This paper presents a novel bidirectional push–pull converter topology incorporating a coupled inductor and clamp capacitors, designed for low-voltage, high-current applications. The proposed topology combines the advantages of conventional push–pull and dual-active-bridge (DAB) converters. Compared to a traditional push–pull converter, it offers higher power density, inherent soft-switching capability, and bidirectional power flow. Relative to a DAB converter, it achieves a reduction in switch count, conduction losses, and overall cost. The use of a coupled inductor, as opposed to two discrete inductors, significantly reduces the volume of the magnetic components. The addition of the clamp capacitor facilitates the recycling of leakage inductance energy, enabling zero-voltage-switching (ZVS) for all power switches. The operational principles and steady-state characteristics of the converter are first elucidated. Subsequently, a mathematical model is derived, and a direct power control (DPC) strategy is proposed to achieve smooth output voltage tracking. Finally, simulations and experimental results based on a prototype validate the superiority of the proposed topology and the effectiveness of the control strategy.
To improve the heat dissipation performance of a dual-stator axial flux permanent magnet machine (AFPMM) with a coreless rotor, a simple rotor structure with ventilation holes is designed and optimized. First, an electro-thermal-fluid coupling model is established through multi-physical finite element analysis (FEA) to investigate the loss distribution and temperature rise of the main components, including the stator, rotor, and permanent magnets. Then, to enhance natural cooling capability, ventilation holes are introduced to the rotor structure. The fluid characteristics on its surface are analyzed, and the influence of both the number and diameter of the holes on the convective heat transfer coefficient is studied, based on which the rotor structure is optimized. In addition, the mechanical strength and electromagnetic performance of the AFPMM are evaluated by FEA. Finally, a prototype is fabricated, and the experiments are conducted to validate the effectiveness of the design method.
Accurate remaining useful life prediction is critical for ensuring the safe and reliable operation of lithium-ion batteries. Conventional deep learning approaches rely heavily on domain-specific fine-tuning, which incurs substantial data collection and training costs. This paper proposes a zero-fine-tuning RUL prediction framework based on a pre-trained TimesFM temporal foundation model. The framework integrates physical covariates including constant current charging time and constant voltage drop transition through a covariate-enhanced autoregressive mechanism, and implements an adaptive strategy for dynamic adjustment of context window and prediction horizon. Experimental validation on NASA, CALCE, MIT, and XJTU datasets demonstrates that the proposed method achieves a maximum RMSE of 0.0153. Ablation and sensitivity results further indicate that the CVDT and CCCT components are essential for stable prediction, while the complete model maintains strong robustness under noise interference, achieving average R2 values of 0.9887 and 0.9944 under 20 dB and 30 dB of noise, respectively. This paper achieves zero-fine-tuning RUL prediction without battery-specific training, significantly reducing deployment costs while maintaining high accuracy. In addition, it provides a practical solution for rapid RUL estimation in diverse battery management systems.
Efficient space utilization in electric drives necessitates optimal motor peripheral dimensions. This paper proposes a Machine Learning (ML) based optimization framework for a five-phase induction motor (FPIM), targeting the inner stator core diameter (D), core length (L), and L/D ratio to enhance key performance parameters, including the efficiency, power factor, and flux density. The motor volume (D²L) is constrained to preserve the existing utility of the machine. Training data for the ML models are generated using a detailed hardware co-simulation approach, while analytical modelling establishes the relationship between motor dimensions and performance metrics, motivating the use of ML for multi-objective optimization. The proposed method incorporates performance-parameter prioritization and accounts for dimensional tolerances in D, L, and L/D to ensure robustness in design and manufacturing. Detailed electromagnetic modeling is performed using ANSYS Maxwell, and the design is validated through a hardware-software co-simulation workflow. Regression-based ML models, including Linear Regression (LR), Decision Trees (DT), Random Forests (RF), Gradient Boosted Trees (GBR), and Support Vector Regression (SVR), are evaluated, with their hyperparameters optimized via grid search to select the best-fit model. Comparative analysis with existing dimension-optimization techniques demonstrates improvements in the flexibility, robustness, and performance-parameter selection capability of the proposed approach.
