The performance of a data-driven fault diagnosis method deteriorates due to insufficient number and unbalanced classes of inverter fault data in the case of weakly supervised learning in few-shot scenarios. To overcome the defect of traditional data augmentation methods which do not consider the data distribution characteristics, an inverter fault diagnosis method based on data augmentation with specific coefficients wavelet reconstruction was proposed. First, the raw small sample data was decomposed by wavelet packet, and a specific coefficient was obtained through calculation. After random fine-tuning, the sample was directly expanded by wavelet packet reconstruction. The process was repeated to obtain a sufficient and class-balanced training set. Then, convolutional neural networks were used to achieve “end-to-end” fault feature extraction and classification. Finally, experimental results show that the proposed method can effectively expand the fault data in the case of weakly supervised learning in few-shot scenarios and improve the fault diagnostic accuracy. Therefore, it can be used as data preprocessing in few-shot fault diagnosis of inverters.

The air discharge plasma technology has become an important research direction of plasma medicine. In general, the plasma discharge power supply takes the mains power or a large-scale regulated power supply as its input. This kind of power supply is too large in size to meet the application conditions of portable, wearable and attachable plasma generators, which limits the practical applications of the plasma technology in biomedical field. Under this research background, a plasma power supply with a piezoelectric ceramic transformer as its core was designed and studied to meet the requirements of miniaturization and portability for power supply. The size of this power supply was 50 mm×15 mm×8 mm, and its weight was 12.1 g. The power drive circuit adopted a chip-driven oscillating half-bridge circuit, and its main circuit included a step-up Boost circuit and a step-down Buck circuit. The results of no-load and on-load discharge tests show that when the power supply was supplied with 3.7 V power input, the no-load discharge voltage can be adjusted from 200 V to 2.2 kV, the output frequency was 52 kHz with fluctuation of 1 kHz, the maximum output power was 7 W, and the power weight ratio was 580 W/kg, which was better than the common plasma power supplies on the market. Through experimental tests, it is found that the application of this power supply can make SDBD and FE-DBD plasma generators discharge to produce cold plasma. Its miniaturization and portability are of significance to the application of wearable and attachable plasma generators.

The topology of a three-level active neutral point clamped (3L-ANPC) converter with hybrid devices of SiC-MOSFET and Si-IGBT was taken as the research object to study the common-mode electromagnetic interference (CM EMI) characteristics in three pulse width modulation modes. First, two combination modes of SiC-MOSFET and Si-IGBT and three modulation modes of the 3L-ANPC converter topology were introduced. Then, the propagation mechanism of CM noise in the 3L-ANPC converter was analyzed. Based on the analysis, an EMI model including parasitic parameters and noise sources was established, and the impedance expression of propagation path in each modulation mode was derived, thereby obtaining the impedance characteristic curves of CM propagation paths. Finally, the conclusion about the CM noise of the 3L-ANPC converter with hybrid devices in three modulation modes was experimentally verified.

As a typical isolation topology, the push-pull converter has been widely applied in high-reliability scenarios because the transformer is bidirectionally magnetized and the input does not have the risk of bridge arm straight through. With an increase in the converter's operating frequency, the influence of common-mode interference becomes obvious. Therefore, the common-mode noise modeling analysis of the converter has important theoretical and application value for its electromagnetic compatibility optimization design. For a four-winding transformer with a central tap on both the primary and secondary sides, its two-capacitance parasitic parameter model is established in this paper. On this basis, a common-mode noise model in the frequency domain of the push-pull converter is established. Through the analysis of the common-mode noise source and the conduction path of common-mode noise, the prediction of common-mode noise is realized. To enhance the practicability of the model, the extraction method for parasitic capacitance parameters of the transformer is discussed for the simplified transformer model, and a common-mode noise suppression method based on a balancing capacitor is proposed. The accuracy of the established common-mode noise model and the effectiveness of the common-mode noise suppression method were verified by experimental results.

A passive electromagnetic interference (EMI) filter can effectively suppress the conducted EMI of an on-board DC-DC converter at the cost of increasing the volume and cost of equipment, which runs counter to the original intention of high-power density design of electronic equipment. The suppression of conducted EMI by the combination of spread spectrum modulation and the filter can effectively reduce EMI and the filter volume as well. First, the method and process of EMI filter design are studied. Second, the principle of reducing EMI by spread spectrum modulation is analyzed. Combined with the calculation method for filter cut-off frequency, the optimal spread spectrum parameters are calculated, which increases the cut-off frequency in the design of a common-mode choke coil and effectively reduces the filter size. Finally, experimental results show that the spread spectrum modulation combined with the filter has a good suppression effect on the EMI of the on-board DC-DC converter, and the volumes of a compact common-mode choke coil and a filter after the spread spectrum modulation are reduced by 63% and 44% respectively compared with those of a common-mode choke coil and a filter used in the fixed frequency modulation.

