In this article, a stability analysis and parameter optimization method is proposed for multiple grid-forming (GFM) converter system. Firstly, a general small-signal state-space model is developed with modular method. Then, the comparative analyses of the stability characteristics and transient response are carried out for the VSG/Droop heterogeneous system and homogeneous control system. Meanwhile, the stable regions of the outer-loop control coefficients are studied, including the droop, virtual inertia coefficients and voltage integral coefficients. The impact of the proportion of VSG to droop-controlled converters on system stability is also investigated. Furthermore, a universal parameter optimization method is proposed for diverse multiple GFM converter systems. Critical parameters that govern stability are first identified via comprehensive parametric sensitivity analysis, after which the particle swarm optimization (PSO) algorithm is utilized for optimization. The objective function comprehensively incorporates the influence of all eigenvalues. The effectiveness of the proposed method is ultimately validated through simulations.

The rapid integration of renewable energy is transforming power systems into weak-grid, low-inertia networks, which increases the risk of oscillations, frequency instability, and voltage collapse. Grid-forming (GFM) inverters can autonomously regulate voltage and frequency while providing synthetic inertia, making them essential for stabilizing weak grids. However, in large-scale renewable energy power stations where multiple GFM inverters operate in parallel, interactions introduced by grid impedances, parameter mismatches, and power coupling may reduce damping characteristics and degrade transient performance under certain operating conditions. Existing control strategies either rely on precise system models, require extensive communication, or provide only local stability, limiting their applicability in dynamic and uncertain grids. This paper proposes a coordinated control framework for multi-inverter GFM stations under dynamic grid conditions. A port-energy coupling model captures system-level interactions, while a port-controlled Hamiltonian based nonlinear disturbance observer mitigates model uncertainties. Based on this framework, a real-time damping-oriented control strategy adjusts the damping parameters of individual inverters according to the identified grid impedance and the power-sharing-dependent equivalent impedance, thereby improving multi-inverter damping without high-bandwidth peer-to-peer communication. The proposed method provides robust and scalable coordination with reduced communication requirements for secure and reliable integration of renewable energy.

In distributed cascaded multilevel converters, communication between converter modules is crucial for enabling distributed control. The most cost-effective approach for this communication is utilizing the series-connected power line. However, switching noise generated by the converter modules significantly disrupts the communication channel, posing challenges to achieving reliable transmission. This article proposes a power line communication (PLC) design method specifically tailored for converter modules operating in noisy environments. In each module, a differential balanced bridge (DBB) structure is implemented to suppress noise, thereby enhancing the signal-to-noise ratio (SNR) of the PLC and enabling fast, low-latency communication. A system channel model is developed to analyze the transmission gains of both noise and signal, facilitating a comprehensive SNR evaluation. Finally, a prototype system consisting of eight H-bridge converter modules is constructed to demonstrate the practicality and effectiveness of the proposed method.

For three-phase cascaded H-bridge (CHB) photovoltaic (PV) inverters, although existing methods can effectively suppress active power backflow under single-phase and two-phase short-circuit faults, they fail to do so over a wide operating range under inter-phase short-circuit conditions. As a result, the abnormal rise in dc-side voltage caused by active power backflow cannot be effectively suppressed across all operating conditions, posing a serious threat to equipment safety and system stability. To address this issue, this paper takes the hybrid CHB (H-CHB) PV inverter as the research subject and proposes a systematic solution that combines topological characteristics with control strategies. First, a classified autonomous control strategy based on coordinated negative-sequence voltage allocation (CNSVA) is introduced, which prioritizes directing the negative sequence voltage to the three-phase bridge inverter with a common DC bus and incorporates zero-sequence voltage compensation to mitigate its impact on the CHB inverter. Second, to address the overmodulation problem that tends to occur under deep grid voltage sags, a composite control strategy integrating CNSVA and multiple harmonic injection is proposed. By injecting harmonics, the linear modulation range is extended, and the feasible region for positive- and negative-sequence voltage distribution coefficients is determined, ensuring stable system operation across all working conditions. Finally, a low-voltage, low-power experimental prototype is built to verify that the proposed control strategy can effectively suppress active power backflow and significantly enhance the transient stability and grid-connected continuous operation capability of the system.