The equivalent impedance of a voltage source converter (VSC) imposes a significant influence on the stability of AC/DC hybrid systems. A positive-damping compensation method, which is equivalent to introducing a feedforward loop, is proposed in this paper. The proposed method aims to compensate for the negative damping region in the VSC DC-side impedance, suppressing the resonance of systems with non-negligible line impedance. The expression for the real component of the VSC DC-side impedance is derived via small-signal modeling and rigorous mathematical analysis. Based on this analytical expression, it is further verified that DC-side resonance arises under both power inflow conditions and power outflow conditions. The impacts of control loop parameters, power flow characteristics, and grid inductance on the formation of negative damping and the occurrence of resonance are theoretically deduced in this paper. Notably, a counterintuitive phenomenon is revealed that under power outflow conditions, the negative damping region expands when the grid inductance and power level decrease. This corresponds to the scenario where the short-circuit ratio (SCR) increases when the AC public grid is regarded as a weak grid. Experimental results under bidirectional power flow conditions further validate the resonance analysis and the effectiveness of the proposed positive-damping compensation method.
In a dual-machine configuration driven by a single three-leg inverter, where a single-phase machine is tied to the neutral node of a three-phase machine, a trade-off inherently arises between the voltage amplitude of the single-phase machine and the linear modulation range of the three-leg inverter. To address this limitation, a third-harmonic voltage injection technique is introduced. The third-harmonic voltage is used as the voltage reference for the single-phase machine drive. This lowers the peak value of the pole voltage, which extends the linear modulation range for the three-phase machine drive. Furthermore, when the voltage amplitude of the single-phase machine is set to one-sixth of that of the three-phase machine, the system achieves the same dc-link voltage utilization as a conventional three-phase machine drive system using a three-leg inverter. Experimental validation confirms the effectiveness of the proposed method and demonstrates improved dc-link voltage utilization.
Dual-active-bridge (DAB) converters have been widely adopted in DC microgrids due to their bidirectional power transfer capability and high power density. Under microgrid disturbances, DAB converters with conventional control exhibit poor dynamic performance and disturbance rejection capability. To alleviate this problem, a bandwidth-adaptive linear active disturbance rejection control (BA-LADRC) strategy is proposed in this paper. This strategy takes the converter output voltage error as the reference for bandwidth modulation, and introduces a sigmoid function into the linear active disturbance rejection controller (LADRC) to achieve smooth adjustment of the adaptive bandwidth parameters. A second-order low-pass filter is also incorporated into this strategy to suppress high-frequency error disturbances, improving the dynamic performance and disturbance rejection capability of the DAB converter. Furthermore, the Hurwitz stability criterion is employed to conduct a systematic stability analysis of the system. Finally, an experimental prototype is established. Comparisons among three different control strategies demonstrate that the voltage overshoot is reduced by approximately 56.25
The increasing penetration of renewable energy sources has accentuated the significance of dynamic variation in power grid strength, commonly characterized by the short circuit ratio (SCR). The stability of grid-following converters (GFLCs) is often compromised under conditions of low SCR, whereas grid-forming converters (GFMCs) may experience instability when operating within strong grid environments. In this paper, frequency-decoupled single–input–single–output models are proposed for both GFMCs and GFLCs. Specifically, a sequence impedance (SI) model for GFMCs with a power synchronization loop is developed, and a sequence admittance (SA) model for GFLCs with a phase-locked loop is formulated. The Nyquist criterion is applied for interaction stability analysis, elucidating the causal mechanisms responsible for GFMC instability in strong grids and GFLC instability in weak grids. Based on insights obtained from the SI model, a virtual impedance control strategy grounded in the argument principle is devised to ensure stable operation of GFMCs across a wide range of SCRs. The validity and effectiveness of the proposed models and control strategy are extensively demonstrated through numerical simulations and hardware-in-the-loop experiments. Results verify the accuracy of the models and substantiate the robustness of the control approach in maintaining converter stability under diverse grid conditions.
The shaftless rim-driven thruster (RDT) has emerged as a promising alternative to conventional shaft-driven propulsion systems due to its highly integrated architecture. The RDT achieves superior compactness by combining the electric motor and propeller into a unified assembly while eliminating the traditional drive shaft. A motor-propeller matching methodology is proposed to rapidly select the optimal RDT drive motor, enabling the RDT to functionally replace shaft-driven thrusters. The corresponding propeller parameters are calculated based on the RDT performance indicators. Then, the motor is designed in accordance with the matching method. A dual-stator single-rotor (DSSR) axial-flux permanent magnet motor is designed as the drive motor. Its performance is evaluated through parameter calculation and finite element analysis, followed by multi-objective optimization. An RDT experimental prototype integrated with a propeller is fabricated and tested, verifying the feasibility of the RDT scheme.