The design of fractional-order PID control of a Boost converter is studied, and the fractional-order inductor and capacitor models of the system are fitted using the Oustaloup filter approximation algorithm. To solve the problems of insufficient learning capability and weak iterative convergence when a particle swarm optimization (PSO) algorithm is used to adjust the parameters of the fractional-order PID controller, an improved PSO algorithm is proposed, which introduces three strategies of adaptive inertia weight, adaptive learning factor and weighted mutation to improve the diversity of particles and enhance the convergence speed and accuracy. The improved PSO algorithm is applied to the design of a PID control system for a fractional-order Boost converter. Simulation results show that the control system designed using the improved PSO algorithm has a faster dynamic response of output voltage and inductance current, a better anti-interference capability of output voltage and a better tracking and adjusting capability of inductance current when the load changes suddenly.

The actual duty cycle of a single-stage Buck converter is affected by the switching speed and dead time, so it is difficult to meet the wide range of voltage regulation applications. On this basis, a multilevel cascade Buck topology is adopted, and an output voltage cooperative control strategy for multilevel cascade converters is proposed. This control strategy takes into account the operation efficiency, and it adopts a hybrid modulation method combining pulse width modulation and frequency conversion modulation, thus realizing the output voltage control switching at all levels of the converter and covering the full range of output voltage. A switching loss model is established, and the efficiency optimization under hybrid modulation is analyzed. Finally, a simulation model was established to verify the correctness of the modulation and control design, and an 8 kW two-stage prototype with 50–75 V input and 1–40 V/200 A output was built to verify the feasibility of the proposed control strategy.

The current inner loop of a peak current controlled Boost DC-DC converter needs to detect the IGBT collector current. However, the existing current detection methods need to connect the components to the main circuit of IGBT, which increases the circuit volume and cost and reduces the efficiency of the main circuit. Based on the IGBT smart driver technology with a collector current detection function, a peak current controlled Boost DC-DC converter without current sensors is proposed. First, a smart drive design is given. Based on the relationship between collector current and gate current during the Miller plateau, a three-point method is used to determine the relationship between collector current and gate current during the Miller plateau of the IGBT device, and the collector current is detected by detecting the gate current. Second, a small signal model of leading edge modulated Boost DC-DC converter considering the effects of sample and hold, quantization error and delay is established, and the stability of the converter is analyzed. Finally, a sensorless Boost DC-DC converter prototype based on current detection on the driver side was built, and the effects of current detection and stability control were verified by experimental results.

In view of the problem that the traditional LLC resonant converters cannot well meet the requirements of wide voltage gain due to their narrow voltage gain range, a DC-DC converter with a wide input voltage range based on a multi-resonant structure is proposed, which achieves a wide voltage gain in both the under-resonant and over-resonant frequency ranges. The topology of the proposed resonant converter and its working principle are introduced, its fundamental wave equivalent mathematical model is established, and the voltage gain and input impedance characteristics are analyzed in detail. On this basis, the influences of each resonance parameter on the voltage gain, input impedance and working efficiency are analyzed, and the parameter design process of a 500 W converter is given. Finally, an experimental platform with rated power of 500 W was built. Experimental results show that the proposed converter not only had a wide voltage gain in both the under-resonance and over-resonance frequency ranges, but also can transmit the fundamental wave and third-order harmonic power, thus improving its working efficiency. In addition, this converter had good soft-start and short-circuit current limiting capabilities.

As an energy manager for spacecraft power supply systems, the combined non-isolated three-port converter (TPC) can transfer and transform the electric energy between the three power ports of solar photovoltaic, battery and load, showing advantages of high-power density, high conversion efficiency, easy system construction and easy on-orbit maintenance. The combined non-isolated TPC is stacked with an interleaved bidirectional Buck/Boost converter and a semi-active full-bridge rectifier to realize the single-stage power conversion between any two ports, and the independent power control of the two ports is achieved by the pulse width modulation and phase-shift modulation strategies. The working principle, modulation strategy and control strategy for TPC were analyzed in detail, the corresponding design considerations were proposed, and its performance was further evaluated by experimental results.