The solid-state transformer (SST) based on a cascaded H-bridge (CHB) topology has gained widespread adoption in grid-to-vehicle (G2V) and vehicle-to-grid (V2G) systems, attributed to its high-efficiency power conversion, flexible bidirectional regulation capability, and compatibility with high-voltage applications. Nevertheless, the CHB-SST inherently suffers from double-line-frequency ripple in the H-bridge (HB) DC-bus voltage and envelope ripple in the isolated DC/DC converter. These issues lead to an increased requirement for HB DC-bus capacitance, exacerbate current stress on power devices, and diminish system efficiency. To mitigate these challenges, a collaborative control strategy is proposed to simultaneously suppress both types of ripples without introducing any additional hardware circuits. Specifically, on the one hand, adaptive compensation of the third harmonic zero-sequence voltage is implemented based on system parameters such as power factor and modulation index, thereby minimizing HB DC-bus voltage ripple. On the other hand, a notch-filter-based control strategy reshapes the output impedance of the isolated DC/DC converter, effectively eliminating both the original double-line-frequency envelope ripple and the quadruple-line-frequency envelope ripple coupled from the DC side due to third-harmonic compensation. Ultimately, a low-voltage, low-power experimental prototype was developed to validate the proposed scheme, with experimental results affirming its effectiveness and optimality.

Grid-forming (GFM) converters have broad application prospects in power electronic-based power systems due to their active support capabilities. GFM converters are limited by the short-circuit current withstand capability of power devices and are typically equipped with current limiters (CLs) to prevent equipment damage. However, CLs may cause the converter to remain locked in the current-limiting control (CLC) mode after fault clearance, thereby losing the superior characteristics of the GFM mode. To address this issue, this paper first establishes transient mathematical models of the converter under different control modes, and based on the changing law of the fault current, analyzes the constraints of the converter entering the CLC mode and recovering to the GFM mode. Then, the transient operation mechanism of the converter under complex fault scenarios is explained, and the essential reason why the converter could not exit the CLC mode is clarified. On this basis, a transient response characteristic improvement method is proposed to ensure that the converter can quickly recover to the GFM mode after fault clearance. Finally, the correctness of the theoretical analysis and the effectiveness of the proposed method are verified based on simulation and experimental results.

The existing topologies of PV generation units have fixed overcurrent capacity for each conversion stage, making it difficult to respond to various disturbances on PV and grid sides. This paper proposes a family of reconfigurable modular PV converter with flexible modules switching between the PV, grid, and energy storage (ES) sides via flexible switches which are composed of thyristors. We further study an optimal topology configuration for the non-ES case. Its control system and adaptive flexible modules dispatch strategy for efficient co-operation between PV and grid are proposed to maximize the utilization of hardware resources under various conditions such as variation of irradiance and low-voltage ride-through (LVRT). The proposed topology and its control strategy bring out enhanced flexibility and overcurrent capacity at limited additional hardware cost. The advantages of the reconfigurable modular PV converter are verified by simulation in MATLAB/Simulink and experimental tests of prototype. Experimental results demonstrate that the proposed system can achieve twice overcurrent capacity via topology reconfiguration in less than 1.85 milliseconds.

Remaining stability and a fast dynamic response under weak grid conditions are vital but challenging to the phase locked loop (PLL)-based inverters due to the PLL's side effects. Moreover, the grid strength can vary significantly due to the high penetration of renewable energy generation (REG). It brings a new challenge for the PLL-based inverters, i.e., achieving robust stability and robust rapid response (RRR) with different short circuit ratios (SCRs). This purpose is hard to achieve by using the existing methods, which mainly focus on specific SCR conditions. Quantitative robust design is seldom considered, especially for achieving RRR. Based on a widely used PLL-based inverter, this paper presents a state-feedback compensator (SFC) that provides sufficient degrees of freedom (DOFs) for pole-placement. Furthermore, a time-domain forbidden region (TFR)-based parameter tuning method is proposed to utilize the DOFs provided by the SFC, by which quantitative robust stability and RRR design can be achieved with different SCR conditions. The effectiveness of the proposed method is verified through a simulation study and a 25kVA prototype experiment.

With the increasing depletion of global traditional energy supply and escalating environmental problems, photovoltaic (PV)-energy storage-based residential power generation systems have gained significant development. These systems, equipped with an energy storage system, can operate both in grid-connected (GC) mode and islanded (IS) mode. To ensure uninterrupted power supply (UPS) for residential loads, seamless transfer between GC and IS modes is critical. Therefore, this article proposes a seamless transfer control strategy based on a unified control structure, which comprises a voltage outer loop and a current inner loop. In the GC mode, the inverter behaves as a current source and the voltage loop is effectively idle. In the IS mode, the inverter serves as a voltage source and both loops are active. The mode adaptive behavior in such a unified control structure avoids voltage and current surges caused by switching between different controllers. Since this control structure belongs to direct control, it has the merits of fast dynamic response and good power quality. To ensure smooth voltage reference during mode transfer, a simple voltage reference reconstruction and pre-synchronization control is proposed. Moreover, considering the safety standard in practical application, the relay is usually used as the grid-side switch; its delay characteristic is also fully considered when mode switching. A step-by-step controller design is provided and a specified time sequence for mode transfer is proposed. Finally, a 5-kVA experimental platform is built and various comparative experimental results are provided to validate the superiority of the proposed strategy.

