dominated weak-grid systems, PMU measurements are particularly critical for monitoring instability caused by low inertia and voltage fluctuations. In practice [7,8], high-resolution PMU data streams are analyzed using frequency and rate-of-change-of-frequency (ROCOF) metrics for early detection of low-inertia and voltage-related instabilities [77,80]. Work [7] demonstrates a HIL configuration integrating PMUs, PDCs, and a real-time digital simulator (RTDS) for hybrid state estimation and control validation, highlighting the importance of such real-time architectures. Work [8] formulates mathematical models for PMU allocation, demonstrating that optimized placement enhances observability and reduces communication load, both of which are critical to real-time monitoring efficiency. At the distribution level, micro-PMUs enable anomaly detection, oscillation analysis, and predictive diagnostics using real-world datasets through physics-informed and unsupervised pipelines [81]. Together, these infrastructures constitute the technological backbone of modern smart grids, where PMUs, PDCs, RTDS, and micro-PMUs work in coordination to ensure continuous observability of renewable-dominated networks. Within these systems, grid-connected inverters are the active front ends that link wind turbines, PV arrays, and energy-storage units to the grid. Their control is implemented by digital microprocessors that switch power semiconductors to deliver required active and reactive power [82]. Key considerations for inverters include ensuring high efficiency, low harmonic distortion, and robust control systems to adapt to fluctuating grid conditions like voltage and frequency deviations. For instance, work [82] reveals that the control loops of inverters, particularly the PLL, are sensitive to weak grid conditions and can negatively impact damping and stability, especially when interacting with other control loops. To address these challenges, new inverter topologies, improved control methods, and integration with energy storage are being explored to enhance power quality, grid stability, and overall performance [82].

Distributed energy resource (DER) penetration further complicates this landscape. As one of the dominant forms of DER, the large-scale expansion of PV generation directly increases the number of distributed generation nodes in both transmission and distribution networks, leading to greater operational uncertainty and control challenges. Work [77] shows that optimal phasor measurement selection enhances distribution-system state estimation under uncertain DER behavior, linking DER proliferation to the increasing demand for synchronized phasor data. In parallel, the rapid global expansion of PV generation has intensified the need for wide-area observation and control. WAMS, enabled by PMUs, provides near-real-time, high-accuracy data crucial for managing large-scale PV integration [7,8]. They allow grid operators to take proactive steps to mitigate voltage fluctuations and frequency deviations caused by solar intermittency, thereby maintaining reliability during the transition toward a renewable-heavy energy future [7,8,77]. Furthermore, advanced inverter controls, including GFM strategies, benefit from and are best validated and coordinated through WAMS-enabled infrastructures [83]. In practice, RTDS-PMU HIL platforms provide real-time observability and dynamic assessment under weak-grid conditions, facilitating the safe tuning and deployment of GFM behavior [7].

Modern WAMS now extends beyond traditional phasor-based monitoring. As illustrated in Figure 4 [11], the wideband oscillation monitoring framework integrates synchronized phasor and waveform measurements obtained from wideband phasor measurement units (WPMUs) and waveform measurement units (WMUs) [84], which are time-aligned through GPS or BeiDou signals. These data are transmitted via existing WAMS communication infrastructures to wideband phasor data concentrators (WPDCs)/waveform data concentrators (WDCs) for aggregation and centralized at the data centralization (DC) and wideband dynamics analysis (WDA) layer. The DC&WDA communicates with the energy management system (EMS) to support oscillation visualization, state estimation, and dispatch coordination, while control signals are issued to local controllers for stability protection and mitigation. This integrated monitoring-analysis-control framework establishes a scalable, real-time infrastructure that can be seamlessly embedded into existing smart-grid WAMS platforms.

3.5. Comparative Summary

The analysis of WBOs in PV power plants requires a balanced combination of model-based, simulation-based, and measurement-based methods. Each category provides complementary insights but also presents specific limitations. Model-based approaches, such as eigenvalue analysis and impedance models, offer theoretical clarity and parametric sensitivity but rely heavily on accurate converter and grid parameters. EMT simulations reproduce nonlinear dynamics and multi-inverter interactions with high fidelity, making them indispensable for validation, though their computational cost limits scalability in large networks. Measurement- and data-driven techniques, supported by WAMS, PMUs, and WMUs, are critical for real-time oscillation detection, early warning, and data-driven risk assessment, though their accuracy depends on data quality, synchronization, and signal processing. To illustrate the research evolution of these methods, Figure 5 presents the publication trends of the discussed WBO analysis approaches from 2019 to 2025. It can be seen that earlier studies mainly focused on model-based analysis and EMT simulations, while recent years show a rapid increase in measurement- and data-driven as well as hybrid approaches. This shift reflects the growing integration of real-time measurement, data analytics, and AI-assisted model validation in WBO research.

