Search

This paper proposes a synergistic approach that traverses the battery optimal size simultaneously against the optimal power management based on deep reinforcement learning (DRL). A fuel cell hybrid electric vehicle (FC-HEV) with the fuel cell/battery hybrid powertrain is used as the study case. The battery plays a key role in current transportation electrification, and the optimal sizing of the battery is critical for both system technical performances and economical revenues, especially in the hybrid design. The optimal battery design should coordinate the static sizing study against the dynamic power distribution for a given system, but few works provided the synergistic consideration of the two parts. In this study, the interaction happens in each sizing point with the optimal power sharing between the battery and the FC, aiming at minimizing the summation of hydrogen consumption, FC degradation and battery degradation. Under the proposed framework, the power management is developed with deep Q network (DQN) algorithm, considering multi-objectives that minimizing hydrogen consumption and suppressing system degradation. In the case study, optimal sizing parameters with lowest cost are determined. Leveraged by the optimal size, the hybrid system economy with synergistic approach is improved by 16.0%, comparing with the conventional FC configuration.

In order to overcome the drawbacks of the conventional protection methods in high‐voltage direct current transmission lines, a deep learning approach is proposed that directly learn the fault conditions based on unsupervised feature extraction to the detection and location decision by leveraging the hidden layer activations of recurrent neural network. The deep‐recurrent neural network boosting with the gated recurrent unit compared with the long short‐term memory unit is used by analysing both the signal presented in time domain and frequency domain. The proposed method is tested based on a modular multilevel converter based four‐terminal high‐voltage direct current system. Various faults under different conditions were simulated against fault resistance, external faults and small disturbance immunity with the validity, and the simulation verified a high accuracy, robustness and fast results because of the utilization of characteristic feature extraction.

Due to the interaction converter control, pre-existing distortion and grid impedance, the harmonic levels caused by renewable energy sources (RESs) are continuously changing, and their assessment requires the use of dedicated computer models. Several time-domain models have been proposed to carry out this analysis, however, they fall short of at least one requirement: either they do not provide accurate results, or they require an excessively long simulation time. This paper presents a novel time-domain model to address the gap described above: the proposed model employs average functions and a novel switching emulator. Therefore, it is referred to as ‘average model with switching emulator’ (AMSE). The proposed model is compared with existing models, and the results indicate that the AMSE meet both requirements stated above, as it accurately represents harmonic distortion while reducing significantly the simulation time. The second part of the paper discusses mitigating solutions to harmonic amplification in systems with a high penetration of VSCs, and shows the effectiveness of using an Active Filter to reduce harmonic levels in system experiencing resonance conditions.

For the first time, we report on the superior performance of the dual-drain gallium nitride (GaN) magnetic field-effect transistor (MagFET) at elevated temperatures. The IV characteristics of the devices reported here were collected under dc conditions and tested at elevated temperatures, 300, 323, 373, and 448 K using a custom-made heating stage, with a thermal feedback loop to accurately control the temperature. Light exposure experiments were conducted during raised temperature levels using an LED light source of wavelength 470 nm. The relative sensitivity of the GaN dual-drain MagFET was calculated and demonstrated a degradation from 9.78%T−1 at 300 K to 8.36%T−1 at 323 K, 6.10%T−1 at 373 K, and 3.79%T−1 at 448 K. This is equal to a small sensitivity decrease of 0.04%T−1/K. It is proposed that the observed reduction in sensitivity reported herein is due to increased phonon scattering in the 2-D electron gas channel. Despite this reduced sensitivity at elevated temperatures, the lowest sensitivity measured at 448 K surpasses those reported for silicon competitors.

In this paper, a vertical device structure and its process flow are proposed for fabrication of Gallium Nitride Trench Field Effect Transistors (GaNTT). A TCAD model is developed to capture effect of bulk and interface traps on the electrical performance of fully vertical Gallium Nitride devices on a Silicon Carbide substrate (GaN-on-SiC). The simulation results show a promising specific on-resistance of R sp,on = 1.4 mΩ . cm2 and a blocking voltage of BV = 540V at V GS =0V for a GaN drift thickness of 4 μm.

This paper proposes a new dual-PM excited doubly salient machine (DPM-DSM) in which the PMs are employed in both the stator yoke and the slot openings. The proposed design is derived from the existing variable flux PM machine with PMs in stator yoke by adding another layer of PMs in the slot openings. The PMs in both the stator yoke and the slot openings can interact with the armature field to generate electromagnetic torque, and consequently the torque capability of the proposed machine can be improved. Therefore, the proposed machine is very suitable for traction applications which require large torque density. Meanwhile, modular design of the stator can be achieved, which can simplify the manufacturing. The rotor only has salient poles, which is mechanically robust and works in high speed. The FEM-TS coupled method, namely the finite element method coupled with a Tabu search algorithm, is used to find the optimal design of the proposed DPM-DSM. The electromagnetic performance of the proposed machine is comprehensively investigated using FEM. The results show that with the aid of slot PMs the electromagnetic torque of the proposed machine can be increased by 62%.