With the continuous growth of renewable energy capacity, power electronic inverters have been widely deployed in power grids, and their total capacity has surpassed that of conventional synchronous generators. To maintain the stability of the power system, some inverters are operated in the grid-forming mode (GFM) instead of the conventional grid-following mode (GFL) mainly through switching over the synchronization schemes. In the GFL mode, the synchronous angle is generated by the phase-locked loop (PLL) scheme, while in the GFM mode, it is achieved through the power synchronization control (PSC). Generally, the GFL mode exhibits better adaptability to strong grid, whereas the GFM mode is more stable to weak grid. However, the specific boundary between these two modes remains ambiguous and varies with control parameters, which poses challenges both to the selection of operation modes and to parameter design for specific grid conditions. Unifying these two operation modes is crucial to dig out of this dilemma. Against this background, the small-signal dynamic performance equivalence between the GFL and GFM modes is established just through adapting control parameters properly, without the need for switching between synchronous schemes. As a result, the characteristics of the inverters adapt to grid strength continuously.

The increasing penetration of grid-connected inverters (GCI), driven by the growing share of renewable energy generation, makes their control performance and stability highly susceptible to grid impedance variations. Therefore, rapid and accurate online grid impedance identification is of great significance. To overcome the limitations of existing estimation methods, such as slow convergence and high computational burden, this paper proposes an improved model reference adaptive control (MRAC) scheme enhanced by deep reinforcement learning (DRL) for online estimation of grid resistance and inductance. Building upon the MRAC framework, the conventional adaptive law is replaced with a deep Q network (DQN). A reward function is designed to minimize the power tracking error, enabling fast and precise identification of the grid resistance and inductance. Simulation results indicate that the proposed method can accurately track line impedance changes under varying grid strengths. Compared with conventional MRAC, the proposed strategy exhibits faster convergence, validating its effectiveness and superiority.

With the large-scale grid integration of renewable energy, the computational and communication burden of traditional centralized control systems has increased, limiting system scalability. Distributed control relies on local communication and autonomous decision-making among inverters, where each inverter adjusts its output power based on local measurements and neighboring information to achieve a global objective. However, distributed control depends on interaction and communication among independent inverters, and the presence of communication and control delays in the system can significantly degrade performance and even lead to instability. Therefore, this paper proposes a consensus active power control strategy for grid-connected inverters based on time-delay Hamiltonian systems. The strategy leverages a Casimir-like function to implement predictive compensation for control delays. Compared to traditional consensus control strategies, the proposed approach ensures the stability of renewable energy power plants over a wider range of delays and exhibits higher robustness in nonlinear scenarios such as plug-and-play and arbitrary switching. Finally, simulation experiments under various operating conditions validate the effectiveness and correctness of the proposed control strategy.

The output admittance model of grid-connected inverter (GCIs) based on frequency sweeping is a critical tool for analysing the interaction characteristics between GCI and the grid. However, variations in grid impedance alter the GCI output admittance characteristics, making it impossible to apply a single admittance model derived under ideal grid conditions to scenarios with a wide range of grid impedances. Moreover, frequency sweeping operations under different grid impedances can lead to system instability near the critical short-circuit ratio, rendering it difficult to determine the most crucial stability boundary and achieve stability prediction. To address this, this paper focuses on how to obtain the admittance model in unstable regions and proposes a data-mechanism fusion-driven approach based on artificial neural networks. By integrating the mechanism model with admittance sweep data under ideal grid conditions, an equivalent sequence admittance identification model is constructed, applicable to a wide range of grid impedances, while avoiding the risk of sweep-induced resonance under critical stability conditions. Compared to data-driven methods, the proposed method reduces root mean square errors of sequence admittance magnitude and phase identification by 77.73% and 48.65%, respectively. Simulations and hardware-in-the-loop experiments validate the accuracy and applicability under wide grid impedance.

This paper proposes an intelligent negative torque control strategy to mitigate frequent start–stop cycles of wind turbines under fluctuating wind conditions. The approach enables stable low-speed operation below the cut-in wind speed by operating the generator in motor mode, thereby reducing mechanical stress and improving operational continuity. To ensure effective performance over the entire operating range, a backpropagation neural network is employed to adaptively tune the control parameters of the negative torque, optimal torque control–maximum power point tracking (OTC–MPPT), and pitch control strategies. In addition, a smooth transition mechanism based on a first-order inertial response is introduced to ensure stable switching between operating modes. The proposed method is validated through simulations of a 1.5 MW wind turbine model in MATLAB/Simulink under varying wind conditions. The results demonstrate improved system stability, reduced mechanical load fluctuations, and enhanced wind energy utilization.