In practice, no single method can fully address the complexity of WBOs under weak-grid conditions. Future developments point toward AI-enhanced hybrid analysis, where ML augments both model-based prediction and measurement-based diagnosis [31]. The recent integration of wide-area and wideband monitoring infrastructures (e.g., PMUs, micro-PMUs, WMUs) further enables hybrid model–measurement validation and adaptive stability assessment. These advances aim to create standardized procedures that ensure robust assessment of PV oscillatory stability and provide the analytical foundation for the mitigation strategies discussed in Section 4. A comparative overview of the main analysis approaches is summarized in

Table 3. Comparative summary of analysis methods for WBOs in PV-dominated weak grids.

Category	Typical Techniques	Strengths	Limitations	Representative Applications
Model-based	Small-signal modeling, eigenvalue analysis; impedance-based stability assessment	Clear mechanistic understanding; parametric sensitivity; standardized framework	Relies on accurate parameters; limited under strong nonlinearity or large disturbances	Eigenvalue analysis of PLL-induced sub-Hz oscillations in PV–hydro systems [55]; reduced-order PV farm models validated against EMT [36]; dq-impedance scanning for compliance (AEMO, IEEE) [10,52]
Time-domain and EMT simulations	Phasor/RMS dynamic models; EMT tools (PSCAD, Simulink, EMTDC)	Captures nonlinearities, multi-inverter coupling, and hardware dynamics; validates analytical models	High computational burden; limited scalability in large networks	EMT reproduction of field oscillations (e.g., Angeles Forest event [65]); resonance clustering in large PV farms [58]; coupled-sequence EMT scanning in weak grids [63]
Measurement- and data-driven	FFT, Prony, wavelet; PMU/DFR/WAMS monitoring; AI/DMD-based identification; micro-PMU and WMU analytics	Real-time monitoring; effective for non-stationary oscillations; scalable with wide-area architectures	Sensitive to noise; requires synchronized data and advanced filtering; may lack physical interpretability	FFT-wavelet detection of harmonics/oscillations in PV plants [67]; PMU-based DMD identification of wideband modes [70]; ANN-based online risk assessment [71]; micro-PMU anomaly detection and risk prediction [81];
Hybrid frameworks	Model–measurement integration; AI-augmented EMT/impedance	Combines strengths of multiple methods; bridges theory and field validation	Still under development; lack of standardized industrial procedures	Black-box impedance validation using operating-point data [72]; AI approximation of inverter dynamics [73]; NREL test-bed impedance calibration [74]

, highlighting their respective strengths, limitations, and practical applications.

4. Mitigation Strategies for Wideband Oscillations

Mitigation of wideband oscillations (WBOs) in PV-dominated weak-grid systems is not merely an auxiliary task but a central challenge for ensuring secure and reliable operation [85]. As outlined in Section 2, the mechanisms of WBOs span multiple frequency bands, while Section 3 emphasized the importance of accurate analysis tools to reveal these dynamics. However, without effective mitigation, PV plants remain highly vulnerable to instabilities that jeopardize system security, power quality, and economic performance. Mitigation strategies are commonly organized into three hierarchical levels. At the converter level, improvements to control loops, virtual impedance, and advanced pulse-width modulation schemes enhance the dynamic robustness of individual inverters. At the plant level, coordinated control among multiple inverters, hybrid operation with BESS [86], and adaptive supervisory frameworks help suppress collective oscillatory modes. At the system level, grid-supportive devices such as flexible AC transmission system (FACTS) controllers, synchronous condensers, and emerging GFM inverters [87] reinforce weak networks against oscillatory instability. These layers are complementary rather than exclusive, forming a multi-layered “defense-in-depth” strategy. The following subsections systematically review mitigation approaches at each level, highlighting their effectiveness, limitations, and prospects for practical deployment.