In this paper, the influence of rotor pole number on the performance of a new kind of slot permanent magnet (PM) machine with complementary rotors is investigated. The major merit of this novel machine concept is the use of complementary rotors, which consist of an inner rotor and an outer rotor that are mechanically staggered by 180 electrical degrees, to provide complementary paths for the PM excitation flux. Two layers of PMs are employed, located on the inside and outside of the slots. Meanwhile, with only salient poles, the rotors are mechanically robust. The investigation is based on 12 stator pole machines with 10, 11, 13 and 14 rotor poles. Electromagnetic performance measures including flux linkage, back electromotive force (EMF), torque characteristics, loss and efficiency are comprehensively studied using the finite element method (FEM). The results show that the machine with 11 rotor poles can achieve the largest torque density and rated efficiency, as well as minimum ratio of cogging torque to rated torque.

This paper presents a methodology to model GaN power HEMT switching transients. Thus, a compact model to predict devices’ pulse switching characteristics and current collapse reliability issue has been developed. Parasitic RC subcircuits and a standard double-pulse switching tester to model intrinsic parasitic effects and to analyze power dissipation of GaN power HEMT are proposed and presented. Switching transient including gate-lag and drain-lag is predicted for ideal (without trap) and nonideal (with trap) devices. The results are validated by and compared to 2-D finite-element technology computer-aided design simulations. The original aim of this exercise is to develop a fast (near-real-time) model which can predict dynamic behavior of single and multiple power GaN HEMTs used for the switching transients of GaN power devices at circuit level.

With increasing penetration of distributed generators (DGs), existing distribution grids are becoming weaker. As a result, voltage regulation in distribution grids impose a significant challenge and can be improved by introducing additional control apparatuses. In this paper, a transformerless static synchronous series compensator (T-SSSC) is adopted to maximise the power transfer between DGs and a weak grid by improving grid voltage stability. A new control strategy is introduced to reduce the number of switching events during the operation of a T-SSSC, thus limiting the switching losses. Furthermore, a new droop based control scheme is proposed to minimise the voltage unbalance between phases. A medium voltage distribution grid with embedded T-SSSCs modelled in Simulink has been used to verify the effectiveness of the proposed control schemes. For completeness, the performance a T-SSSC is investigated following a sudden change in grid short circuit level.

This letter presents first-ever fabricated GaN split-current magnetic sensor device. Device operation and key manufacturing steps are also presented. The measured relative current sensitivity is constant at 14 % T−1 for wide mT range of the magnetic field. Constant sensitivity of a fabricated sensor can be attributed to device’s 2DEG nature, i.e., its high electron concentration and mobility, and very small layer thickness.

Gallium nitride (GaN) magnetic high electron mobility transistors (MagHEMTs) with different gate lengths intended for integration with magnetic flux concentrator for galvanic isolation are presented. Detailed discussions on the physical mechanisms behind the sensitivity change at room temperature with respect to gate geometry are given. The relative sensitivity of dual-drain GaN MagHEMTs with a device length of ${L} = {65}\,\,\mu \text{m}$ and a width of ${W} = {20}\,\,\mu \text{m}$ is measured at the highest of ${S} = {17.21}$ %/T and the lowest of ${S} = {7.69}$ %/T at ${V}_{{\text {GS}}}= -{2}$ V and ${V}_{{\text {GS}}}= {0}$ V, respectively. In addition, a novel spiral magnetic flux concentrator with the conversion factor of up to ${F}_{C}= {96}$ mT/A is designed for improving the performance of the optimized MagHEMTs in ICs. It is predicted that a spiral configuration is a necessity to enhance the conversion factor for a long MagHEMT.

In this work, we have shown that betagallium oxide (Ga2O3) thin films of differing thickness could be obtained by physical vapor deposition method, employing proper annealing conditions. These enable us to compare the variation of optical properties like transparency, band gap in these phases. Apart from these, our analysis of transmittance spectra of beta-Ga2O3 indicated the reduction of structural disorders (amorphous to crystalline) with increase in annealing temperature. The calculated band gap based on Density Functional Theory (DFT) for bulk beta-Ga2O3 thin films ~ 4.9 eV (direct) at room temperature is in excellent agreement with our experimentally measured values. This work will serve as design guidance for the new Ga2O3 based thin film electronics.

