In order to improve the accuracy of photovoltaic array fault diagnosis, an improved artificial neural network (ANN) method for photovoltaic array fault detection is proposed. In ANN, a gated convolutional layer, a flat layer and a fully connected layer are added to replace hidden layer, which solves the problem of gradient disappearance easily in existing ANN. Adaptive Softmax is used as the activation function of the output layer of the model to reduce the computational time incurred due to the large data set size. The simulation model of photovoltaic array was established by using MATLAB to obtain the simulation operation data of different states. Meanwhile, the actual unbalanced operation data set of photovoltaic array was obtained from the literature, and the diagnostic performance of the improved ANN was compared with the six machine learning methods mentioned in the literature, showing the diagnostic performance advantages of the proposed method.

In order to address the model performance degradation caused by sample sparsity and class imbalance in multimode fault diagnosis of automatic guided vehicle (AGV) power batteries, a generative adversarial network incorporating gradient penalty and an independent classifier (GPSC-GAN) was proposed. First, the discriminator loss function was optimized through the introduction of Wasserstein distance with gradient penalty, whereby the distribution divergence between generated samples and real battery fault features was constrained to enhance training stability. Second, an independent classifier module was designed to decouple classification and discrimination functions, enabling high-quality multimode fault samples to be generated for constructing a balanced dataset. Experimental results on AGV battery fault datasets demonstrated that the generated samples achieved structural similarity index measure (SSIM) and maximum mean difference (MMD) values of 0.5 and 0.2 respectively, showing 25% to 66% improvements compared with WGAN-GP, ACGAN, SMOTE, ADASYN, Diffusion GAN and StyleGAN-T models. After integrating synthetic samples, the F1-score of the fault diagnosis model was enhanced to 0.938, with precision and recall rates reaching 94.6% and 93.0% respectively. These metrics outperformed six other comparative methods, verifying the proposed method’s efficiency and robustness in complex industrial scenarios.

High-power high-voltage high-frequency transformers (H³T) are typically subjected to high du/dt pulsed voltage inputs. Their unique operational environment presents significant insulation challenges, making them highly susceptible to partial discharge (PD) initiation. The intense electromagnetic interference generated by high du/dt pulsed voltages substantially compromises the accuracy of PD detection. A novel joint detection method integrating ultra-high frequency (UHF) antennas with photomultiplier tubes (PMTs) is proposed, combining UHF’s advantages of high sensitivity and wide bandwidth with PMT’s superior anti-interference capability and rapid response characteristics. Signal fusion is achieved through amplitude correlation analysis by comparing temporal and spectral consistency between discharge signals acquired from UHF and PMT. The research reveals remarkable consistency in spectral steepness S and kurtosis K between the two detection modalities. The fusion method demonstrates average discharge count enhancements of 66.3% compared to PMT alone and 49.7% compared to UHF alone, significantly improving detection integrity. Furthermore, the analysis of PD characteristics under varying rise times demonstrates that shorter rise times induce earlier and more concentrated PD occurrences. Specifically, the average and maximum discharge amplitudes increase by 37.9% and 37.2%, respectively, while discharge repetition rates and cumulative discharge amplitudes show improvements of 37.1% and 22.1%. This study provides innovative methodology and technical support for achieving high-sensitivity, interference-resistant online PD detection under high du/dt pulsed voltage conditions.

With the widespread random access of distributed resources in the new power system, precise measurement of bus current under complex operating conditions has brought new challenges. Tunnel magnetoresistance (TMR) sensors, as the latest generation of magnetoresistance sensors, integrate the advantages of high sensitivity, low power consumption, and small size of traditional magnetic sensors. They have a wide range of applications in magnetic fields detection, current detection, magnetic memory and other fields. Due to the skin effect of bus current at high frequencies, the current is concentrated in the “skin” part of the conductor, and the surrounding magnetic field exhibits nonlinear characteristics that affect the measurement of the sensor. An optimization method for arranging the position and quantity of tunnel magnetoresistance chips around the distribution bus is proposed, based on the physical characteristics of tunnel reluctance sensors and the disturbance law of skin effect. The corresponding sensor structure is designed to reduce the influence of bus skin effect and stray magnetic field on sensor measurement accuracy.