Modular multilevel converters (MMCs) are widely recognized for their use in flexible DC transmission and medium-voltage motor drive applications due to their rapid dynamic response, minimal output harmonic content, and flexible control. However, due to the complexity of operating conditions, most faults cause MMCs to operate in an unbalanced state, resulting in distorted output waveforms and system instability. This paper analyzes the operational characteristics of MMCs under unbalanced conditions and the mechanism of power fluctuations caused by this imbalance. To address the safety operation of MMCs and reduce the power fluctuations caused by unbalanced operating conditions, a comprehensive control strategy is proposed. on the basis of traditional control methods. This strategy introduces inter-phase energy balance control using the average energy accumulated in the arm as a reference value, and combines it with arm voltage balancing control, effectively suppressing power fluctuations in the MMC. Simulation and experimental results indicate that the control strategy can achieve safe MMC operation under asymmetric conditions.
A resonance suppression strategy for bidirectional LLC resonant converters is proposed in this paper to address the non-monotonic voltage gain problem inherent in conventional phase-shift modulation (PSM). The mechanism of parasitic resonance between the transformer magnetizing inductance and the parasitic capacitance of synchronous rectifiers (SRs) is mathematically investigated through time-domain analysis. The analysis reveals that this resonance distorts the voltage gain curve during the freewheeling interval, leading to control instability. To resolve this, the proposed topology utilizes a secondary-side auxiliary bidirectional switch. By actively clamping the secondary voltage to zero exclusively during the resonance period, the proposed control strategy effectively eliminates parasitic oscillation and restores a monotonic voltage gain curve across the entire phase-shift range. Experimental results verify that the proposed method ensures stable constant voltage control under light-load conditions and achieves higher efficiency than the conventional PSM operation.
A dual-output neutral-point-clamped CLLC resonant converter (DO-NPC-CLLC) based on flying capacitor neutral-point clamping topology is proposed in this paper. By adding two switches, this topology can realize dual-channel output voltage while completely retaining the inherent bidirectional power flow capability of the CLLC resonant converter. In this paper, the structure of the dual-output topology is described in detail, and its working principle under various working conditions is analyzed. Based on first harmonic approximation (FHA), an optimized FHA model is proposed to solve the problem of the low precision in a parameter solution when it works under resonance. In addition, the current path, resonance current characteristics, and soft switching realization capability under different modulation strategies are studied. According to this topological characteristic, a hybrid modulation strategy combining pulse frequency modulation (PFM) and primary-side phase-shift modulation (PSPSM) is proposed. This strategy effectively supports the stable operation of the converter under a variety of working conditions and a wide range of dual output voltages. Finally, the feasibility of the proposed dual-output topology and the effectiveness of the hybrid modulation strategy are fully verified on an experimental prototype.
Conventional model-free predictive control (MFPC) for three-level NPC rectifiers suffers from two inherent drawbacks: chattering induced by the sliding-mode observer (SMO) and poor dynamic performance of the voltage regulation loop. To address these issues simultaneously, this paper proposed a unified solution combining a high-order fast terminal SMO (HOFTSMO) for chattering suppression and a non-singular fast terminal sliding-mode controller (NFTSMC) with a variable reaching law for the voltage loop. The HOFTSMO fundamentally mitigates chattering to ensure superior current quality, whereas the NFTSMC, with its variable reaching law, guarantees fast and robust voltage regulation under transients. Experimental results demonstrate that the proposed method significantly reduces the current total harmonic distortion (THD) and improves the transient performance compared to conventional MFPC approaches. This work provides a robust and parameter-immune solution for high-performance power conversion systems.
In this study, a modulation optimization scheme is proposed for Vienna rectifiers in high switching frequency application scenarios such as communication power supply and electric vehicle. Although the existing carrier-based discontinuous pulse width modulation methods can reduce most of the switching losses, there are still two key problems. One is that the current distortion in the zero-crossing region has not been solved. The other is that average current fluctuation at the midpoint of the DC side causes midpoint potential oscillation. Therefore, this paper proposes a carrier modulation optimization algorithm for the Vienna rectifier. By adding a compensation component in a specific clamping area and introducing an optimized zero-sequence component in a non-clamping continuous area, the problem of zero-crossing distortion of the AC side current and unbalance of the DC side midpoint potential is alleviated. At the same time, the action times and switching losses of the switching devices are greatly reduced, and the working efficiency of the rectifier is effectively improved. Finally, the effectiveness and practicality of the proposed optimization method are verified by simulation and experimental results.