This paper reports on the 1st gated gallium nitride (GaN) magnetic sensitive high electron mobility transistors (MagHEMTs) operation and their key manufacturing steps. The output characteristic, change of sensitivity with respect to biasing conditions and ambient temperatures are presented here. The sensitivities of 19 %/T and 4.68 %/T are reported as a response of the GaN MagHEMTs to uniform magnetic field of 30 mT at ambient temperatures of 25 °C and 175 °C, respectively.

This paper explores and discusses various design considerations for inverter-based systems. Different load sharing techniques are presented for the integration of renewable energy sources within islanded microgrids. In off-grid connection, renewable energy sources are often configured to share power based on their rated capacity. This paper explores both conventional and dynamic load sharing interaction between distributed generation units, both in an inductive (high voltage) and resistive (low voltage) networks. Load sharing based on the proper design of virtual impedance is also reviewed.

This paper presents a dynamic droop load sharing scheme based on the available generation capacity of the distributed generation (DG) units. Since conventional droop schemes share loads proportional to units’ ratings, they suffer from the inability to maintain an efficient operating point when their input renewable power varies, without imposing their new operating point on other connected DGs in the microgrid. This problem is mainly due to the insensitivity of the droop scheme to the varying nature of the renewable resources used, including wind and solar photovoltaic (PV). A control method is proposed for PV systems; however, it is applicable for all types of droop-controlled renewable DG. A stability analysis of the proposed scheme on DG units is also presented to identify theoretical and practical limits. The proposed scheme identifies the dc operating zone of the inverter-based source as irradiance level changes and conditions the droop parameters appropriately for an efficient load sharing based on available generation, while the rating of each unit is also taken into account. The proposed scheme provides energy saving, since energy demand from a local auxiliary generator is reduced. The proposed method is validated using MATLAB/SIMULINK simulations.

The three-dimensional simulations, calibration, measured currents and calculated relative sensitivities of the first-ever fabricated double-drain gallium nitride (GaN) magnetic field effect transistor (MagFET) are given in this work. The MagFETs are GaN high electron mobility transistors (HEMTs) capable of operating under harsh environments. Geometrical and operational analysis are carried out on MagFETs using commercial simulation software Silvaco. The analysis shows promising relative sensitivities of 6.84%T−1 and 5.04%T−1 at ambient temperatures of 400 K and 500 K, respectively. In addition, the relative sensitivity of fabricated device is improved from 12%T−1 to 24%T−1 at 300 K by optimising device geometrical parameters.

The compact model of an injection enhanced insulated gate bipolar transistors based on the exponential solution of the ambipolar diffusion equation is presented in this paper. To model plasma carrier distribution, an exponential shape function is used, and in steady-state forward bias operation, the plasma carrier concentration has a distribution of catenary form with just two exponential basis functions, while in transient operation, more complex profiles can be approximated using a number of exponential basis functions with a range of decay length parameters, shorter than the steady state ones. The device model developed has been implemented in Saber circuit simulator and successfully tested against complete set of high current, high voltage experimental results.

The DMAIC (Define, Measure, Analyze, Implement and Control) approach is a well-established approach used to improve industrial processes and products. The DMAIC approach is mostly used as part of quality assurance processes, but this paper will show that it can be applied to investigation of root cause analysis of equipment failure. Specifically, the piece of equipment under investigation is a circuit breaker used in a distribution system. A second contribution of this paper is to provide suggestions on approaches to overcome modeling challenges related to lack of data, both in terms of measurements and in terms of system parameters.

This paper presents state of the art split drain current magnetic sensors fully compatible with the bulk silicon and SOI technology. While lateral double-diffused magnetic sensitive MOSFET with integrated n-type Hall plate exhibits an order of magnitude higher relative current magnetic sensitivity compering with the conventional split drain silicon MagFETs, novel dual-gate four-terminal magnetic sensor brings a step-change in the integrated magnetic sensor's performance. Finally, a potential of GaN split current magnetic sensitive HEMT has been discussed.

This paper reports simulation of punch-through gallium nitride (GaN) insulated gate bipolar transistors (IGBTs) with different drift layer thicknesses of t=12 μm and t=17 μm delivering breakdown voltages of V BD =600V and V BD =700V, respectively. The simulated devices are enhancement-mode with the threshold voltage of +3.5V. The leakage current in proposed devices are in the range of 10−9 A/μm. Using GaN, a significant improvement in blocking capability of IGBTs for a given epithickness is observed. The GaN IGBTs have been simulated and analyzed by Silvaco atlas toolbox.

In this paper a novel photovoltaic (PV) emulating scheme for testing maximum power point tracking (MPPT) algorithms and PV inverters has been proposed. It is constructed by the parallel connection of conventional solar panels with a DC power supply operating in current source mode. The advantages of the proposed scheme are cost-effectiveness, relatively easy implementation and indoor testing of MPPT algorithms and power converters avoiding weather and time of day dependency on solar irradiation levels. Furthermore, the proposed PV emulator avoids the bandwidth problem associated with the dc converter based PV emulating systems. Detailed circuit connection, parameters, electrical characteristics and mathematical model of the PV emulator are presented and discussed. Proposed PV emulating system has been used to test a boost DC/DC converter controlled by Perturb & Observe (P&O) MPPT algorithm. Test results confirmed the effectiveness of the proposed PV emulation system and all achieved results correspond well to the original designed values.

