IET Power Electronics最新文献

This paper presents a novel deep reinforcement learning based extended fractal radial basis function (DRL-EFRBF) network for accurate state-of-charge (SOC) estimation in lithium iron phosphate (LiFePO₄) batteries. Unlike conventional methods such as open circuit voltage (OCV) and unscented Kalman filter (UKF), the proposed DRL-EFRBF framework eliminates the need for equilibrium conditions and predefined battery models. The proposed DRL-EFRBF approach combines adaptive policy learning with fractal-based nonlinear modeling to capture dynamic battery behaviors without relying on predefined battery models or requiring equilibrium states. The system was implemented using a Texas Instruments BQ76930EVM development kit and interfaced via an STM32 microcontroller. Experimental evaluations under constant, pulsed, and dynamic load profiles demonstrate that DRL-EFRBF achieves SOC estimation errors below 0.5%, outperforming OCV and UKF by over 90% and 70%, respectively. The proposed method exhibits rapid convergence, adapts to fluctuating conditions, and integrates effortlessly with IoT systems. These findings underscore its appropriateness for real-time battery management systems in electric vehicles and intelligent energy applications.

This study proposes a single-inductor multiple-output (SIMO) converter with adaptive reference inductor current (ARIC) control for low-power light-emitting diode (LED) applications. Owing to the benefits of the SIMO architecture, component counts and circuit size are reduced significantly. The ARIC control method inserts a freewheeling duration into every switching cycle to suppress cross-regulation issues among multiple outputs, while dynamically adjusting the freewheeling level of the inductor current based on load conditions to maintain high efficiency across the entire output range. Furthermore, by rearranging the locations of the output switches, non-isolated gate drivers can be adopted. The experimental results demonstrate that the maximum LED current error rate is less than 1.79% at an LED current of 100°mA. Compared with discontinuous conduction mode control, ARIC control provides higher efficiencies across the entire measured LED current range. Moreover, the maximum efficiency improvement is 3.7%, and the maximum reduction of the inductor current ripple is up to 47.4%. The ARIC control achieves a peak efficiency of 92.9% at the LED current of 150°mA.

The proliferation of distributed generation promotes active distribution networks (ADNs) while challenging power quality management. The topologically re-construable hybrid distribution transformer (ReC-HDT), capable of compensating for voltage sags, suppressing harmonics, and eliminating imbalances, has garnered considerable attention. However, although the voltage sag compensation capability of ReC-HDT is superior to conventional HDTs, it remains constrained by capacity limitations. This constraint may cause insufficient compensation for downstream sensitive loads in critical scenarios. Since voltage sags are sporadic quality issues, simply increasing converter capacity to address these demands would incur substantial costs. Therefore, to address this without capacity upgrades, interconnecting two ReC-HDTs via DC links is proposed to expand the compensation range. However, multiple operational modes introduced by interconnection complicate control coordination. This paper analyzes the structure of a dual-station-area interconnected system, detailing the topology reconstruction method and the operational mechanisms of the ReC-HDT. On this basis, four interconnection modes and their compensable voltage sag ranges are derived. Furthermore, a multi-mode coordinated control strategy for the interconnected system is proposed. Finally, the effectiveness of the proposed strategy is validated through digital simulations on the PLECS platform and hardware-in-loop experiments.

The medium-voltage (MV) modular multilevel converter (MMC) with hybrid Si and SiC sub-modules (SMs) is investigated in this paper to reduce switching losses. Unlike existing hybrid MMCs with full-bridge SMs, stable operation is achieved by adopting only half-bridge SiC SMs, thereby preserving the modularity of the MMC. The main contribution lies in the proper distribution of the switching actions, which comprehensively coordinates three objectives: reducing total switching loss; ensuring the capacitor voltage balance among the SMs, and limiting the capacitor voltage ripple of Si SMs. The effectiveness of the proposed distribution has been verified by both simulation and experimental results.

This paper introduces a single switch high-step-up DC-DC converter designed for photovoltaic energy system applications with the primary purpose of increasing the relatively low output voltage of photovoltaic panels to the DC-bus voltage of 400 V. The proposed high step-up converter is centered on a coupled inductor, thereby providing flexibility in the output voltage range, enabling the attainment of higher voltage gain. To recycle the leakage inductance energy, the resonance method is employed. Additionally, the clamped capacitor effectively minimizes voltage stress on the main switch by clamping the capacitor voltage on the switch, thereby alleviating large voltage spikes, resulting in the implementation of a low on-resistance single switch, leading to simplified switch control, and ultimately enhanced efficiency. By incorporating the non-isolating approach and hence having a common ground structure, the system's power density is enhanced, and electromagnetic interference (EMI) effects are reduced. A comprehensive and comparative analysis of the steady-state operation and design of the proposed converter is presented, with detailed examination of its performance metrics and thorough comparison to various existing converter topologies. Experimental results have been obtained from a 25 V/402 V/200 W prototype, validating the correct operation and feasibility of the presented converter.