For high-frequency transformers with nanocrystalline cores, a core loss correction model that accounts for the characteristics of non-sinusoidal excitation waveforms is proposed in this paper. This approach enhances the accuracy of loss prediction under complex operating conditions, such as high frequencies and non-sinusoidal excitation. The formulation incorporates key waveform parameters, including the rise time constant and duty cycle, to derive loss expressions for both trapezoidal and rectangular waveforms. Furthermore, a conversion coefficient is introduced to correlate these losses with those under sinusoidal excitation, leading to a refined predictive model. The magnetic flux density and loss characteristics of the nanocrystalline core are evaluated using a finite element model. Experimental results confirm the accuracy of the model, with a maximum prediction error of 6.372
This paper proposes a bidirectional single-stage DC/AC converter with high-frequency isolation, resonant operation, Power Factor Correction (PFC), and reduced component count. By applying a switch multiplexing concept, the proposed topology utilizes only four active switches on the AC side, compared to the six or eight switches in conventional converters, while replacing the split capacitor with a single capacitor to eliminate voltage imbalance issues. However, conventional sinusoidal PWM introduces low-frequency harmonic components that increase grid current THD. To address this, the harmonic components are analyzed and a compensation strategy is proposed. In addition, a novel small-signal model incorporating beat-frequency dynamics is developed, and a corresponding controller design is presented to avoid instability and oscillations. Analyses of the modulation strategies, operation modes, and ZVS regions are provided, and component design guidelines are given. Finally, a 0.85-kW proof-of-concept prototype is implemented, which is widely adopted in electric vehicle (EV) battery charging applications.
With the large‒scale grid integration of power converter systems (PCSs), the synchronous inertia support capability of traditional power systems has weakened, leading to reduced system strength and increased challenges in terms of frequency and voltage regulation. Grid‒forming PCSs (GFM-PCSs), which emulates rotating machine characteristics to provide virtual inertia support, are considered an effective solution to address the decline in system inertia. However, the response characteristics of GFM-PCSs differ from those of synchronous generators and synchronous condensers, and there is currently no authoritative technical method to evaluate their dynamic performance. To accurately assess the active power‒frequency droop performance, inertia constant, and damping performance of GFM-PCSs, this paper proposes a method based on the network frequency perturbation (NFP) approach to evaluate the dynamic response of GFM-PCSs. The inertia constant and damping coefficient of a 1.725 MW power conversion system (PCS) prototype were validated on a hardware‒in‒the‒loop (HIL) simulation platform, and the impact of parameters on the output power response of GFM-PCSs under grid disturbances like phase jumps and RoCoF was analyzed. This study demonstrates that the NFP‒based method can be a valuable tool for system operators and equipment manufacturers to evaluate the inertia and damping performance of GFM-PCSs.
This paper proposes a nine-level (9 L) hybrid T-type nested neutral-point-clamped (HTNNPC) inverter composed of a four-level (4 L) T-type nested NPC (TNNPC) converter cascaded with an H-bridge cell. Owing to this configuration, the proposed inverter significantly reduces the total component count, enhances DC-bus voltage utilization, and eliminates the need for DC-link capacitor voltage balancing control. For a 3-kV/1-MW system, the proposed 9 L-HTNNPC inverter achieves main component cost reductions of 16.7
Switched capacitor techniques provide a simple way to achieve high voltage levels using minimal DC sources. They efficiently store and transfer energy through capacitors. However, designing high-voltage multilevel inverters with fewer switches and sources remains challenging. This paper proposes a scalable high-gain switched capacitor multilevel inverter (HGSC-MLI) to address this issue. The proposed inverter produces nine output levels with four times the input voltage. It requires only a DC source, two capacitors, two diodes, and nine switches, significantly reducing the overall hardware count. Additional voltage levels and higher gains can be achieved by adding switched capacitor units (SCUs), each containing one capacitor, one diode, and two switches. The design includes a self-balancing mechanism for capacitor voltages, eliminating the need for external sensors. Comparisons with existing topologies show that the proposed HGSC-MLI has better voltage boosting capability and fewer components. Simulation and experimental results from a laboratory-scale prototype validate the feasibility and effectiveness of the design.
A novel interleaved double LLC resonant converter for renewable energy applications is proposed in this paper. The proposed converter achieves a low input-current ripple without the need for additional bulky filters. It ensures soft- switching operation for all of the primary-side switches and secondary-side diodes throughout the entire input range. Moreover, it significantly reduces transformer primary winding losses, which results in remarkably improved power conversion efficiency. In addition, the proposed converter can drive two LLC resonant tanks with just two switches in a half-bridge (HB) configuration, reducing the number of switches and simplifying the circuit. The performance of the proposed converter is validated through both simulation and experimental results.