4.1. Converter-Level Strategy

Converter-level strategies constitute the first line of defense against WBOs in PV systems, since inverter control loops directly govern dynamic interactions with the weak grid. These methods are generally low-cost, easy to implement, and provide rapid local stabilization. The literature reveals several mainstream directions. Virtual impedance and resistance fitting approaches are widely studied. Du et al. propose a virtual resistance fitting method to improve damping under variable irradiation [88], while reference [89] shows that tailored damping controllers effectively suppress subsynchronous oscillations. Extensions include adaptive virtual impedance combined with ML for microgrid-connected PV units [16], and PV-based shunt active power filters [90] that provide harmonic and oscillation mitigation.

Another important group of methods involves modulation and compensation schemes. Semi-discontinuous PWM has been developed for three-level PV inverters to mitigate switching-induced oscillations [17]. In parallel, quasi-harmonic voltage compensation [45] reshapes the inverter output to suppress sub- and super-synchronous modes in weak grids. New converter topologies, such as the self-excited three-port PV converter [91], also contribute to enhanced oscillation damping. Impedance-based modeling and control retuning represents a further major strand. Investigations into PV inverters, both single-stage [39] and two-stage [61], have revealed the critical role of PLLs and voltage controllers in oscillatory stability. Corresponding tuning and compensation strategies have been verified through impedance-based analysis [53,60]. Such studies, together with stability reviews on PV inverters under weak grids [23,25], provide valuable design guidelines. Lastly, intelligent and adaptive controllers have emerged as a frontier. Reinforcement learning methods, such as deep Q-networks, have been applied to PV inverters for adaptive subsynchronous oscillation damping [92].

To provide a structured comparison, the main converter-level mitigation strategies reported in the literature are summarized in

Table 4. Summary of converter-level mitigation strategies.

Reference	Target PV System	Mitigation Method	Main Contribution	Limitation
[88]	PV inverter under variable irradiation	Virtual resistance fitting	Enhanced damping and stability under irradiance fluctuations	Effectiveness depends on accurate parameter fitting
[89]	PV grid-connected system	Damping control	Suppressed sub-synchronous oscillation	Case-specific tuning required
[16]	PV units in microgrids	Adaptive virtual impedance + ML	Robust high-frequency resonance mitigation	Depends on ML training quality
[90]	PV-based shunt active power filter	Decoupled double synchronous reference frame control for harmonics	Effective harmonic/WBO mitigation in weak grids	Limited to specific filter-based applications
[17]	Three-level PV inverter	Semi-discontinuous PWM	Reduced oscillations via modulation redesign	Hardware-specific
[45]	PV grid-connected	Quasi-harmonic voltage compensation	Suppressed sub-/super-synchronous oscillations	Additional computational overhead
[91]	PV three-port converter	Novel topology with coordinated port control	Improved stability margin with multi-port design	Hardware complexity
[39]	Single-stage PV inverter	Impedance model-based tuning	Stability ensured through parameter adjustment	Requires precise modeling
[61]	Two-stage PV inverter	Impedance model + control tuning	Improved stability via controller retuning	Sensitive to grid strength
[53]	PV inverter system	Frequency-coupled impedance control	Improved dynamic robustness under weak grids	Implementation complexity
[60]	PV inverter	Impedance-based suppression	Identified unstable impedance interactions and proposed compensation	Modeling simplifications
[92]	PV grid-connected	Deep Q-Network damping control	Adaptive real-time suppression of sub-synchronous oscillations	Computational cost

, which outlines their target PV systems, technical approaches, key contributions, and limitations. As illustrated in Table 4, converter-level mitigation techniques exhibit diverse mechanisms, advantages, and limitations, each tailored to specific grid conditions and control objectives. Virtual impedance and damping-based controls are advantageous when rapid and low-cost stabilization is needed, especially under weak-grid or irradiance-varying conditions. Modulation and compensation schemes enhance waveform quality and suppress switching-induced oscillations, though their hardware dependence may limit scalability. Impedance-based tuning and model-informed methods are preferable when detailed system models are available, offering high accuracy at the cost of greater complexity. Meanwhile, intelligent and adaptive controllers, such as reinforcement learning–based designs, demonstrate strong adaptability to dynamic environments but remain computationally demanding for real-time deployment.

Overall, there is no universally superior converter-level strategy. Each method must be selected according to the specific oscillation source, grid characteristics, and control infrastructure. The optimal choice depends on the desired trade-off between implementation cost, adaptability, and robustness. In practice, these converter-level techniques often serve as the first layer of defense and can be complemented by plant- or system-level coordination for more comprehensive stability enhancement.

