Recently, four-level inverters for low and medium voltage applications have been widely investigated thanks to their simple structure, low dv/dt, low switching loss, etc. However, the existing topologies have the disadvantage of unbalanced power loss distribution. For this reason, this article proposes an improved four-level active neutral point clamped inverter topology which is featured with low total voltage stress and uniform loss distribution under multiple operation conditions. These merits can fully utilize all power devices and expand the output capacity of the inverter. The operating principle and modulation strategy of this topology are described in detail and a comprehensive comparison with other four-level and three-level inverter topologies is analyzed. Finally, experimental results have been carried out to validate the proposed topology and its advantages in improving loss distribution.

Hydrophobicity degradation is a critical factor leading to pollution flashovers in composite insulator systems. The International Electrotechnical Commission (IEC) 62073 visual inspection method suffers from ±1 level measurement uncertainty, while existing deep learning methods lack physical interpretability for maintenance decision-making. This paper presents a hierarchical framework employing a physically grounded intermediate representation. A five-level droplet taxonomy (C1-C5) based on contact angle physics decouples droplet morphology from surface assessment. An enhanced Faster Region-based Convolutional Neural Network (Faster R-CNN) with Squeeze-and-Excitation Residual Network-50 (SE-ResNet-50) backbone achieves 75.8% mean Average Precision on 16,535 labeled droplets. The fusion of 21 detection-derived and five image-processing features yields a 4.08% improvement over the naive combination. Evaluated on 1,160 independent images spanning four climate zones and 2-13 years of service life, the method achieves 77.09% Full Hit Accuracy (FHA, exact HC level match) and 99.13% Tolerant Accuracy (TA, predictions within ±1 level), compliant with IEC 62073 uncertainty tolerance. The hierarchical design enables verification of classification reasoning through inspectable C1-C5 intermediate outputs.

Severe corrosion and blockage in copper heat exchangers of generator stators from a nuclear power plant motivated this study. The accumulation of corrosion products within narrow cooling channels led to flow restriction and heat transfer degradation, posing risks to the safe and efficient operation of water-cooled generators. To address this practical thermal engineering problem, a case-based investigation was conducted using real corroded components dismantled from the site. Dynamic simulation experiments were designed to reproduce actual cleaning conditions and to examine the effectiveness of ethylenediaminetetraacetic acid (EDTA)-based chemical cleaning. The dissolution behavior of copper corrosion products was systematically evaluated with respect to key operational parameters, including H2O2 addition, temperature, EDTA concentration, and solution pH. The results identified optimal cleaning conditions as no H2O2 addition, a temperature of 40 degrees C, an EDTA concentration of 0.3 g/L, and operation at the natural equilibrium pH. To overcome the decline in cleaning efficiency during operation, a constant-EDTA cleaning strategy was further developed using an EDTA-type anion exchange resin for in situ regeneration of EDTA via continuous removal of Cu-EDTA complexes. Under these conditions, complete removal of the corrosion layer was achieved within 2.5 h. This case study demonstrates an effective and gentle cleaning solution for corrosion-induced performance degradation in heat exchangers. The proposed strategy provides a practical, field-applicable approach with advantages including minimal substrate corrosion, stable cleaning driving force, and simplified operation. The findings offer transferable insights for the maintenance and performance recovery of heat exchangers and similar thermal systems affected by fouling or deposition.

Four-level neutral-point-clamped (4L-NPC) inverters are widely used in low- and medium-voltage applications. However, existing variable-reference PWM and carrier-overlapped PWM methods often lead to increased switching losses and uneven loss distribution among power devices. To address this issue, this paper proposes a unified variable-reference discontinuous PWM (VRDPWM) framework that enables two clamping modes: full-clamping VRDPWM-I method and partial-clamping VRDPWM-II method. In the full-clamping method, the phase leg is clamped to the DC-link voltage levels during the discontinuous interval, thereby reducing the number of switching transitions. In contrast, the partial-clamping method selectively suppresses the switching activity of the inner switch pair to redistribute switching losses by exploiting the switching redundancies of the 4L-NPC topology. Furthermore, the influence of different clamping states on capacitor voltage dynamics is analytically investigated, and a compatible capacitor voltage balancing method is developed. Experimental and simulation results verify that the proposed VRDPWM-II method effectively reduces the switching loss of the inner switch pair while maintaining high output current quality and stable DC-link capacitor voltages.1.

In modern military applications such as directional energy and electromagnetic emission, lithium-ion batteries (LIBs) need to be operated under ultrahigh-rate discharge conditions, where the temperature difference between the core and the surface of cell exceed tens of degrees celsius. Such thermal gradients pose great difficulties for the conventional thermal management techniques that rely on cell surface temperature, leading to severe underestimation of the maximum temperature and non-negligible control delay in thermal regulation, thereby increasing the risk of overheating. To address this critical challenge, this study proposes a self-adaptive multimagnitude physics-informed neural network (SA-MMPINN) framework for internal temperature sensing. This framework integrates thermal model into neural networks architectures, enabling high-accuracy sense of internal temperature based solely on surface temperature while simultaneously identifying convective heat transfer coefficients to adapt to different heat dissipation conditions. Experimental validation using instrumented cells with embedded thermocouples shows that the framework can achieve root mean square error below 0.76 degrees C for core temperature sensing under 20C discharge conditions with temperature ranging from -5 degrees C to 35 degrees C. Meanwhile, the convective heat transfer coefficient can be identified with over 97% accuracy. The proposed SA-MMPINN framework enables accurate sensing for the internal temperature evolution of LIBs, which is crucial for informing the design of next-generation intelligent thermal management systems.