The paper aims to review the research area of the IGBT compact modelling and to introduce different device models. The models are separated in two groups, one that solves ambipolar diffusion equation (ADE) and one that does not. Both types of compact models have been successfully used in the past for power electronic circuit design.

The effect of self-heating and polarisation is studied in AlGaN/GaN Transmission Line Measurement (TLM) structures with varying the contact spacing between the source and drain. The measurement results of the I-V characteristics are calibrated and investigated by TCAD Atlas-Silvaco. The self-heating simulations show a hotspot at the vicinity of the drain side. The electrical stress that is applied on the Ohmic contacts decreases the polarisation as the source-drain distance is reduced, causing the inverse piezoelectric effect.

The effects of electric field induced traps generation in the drain access region is studied using industry standard TCAD, Atlas by Silvaco [1]. We show that the reduction in the cut-off frequency of the device from 13.9 (GHz) to 11.25 (GHz) could be linked to the electric field induced traps. We have used acceptor traps at an energy level of E T = E v + 0.9 (eV), corresponding to substitutional carbon in GaN, and a concentration of N IT = 5 ×1017(cm−3) to model the induced traps. Although vertical scaling has been used to reduce short channel effects, we observe that this leads to a reduction in the current arising from the reduced ionised surface donors [2].

This paper explores the use of a dynamic virtual impedance approach to load sharing as an alternative to the more generic droop load sharing scheme for low-voltage islanded micro-grids. The approach is based on taking advantage of the power decoupling capability of the virtual impedance scheme to facilitate appropriate load sharing based on the available generation capacity of distributed generation (DG) units. The proposed scheme identifies the DC operating zone of the photovoltaic inverter-based unit as irradiance levels change and conditions the virtual impedance loop for an efficient load sharing based on available generation. This dynamic approach allows taking into account the rating, output impedance and voltage limits of each unit. The proposed scheme provides energy saving; since energy demand from a local auxiliary generator is reduced. Results from the proposed dynamic scheme were tested on a P-f/Q-V droop, in the presence of virtual impedance, and results were compared with the conventional-static virtual impedance loading method. The proposed method is validated using MATLAB/SIMULINK simulations.

In this paper, load current observer based feedforward control of DC bus voltage for a three-phase active rectifier is presented. The paper focuses on: (1) design of conventional proportional-integral (PI) current controller with fastest possible response, (2) the digital current control loop time-delay compensation, (3) design of dc-link-load current observer for identifying the dc load current without using actual current sensors and (4) the feedforward control of dc bus voltage based on the designed current controller and dc link load current observer. A prototype is developed based on the proposed control scheme and expected results have been achieved.

A standard 0.35 micrometer CMOS technology has been extended for 100 V Power IC applications by accommodating reduced surface field (RESURF) LDMOSFET device with p-well block region or extended poly-overlap region for suppression of the drain wrapping potential. A 100 V integrated H-bridge circuit suitable for driving a brushless DC motor has been designed, manufactured and tested to prove the technology. To streamline the design and integration of this power device 2D and 3D simulations have been performed. Different electrical isolation schemes to provide technology compartmentalization have also been investigated experimentally and results are discussed.

The physics-based IGBT sub-circuit compact model which can successfully include the effects of the localized lifetime control (LLC) on device electrical performance has been described in this paper. The model of non-punch trough IGBTs with different locations of LLC region is developed and its accuracy is verified based on the agreement with the results of two-dimensional numerical simulations of LLC effects in IGBTs. The models have been implemented in SPICE for investigation of the total power losses in the single phase Pulse Width Modulated (PWM) inverters.

In this article, a high-speed electro-thermal (ET) modelling strategy to predict the junction temperature of insulated gate bipolar transistor (IGBT) devices of a three-phase inverter power module is presented. The temperature-dependent power loss characteristics of IGBT and diode devices are measured and stored in lookup tables, which replace the conventional complicated physics-based compact models. An inverter is modelled as a voltage controlled voltage source, which allows the inverter-based power train simulation to be carried out in the continuous time domain with a large simulation time-step (1 ms). Using the simulated sinusoidal voltage and current components of the inverter output, the given pulse width modulation mode, the conduction time (duty ratio) and the current of the devices are extracted. Based on the lookup tables, on-times and conduction currents of devices, the average power loss over each simulation time-step is calculated, which is then fed into the inverter thermal model to predict...