4.2. Plant-Level Strategy

At the plant level, mitigation of WBOs in PV power stations emphasizes the coordination of multiple inverters and hybrid resources [27]. Unlike converter-level measures that focus on single devices, plant-level strategies address collective dynamics and interaction among PV inverters, energy storage, and controllable loads within an integrated station [70]. Multi-inverter coordination plays a central role. Zou et al. highlight the transition from decentralized to centralized control for renewable-dominated distribution grids [93], where coordinated supervisory control effectively suppresses resonance in large PV clusters. Similarly, reference [50] provides an overview of resonance characteristics in multi-parallel PV inverters and summarizes resonance suppression strategies applicable to utility-scale PV plants.

Another significant direction is the integration of PV with BESS. Zhu et al. propose optimization-based suppression methods by coordinating PV inverters with controllable nonlinear loads [94], while the design of new hybrid controllers in PV-BESSs [18] demonstrates improved transient stability and effective oscillation damping. These approaches underline the benefits of coupling PV generation with fast-response storage units at the plant level. Finally, hierarchical and mode-dispatching control has been developed to handle complex interactions in large-scale stations. A hierarchical dispatching framework [6] enables adaptive allocation of GFM and GFL modes among inverters, thereby preventing unstable modes and enhancing collective stability.

A concise summary of representative plant-level strategies is provided in

Table 5. Summary of plant-level mitigation strategies.

Reference	Target PV System	Mitigation Method	Main Contribution	Limitation
[93]	PV-based distribution grid	Centralized/transition control	Suppressed resonance through coordinated supervisory control	Requires advanced communication and centralized control
[50]	Multi-parallel PV inverters	Resonance suppression strategies	Comprehensive overview of station-level strategies	Mostly conceptual, limited experimental validation
[94]	PV with controllable nonlinear loads	Optimization-based suppression	Improved damping via coordinated optimization	Sensitive to modeling accuracy
[18]	PV-BESS hybrid station	Novel controller design	Enhanced transient stability and oscillation damping	Added complexity and cost
[6]	Multi-inverter PV-BESS station	Hierarchical mode allocation (GFM/GFL)	Improved plant-level stability via adaptive dispatch	Requires reliable communication and supervisory framework

, emphasizing their critical features. Plant-level mitigation strategies exhibit complementary advantages across different system configurations and coordination schemes. Centralized or hierarchical control frameworks are particularly effective for large PV clusters where resonance arises from inter-inverter coupling; they provide strong global damping but depend on reliable communication infrastructure and supervisory coordination. Hybrid PV–BESS configurations offer a flexible means to mitigate both transient and steady-state oscillations by leveraging fast-responding storage units, though their capital cost and control complexity increase accordingly. Optimization- and dispatch-based strategies can adaptively allocate control resources or inverter operation modes (e.g., GFM/GFL) to improve collective stability, yet their performance is sensitive to model accuracy and real-time data availability.

In summary, each plant-level strategy is suited to a distinct operational context rather than being universally preferable. Centralized and hybrid schemes are ideal for large-scale or grid-connected PV plants requiring coordinated response, whereas optimization-based and mode-dispatching controls are more suitable for adaptive operation in mixed or evolving network conditions. The choice among these approaches should therefore be guided by the specific grid architecture, communication capability, and desired trade-off between control accuracy, reliability, and cost.

4.3. System-Level Strategy

System-level mitigation strategies aim to enhance the oscillatory stability of power systems with large-scale PV integration. Unlike converter- or plant-level measures, these approaches consider the broader grid dynamics and typically require advanced coordination, supervisory control, or supporting devices such as energy storage systems.

A first category concerns WADCs and coordinated system-level control. Work [85] demonstrates how wide-area signals from PV plants can be exploited to design damping controllers with adaptive delay compensation. Similarly, a robust WADC design for large-scale PV farms [19] has proven effective in suppressing low-frequency oscillations at the system scale. Another group of methods focuses on frequency regulation and virtual inertia. Rajan et al. review primary frequency control techniques applicable to large-scale PV integration, including droop-based and virtual synchronous machine methods [86]. Complementarily, system-level studies have shown that equipping PV inverters with virtual inertia can significantly affect low-frequency oscillatory modes, with appropriate parameter tuning serving as a key mitigation strategy [68].