Based on the technology computer-aided design (TCAD) simulation, a SPICE model of vertical silicon carbide power metal-oxide-semiconductor field effect transistor has been proposed with an improved description of the junction-type field effect transistor (JFET) region. The model describes a tilted current channel through the gate voltage V_G-controlled depletion layer and builds a direct correspondence between the resistance in the undepleted part and the current path in the JFET region. It is shown that in JFET region, the diffusion current takes about 18% of total current and has a large magnitude at high gate voltage. However, the diffusion current cannot be described by one-dimension approximation. In addition, at high gate voltage ($�>$14 V) and high drain voltage, excess electrons diffuse to the JFET region and form a high density of accumulation electrons. In this case, a constant electron mobility is not good enough to describe the drain current. A two-segment model of electric field and electron mobility through the one-dimension drift approximation is adopted to improve the agreement between the SPICE model and the TCAD simulation by using slightly higher values of electron mobility and the injected electron density. A new SPICE model of C_gs, C_gd and C_ds is also presented. The SPICE simulated two-pulse transients show very good agreement with TCAD simulation.

In grid-connected photovoltaic (PV) systems, string-type PV optimizers can realize maximum power point tracking (MPPT) based on the characteristics of single PV panels under irradiation imbalance, so as to improve the energy utilization rate of PV panels. However, the power loss of the PV optimizers cannot be ignored. Meanwhile, the coordinated start/stop control of PV optimizers with PV inverters deserves further investigation. In this paper, a high-efficiency coordinated control method for grid-connected PV system is proposed. Based on the operating modes and loss of both PV optimizers and inverters, the optimal operating mode of PV optimizers under different irradiation conditions is obtained, especially in case of irradiation imbalance of PV panels. The proposed method also realizes higher reliability and lower power loss in the coordination of the first and second stages of the PV system with the control logic design during the starting, stopping, and fault cases. Experiments are carried out with three 700 W string-type PV optimizers, and the results verify that the proposed method realizes higher conversion efficiency and reliability under different operating cases.

The flexible transmission system for the renewable energy contains multiple modular multilevel converters (MMCs) working in different modes. The reliable operations of MMCs are vital to the premium operation of the transmission system. The fault detection scheme is an available method to guarantee the reliable nature of MMCs. This paper presents a fault detection scheme for the insulated gate bipolar transistor (IGBT) open-circuit failure inside the submodule (SM). The proposed detection consists of fault monitoring and fault confirmation. The first part is based on the feature that abnormal extra components emerge in the circulating current under IGBT faults. To precisely capture this characteristic, the voltage reconstruction and discrete Fourier transformation based on the sliding windows are adopted. The general characteristics of IGBT faults in the two operation modes are analysed and the pulse signal of the specific SM is used in the fault confirmation part. The two functional parts can significantly relieve the calculation burden by shrinking the fault scope and precisely detect the faulty arm without being influenced by the operation disturbance and fault scenarios. Besides, this detection is efficient in both inverter and rectifier MMCs. The effectiveness of the detection scheme is demonstrated by the studies in Matlab/Simulink.

In this paper, a high step-up DC-DC converter is presented. The input current ripple is reduced by adding the interleaving technique on the input side. Moreover, the combination of coupled inductors (CIs) and built-in transformer (BIT) techniques is applied to achieve ultra-large gain with minimal duty cycles. By combining CIs and BIT, the component count is decreased. In addition, the applied active clamp circuits provide zero-voltage-switching (ZVS) turn-ON conditions for the MOSFETs, which absorb the energy of the leakage inductances of magnetic means and decrease the voltage stress of switches to a value lower than the output voltage. Also, the diodes turn off at zero-current-switching (ZCS), which greatly reduces their switching and reverse-recovery losses. In addition, the introduced converter has common ground, continuous input current, and automatic current sharing between phases. The proposed simplified structure with all mentioned factors results in high efficiency. The introduced converter operation description with its steady-state analysis is discussed. Finally, experimental results with a 24 V input voltage to supply a 480 V–600 W load are presented to validate the theoretical analysis.

Two-vector-based model predictive control (MPC) with optimal cycle has been well studied and widely applied in induction motor drives. The past study mainly focuses on the vector selection and duty cycle optimization. It is shown that the steady-state performance of MPC using two arbitrary voltage vectors during one control period is better than that using a fixed combination of a non-zero vector and a zero vector, especially at high speeds. However, the optimal pulse pattern is often ignored or difficult to be balanced in the state-of-the-art methods. This paper proposes an improved symmetric generalized two-vector-based MPC (SGTV-MPC) method to further reduce the current harmonics at steady state. At first, a non-zero vector is distributed equally on both sizes of the middle vector, which is not restricted to a zero vector but possibly a non-zero vector. An efficient way is employed to select the optimal vectors by comparing the sum of their respective duty cycle from deadbeat control based on space vector modulation. Then, not only the middle vector may be a non-zero vector, but also the first and last non-zero vectors are no longer equally distributed on both sizes of the middle vector. In fact, the duty cycle of the first vector is calculated by minimizing the current harmonics during one control period. Finally, the experimental results show that the proposed SGTV-MPC can reduce the current THD by 23.9% at most compared to the conventional generalized two-vector-based MPC, while maintaining similar switching frequency.