At present, most of the traditional Cauer models assume that the shape of the effective thermal conduction surface of each layer in an insulated gate bipolar transistor (IGBT) module is square or circular, which leads to large errors of the calculation results of IGBT module junction temperature. On this basis, the shape of the effective thermal conduction surface of each layer in the IGBT module are studied, and a Cauer model considering the deformation characteristics of thermal diffusion boundary is proposed to make it more suitable for the actual situation. It is found that a phenomenon of offset occurs at the thermal diffusion boundary of the IGBT module. By capturing the diameter of the selected dynamic thermal diffusion boundary and modifying the parameters of the Cauer model which considerers the deformation and offset characteristics of dynamic thermal diffusion boundary, the junction temperature of the IGBT module can be accurately estimated. The correctness and feasibility of the proposed optimization method for Cauer model consideringthe deformation and offset characteristics of dynamic thermal diffusion boundary were verified by simulation and experimental results.

Aimed at the problem that various characteristics cannot be well compromised in the design of reverse blocking integrated gate commutated thyristor (RB-IGCT) devices, based on the research on the effect of electron irradiation on the device performance, a double-sided proton irradiation method is proposed, which is further verified by simulations. Results show that electron irradiation will significantly shorten the overall lifetime of the device, and greatly decreasing the internal carrier concentration in its on-state. Therefore, electron irradiation has a positive effect on the turn-off characteristics of the device at the cost of deteriorating all of its other characteristics. In contrast, double-sided proton irradiation can locally modulate the carrier concentration distribution in the on-state of the device, thus ensuring its on-state characteristics and surge current withstand capability. Meanwhile, this irradiation method reduces the leakage current of the device at high temperature (400 K) by nearly 50%, indicating that it is an effective method for handling the trade-off relationship between various performance parameters of the device.

By improving the traditional non-isolated Γ-Z source converter, a novel high-frequency isolated anti-Γ source AC-AC converter is proposed, which can be applied to AC transformers, dynamic voltage restorers, etc. This converter uses a high-frequency coupled inductor which acts as a high-frequency transformer, and its size is thus reduced. The addition of an anti-Γ source makes the voltage gain of the converter flexibly adjustable, so that its Buck or Boost voltage range is extended, and the output and input voltages can be maintained in-phase or out-phase. The use of a safe commutation pulse width modulation (PWM) strategy can suppress the voltage spike of a switch tube and reduce the switch loss. Through the analysis of working states of the converter, the relationship between the output voltage and input voltage is calculated for each device, and the performance parameters of the proposed converter are compared with those of other AC converters, indicating the advantages of the novel converter. A simulation model was established by MATLAB/Simulink, and the printed circuit board (PCB) of the converter was drawn for experiments. The proposed converter was discussed and verified based on simulation and experimental results.

A switch-clamped high-gain coupled inductor DC-DC converter is proposed, which co-adjusts the voltage gain by coupling the turns ratio of the coupled inductor and the duty cycle. Therefore, it can realize a very high voltage gain at a low duty cycle. A switch-clamped structure is used to eliminate voltage spikes on power switches, and the energy stored in leakage inductance is recovered through switch capacitors, thus improving the lifespan of the converter. The voltage stress of power switches in this structure is low, and the required duty cycle is small, which can effectively reduce switch losses and improve the efficiency of the converter. A comparison with other converters was performed. In addition, a 200 W low-power prototype was built for experimental analysis, and results verified the feasibility and theoretical correctness of the proposed topology.

A multi-input inverter is capable of converting the direct current (DC) generated by multiple new energy sources into the alternating current (AC) required by power grids, effectively solving the problems caused by single new energy power supply such as low power supply reliability, poor power quality and significant power fluctuations. Based on the research status both at home and abroad, the topologies of multi-input inverters are classified into three categories according to the types of DC-DC converters and stages of power conversion, i.e., two-stage single-input DC-DC converters, two-stage multi-input DC-DC converters and single-stage multi-input inverters. A detailed discussion on various types of multi-input inverters is carried out, including their advantages and disadvantages, as well as the corresponding energy management under different application scenarios. Furthermore, a comparison of circuit complexity, control difficulty, cost and conversion efficiency among the multi-input inverters with different topologies is performed, providing a reference for further research and development in this field.

To meet the requirement of lightweight design for modular multilevel converters (MMCs), the unified structure of arm-multiplexing-type MMCs based on a single-arm multiplexing MMC (SAM-MMC) is summarized. Then, a dual-arm multiplexing MMC (DAM-MMC) is proposed, and the corresponding structure of arm selection switch, modulation strategy and low-voltage switching strategy in multiplexing mode are designed. The DAM-MMC requires a total of 4N in three phases, with a submodule utilization rate of 75%. Through the analysis of arm energy fluctuations, the capacitance in multiplexing arms of the DAM-MMC is reduced by about 31.9% compared with that of the SAM-MMC, and the capacitance in upper and lower bridge arms is reduced by about 11.5%, which further reduces the volume, weight and cost of the converter. Based on an experimental prototype, the feasibility of the DAM-MMC topology, parameter design and proposed strategies was verified.