Grid-forming (GFM) control enables doubly fed induction generators (DFIGs) to operate as voltage sources in islanded systems, but the stator voltage is highly sensitive to load transients, electromagnetic coupling, and parameter variations. This article proposes a disturbance-compensated single-loop variable-gain super-twisting control (DC-SLVGSTC) strategy for islanded GFM-DFIGs. The stator-voltage dynamics are first transformed into a single-loop matched-disturbance model, allowing the indirect rotor-side actuation, coupling effects, load variations, and model uncertainties to be handled in a unified compensation channel. Based on this model, a voltage-error-normalized variable-gain super-twisting controller is developed to improve transient recovery while suppressing rotor-voltage chattering. An adaptive disturbance observer (ADOB), which combines extended state observer (ESO) and generalized proportional-integral observer (GPIO) estimates through hysteresis-based fusion and smoothing, is further introduced for feedforward compensation. Lyapunov analysis establishes finite-time stability under bounded residual disturbance derivatives. Hardware experiments verify that the proposed method improves startup and load-step performance while maintaining robustness against parameter mismatch, DC-link voltage variation, and rotor-speed disturbance.

Grid-connected inverters (GCIs) can operate in both grid-following (GFL) and grid-forming (GFM) modes. However, GFL control tends to be unstable in weak grids, whereas GFM control is prone to resonance in strong grids. Recently, a hybrid control strategy based on the equivalent parallel connection of a GFL inverter and a GFM inverter has been proposed. Nevertheless, it still struggles to maintain stable operation across a wide range of short-circuit ratios (SCRs). This paper proposes a novel hybrid control based on the linear active disturbance rejection controller (LADRC). The LADRC-based approach effectively suppresses the circulating currents between the GFL and GFM controllers. The state-space model of the LADRC-based hybrid control is established, and the impact of control parameters on the eigenvalues of the proposed hybrid control is analyzed. Through the design of LADRC, the proposed hybrid control can achieve stable operation under both weak and strong grids. The grid current harmonic disturbance rejection capability is enhanced, and the voltage support ability is also maintained. Experimental results validate the effectiveness of the proposed LADRC-based hybrid control.

To solve the problem of low efficiency of triple active bridge (TAB) converter under light load and port voltage mismatch, a reactive power optimization strategy is proposed in this paper. Firstly, the model of TAB converter is established by Fourier transform, and the reactive power expressions are established. Secondly, to meet the requirements of zero voltage switching (ZVS) and power transmission, optimization constraints are set. Then, particle swarm optimization (PSO) is used to solve the reactive power expression globally. Finally, the optimal operating points are subjected to polynomial curve fitting to ensure that the phase shift angle and the transmission power are continuously varied and can be smoothly switched between different modes. Compared with the existed strategies, the proposed strategy can significantly reduce reactive power and realize ZVS in the full power range, improving the overall efficiency. Experimental test verifies the effectiveness of the proposed method.

Active Dampers (ADs) are power electronic devices that suppress harmonic oscillations in parallel grid-connected inverters by emulating virtual resistance. Although typically installed at the point of common coupling, their optimal placement for achieving the best resonance suppression performance has not been systematically investigated. Traditional resonance modal analysis faces challenges due to the requirement of detailed system modeling and computationally intensive eigenvalue decomposition. This paper proposes an improved RMA-based siting strategy, which integrates impedance scanning with offline inverter data to construct a node self-impedance model. The node with the highest participation factor is identified as the optimal installation location for the AD, and the proposed method directly derives participation factors from self-impedance, thereby avoiding complex eigenvalue decomposition. MATLAB/Simulink simulation results demonstrate that following this siting strategy for impedance adapter placement achieves optimal harmonic resonance suppression. RT-LAB experiments further validates the effectiveness of the proposed strategy. The conventional practice of installing impedance adapters at the PCC is shown to be non-optimal, while the proposed AD siting strategy provides effective theoretical support and a practical solution for optimizing impedance adapter performance, filling the gap in modal analysis-based siting for black-box inverters.

To address the problem of grid equivalent impedance estimation for multi-plant systems in the large-scale renewable energy base, this paper conducts research on modeling, impedance estimation and error mechanism. First, it clarifies the typical grid structure of multi-plant systems characterized by series-connected main feeders and parallel-connected branch plants. From the observation perspective of a single plant, a Thevenin equivalent model considering line impedance and output coupling between plants is established based on the superposition theorem, and the theoretical expression of equivalent impedance is derived accordingly. Second, aiming at the applicability limitation of the traditional power perturbation method in multi-plant scenarios, theoretical derivation reveals that the error arises from the inconsistent relative changes in the output currents of individual plants. This inconsistency leads to the failure of estimation results to reflect the output coupling effect among plants, with the results only representing the physical line impedance. Finally, simulation results validate the accuracy of the proposed model and the rationality of the derived error mechanism.