GFM inverters and energy storage integration represent another central line of research. Work [87] proposes GFM control for PV inverters with power reserves, while study [91] demonstrates how coupling PV with ESS improves frequency modulation and stability. Reference [95] investigates adaptive droop-controlled GFM converters that suppress oscillations in GFL PV units, and Zong & Xu show that partial substitution of GFL by GFM converters enhances small-signal stability in weak grids [96]. Mid-frequency oscillation suppression using coordinated PV and GFM-energy storage system (ESS) systems has also been validated [43]. Importantly, real-world cases confirm the value of such strategies, as demonstrated by a large Australian PV plant where GFM-BESS successfully mitigates oscillations [46]. Finally, several comprehensive reviews and studies provide broad perspectives. Study [97] summarizes the state of broadband oscillation suppression in renewable power stations, while system-level reviews focusing on PV farms [20] highlight control methods for damping low-frequency oscillations. Recent studies also emphasize estimation, localization, and mitigation of wideband oscillations across PV-integrated systems [14].

Table 6. Summary of system-level mitigation strategies.

Reference	Target PV System	Mitigation Method	Main Contribution	Limitation
[85]	PV plants with wide-area measurement	Wide-area damping controller with adaptive delay compensation	Effective suppression of power oscillations at system scale	Requires reliable WAMS infrastructure
[19]	Large-scale PV farm	Robust wide-area damping controller	Enhanced damping of critical low-frequency oscillations	Complexity of implementation
[86]	Large-scale PV systems	Primary frequency control techniques	Comprehensive review of droop, freq-watt, virtual synchronous generator (VSG) methods	Mainly theoretical analysis
[68]	PV systems with virtual inertia	Parameter tuning of VSG	Mitigation of low-frequency oscillations	Sensitive to parameter design
[87]	Solar PV with reserves	GFM control	Stability improvement under disturbances	Reserve requirement limits scalability
[91]	PV + ESS	GFM inverter with frequency modulation	Improved frequency stability and damping	Relies on ESS availability
[95]	PV GFL converters	GFM with adaptive droop control	Suppressed oscillations in weak grids	Tuning complexity
[96]	Weak-grid PV plant	GFM partial substitution	Improved small-signal stability	Requires partial redesign of inverter fleet
[43]	PV + GFM-ESS hybrid	Impedance reshaping and hybrid coordination	Mid-frequency oscillation suppression	Dependent on ESS capacity
[46]	Real-world PV plant (Australia)	GFM BESS	Practical validation of WBO suppression	High capital cost
[97]	Renewable stations including PV	Review of suppression methods	System-level analysis of approaches	Review only
[20]	Large-scale PV farms	Review of damping control methods	Overview of strategies for PV farm oscillation control	Conceptual, lacks new validation
[14]	PV-integrated systems	WBO estimation, localization, mitigation	Comprehensive view of system-level measures	Generalized to mixed renewable energy systems

summarizes representative system-level strategies, outlining their mechanisms, advantages, limitations, and practical deployment cases. As shown in Table 6, system-level mitigation strategies are characterized by their wide-area coordination capability and ability to address oscillations that extend beyond individual plants. WADCs and supervisory coordination frameworks are particularly suitable for large interconnected systems where multi-point oscillations must be mitigated in a synchronized manner. Their main advantage lies in global observability and coordinated response, but this comes at the expense of significant communication and computational infrastructure requirements. Virtual inertia and frequency regulation–based controls (e.g., droop or virtual synchronous machine methods) are effective for low-frequency oscillations and grid-inertia restoration but require careful parameter tuning to avoid adverse interactions. GFM and hybrid PV–ESS solutions provide strong damping across multiple frequency ranges and have shown promising real-world performance; however, their large-scale deployment is limited by hardware cost, control complexity, and standardization challenges.

System-level strategies are most effective in transmission-scale networks or regional grids where broad coordination and hierarchical control are feasible. In practice, combining system-level coordination with lower-level local damping mechanisms yields the most robust and cost-effective framework for wideband oscillation mitigation.

To provide an integrated view of research progress, Figure 6 presents a quantitative and structural synthesis of the mitigation strategies discussed above. Figure 6a illustrates the publication distribution of converter-, plant-, and system-level mitigation studies from 2019 to 2025, revealing that converter-level methods remain dominant due to their practicality and low implementation barriers. However, the accelerating growth of system-level studies after 2023 indicates a clear research shift toward wide-area coordination, GFM integration, and hybrid PV–ESS control for system-wide stability enhancement. Figure 6b further classifies these strategies into three hierarchical layers. Converter-level techniques focus on local control enhancement and adaptive damping; plant-level coordination addresses collective oscillations within PV clusters through hybrid PV-BESS operation; and system-level frameworks emphasize wide-area damping, GFM integration, and standardized compliance. Together, these strategies form a multi-layered defense architecture that enhances oscillatory stability across device, plant, and grid scales.

5. Outlooks

Despite considerable progress, several critical gaps remain that require further research and industry collaboration:

(a)

Transition to GFM operation: Most current strategies are tied to GFM dynamics, limiting their effectiveness under high renewable penetration. GFM inverters offer a more fundamental pathway by eliminating PLL dependence and providing synthetic inertia, yet challenges of interoperability, coordination, and large-scale deployment remain. Future research should prioritize standardized GFM frameworks and mixed fleet demonstrations in 100% inverter-based resources scenarios;

(b)

AI-driven adaptive stability management: While ML has been applied mainly to oscillation detection, its greater potential lies in adaptive stability control. Real-time tuning of inverter and plant-level parameters requires high-quality field datasets and the adoption of explainable AI (XAI) approaches to ensure interpretability and trustworthiness in safety-critical power systems;

(c)

Hybrid PV–BESS–hydrogen integration: Beyond short-term battery damping, long-term stability can benefit from integrating hydrogen-based storage, aligning oscillation mitigation with decarbonization and sector coupling. Research should focus on co-optimization frameworks where BESS addresses fast oscillations while electrolyzers and hydrogen storage enhance system resilience and energy utilization;

(d)

Standardization and compliance: Current oscillation testing practices are fragmented and mainly rely on ad hoc EMT or impedance scans. There is an urgent need for harmonized protocols, similar to fault ride-through standards, that explicitly cover broadband oscillatory stability. Initiatives such as IEEE Std 2800-2022 guidelines provide important starting points for establishing robust and globally applicable compliance frameworks. Meanwhile, coordinated efforts from national regulators are equally essential to ensure that such standards are globally harmonized and practically enforceable. Establishing such frameworks would facilitate fair evaluation of mitigation strategies, accelerate industrial adoption, and enhance the resilience of PV-dominated weak-grid systems.

(e)

Digitalized wide-area monitoring and data-centric validation: The deployment of WAMS, PMUs, micro-PMUs, and WMUs will form the backbone of real-time situational awareness in future smart grids. Upcoming research should focus on unified architectures that merge wideband measurement, edge computing, and AI-assisted data analytics to enable early-warning, adaptive damping, and self-healing control. Standardized communication protocols and cyber-secure data-sharing frameworks are equally crucial to ensure reliability, interoperability, and scalability in next-generation PV-dominated systems.

6. Conclusions

This review addresses a critical yet underexplored challenge in smart grids: the analysis and mitigation of WBOs in PV-dominated, weak-grid systems. While oscillatory stability of inverter-based resources has been widely studied in general, existing reviews have often focused on wind energy systems, generic converter oscillations, or analytical modeling without a targeted focus on PV-specific mitigation. Thus, this work aims to provide a PV-centered, multi-layered analysis of mitigation strategies, systematically linking oscillation mechanisms, analysis methods, and engineering applications.

First, this work clarifies the principles and mechanisms of WBOs in PV plants. Oscillations are systematically classified across frequency ranges, linking their underlying drivers with PV-specific characteristics and practical engineering impacts, including instability, power quality deterioration, equipment stress, etc. This mechanism-based perspective offers a clearer foundation for understanding how multi-scale oscillations uniquely emerge in PV systems. Then, various analysis methods, covering model-based approaches, EMT simulations, and measurement- and data-driven diagnostics, are critically compared, with their strengths, limitations, and suitable application contexts highlighted to guide both theoretical studies and practical assessments. Third, this work adopts a structured classification of mitigation measures across converter-, plant-, and system-levels, each supported by a critical analysis of mechanisms, effectiveness, and limitations. Together, these insights form a comprehensive roadmap that bridges academic research with engineering practice, providing both mechanistic understanding and practical guidance for future PV deployment in weak grids.