With the advancement of new power system construction, switching operations for substation equipment are gradually adopting the one-touch sequential control operation mode. Among them, the dual-confirmation mode for disconnector sequential control based on video images utilizes on-site cameras to capture disconnector images for recognition and confirmation. However, the images are often affected by adverse shooting weather conditions, especially under rainy or backlit conditions, where key targets in the image are obscured, leading to unrecognizable targets and severely impacting the efficiency of switching operations. To address this, this paper employs a Vision-Language Large Model (VLLM) to achieve image restoration for substation equipment sequential control under inclement weather. Through comparative studies of different models and prompt words, the results demonstrate the effectiveness of the proposed method.

Four-level inverters can output multiple voltage levels, which can mitigate the negative impacts of dv/dt. However, high-frequency switching of the power device is a primary cause of dv/dt. Periodic switching frequency pulse width modulation (PSF-PWM) is widely used in power converters to reduce electromagnetic interference. In this study, a PSF-PWM method is introduced for four-level inverters. A triangular carrier wave generates a switching frequency sequence based on three periodic functions, while the central switching frequency remains consistent with that of conventional PWM methods. The fundamental principles behind the harmonic dispersion effect of the PSF-PWM method are analyzed by deriving an expression for the power spectral density. An experimental evaluation of the proposed method is conducted with a simulation and real measurements of the operation of a four-level active neutral point clamped (4L-ANPC) inverter as an example. The results show that the PWM harmonics near the switching frequency and their integer multiples are dispersed to the adjacent frequency bands. Thus, the device exhibits reduced electromagnetic interference compared to conventional PWM techniques in four-level ANPC inverters.

The streamer discharge is the earliest discharge stage under impulse voltage and usually carried out in stationary air in the previous studies. However, the breakdown discharge characteristics of high-speed moving objects or those exposed to strong winds differ from those of traditional stationary targets, and the streamer discharge is important for subsequent discharge process. Therefore, it is significant to study the effect of air flow on the initial characteristics of streamer discharges. In this study, positive streamers under impulse voltage were investigated using a 2.5 mm diameter rod electrode in a wind tunnel, with a maximum airflow speed of 100 m/s. The tests were carried out in a darkroom, where discharge voltage and current were measured, and streamer morphology was captured. The experimental results indicate that regardless of the airflow speed, the initiation time and initiation voltage of the streamer decreases with increasing voltage rise rate. As the airflow speed increases, the initiation time of the streamer also decreases, accompanied by a reduction in dispersion. The streamer initiation voltage follows a similar trend to the initiation time, but with less consistency. From the discharge images, the streamer stem and its branches form a conical distribution, and increasing airflow causes the streamer stem to shorten and narrow. The injected charge of the streamer depends quadratically on the inception voltage; while the airflow alters the operating point along this relationship, the functional form of the dependence remains unchanged. Finally, the concept of critical volume was used to explain the experimental results.

Although SHAP and other explainable artificial intelligence (XAI) techniques have been applied in several power system scenarios, lightning protection strategies in transmission lines still rely heavily on the practical experience of operation and maintenance personnel. To address this gap, this study proposes a quantitative diagnostic framework based on SHapley Additive exPlanations (SHAP) to identify the root causes of high lightning risks. To facilitate the extensive computations required by SHAP, a dual-hidden-layer neural network is integrated as a surrogate model to efficiently approximate lightning trip rates. This framework is then applied to analyze the specific risk factors of transmission towers along a line in Yunnan Province, China. The results obtained using the SHAP method were also used to formulate targeted lightning protection measures for these towers, thereby effectively reducing their risks to a reasonable level. The quantitative weights derived using the SHAP method aligned well with theoretical principles of lightning trip rate calculations, validating the rationality and effectiveness of this method in diagnosing causes of lightning risks for transmission lines.

For high-speed moving objects such as aircraft and wind turbine blades, discharge occurs under strong airflow. Research on lightning attachment characteristics has been extensive, whereas the role of corona discharge in pre-breakdown streamer development has received less attention. This paper presents wind tunnel experiments on positive corona discharge in a needle-to-plate configuration under DC voltage, with airflow ranging from 0 to 50 m/s. The results show that airflow induces significant deviations in the streamer channel, with higher airflow velocities causing the corona to tilt further upwind. Additionally, airflow affects both the discharge intensity and the dispersion of space charge. This study reveals the displacement characteristics of streamer-corona discharge and discusses the influence mechanism of airflow on space charge distribution.

Rotating wind turbines are more vulnerable to lightning strikes than static wind turbine, possibly due to the different distribution of tip corona charges. However, there is a lack of research on the characteristics of blade tip corona charge distribution. This paper establishes a simulation model of corona charge distribution of rotating wind turbine. The long gap discharge test of the scaled wind turbine is carried out to verify the effectiveness of the simulation model. The simulation results show that the corona current increases with the increase of wind turbine speed, which is consistent with the test results. When the wind turbine rotates, the corona charge will migrate to the arc trajectory through which the blade rotates, which greatly reduces the charge density accumulated at the blade tip. Compared with the static state, the electric field intensity of the blade tip is higher when the wind turbine rotates, which will make it easier for the blade tip to initiate the upward leader. Moreover, the influence of different factors on the corona charge distribution of rotating wind turbine is calculated, and the function relationship between the maximum electric field intensity of the wind turbine tip and the rotation speed is established.

This study identifies a new failure mechanism in glass insulators caused by a manufacturing defect: a missing internal thread structure. This defect results in a “mechanical–electrical decoupled failure,” where electrical performance remains intact while mechanical strength is severely reduced. Analysis of a failed insulator from a 220 kV transmission line, together with laboratory tests on 17 units, showed that insulators without threads pass standard electrical tests but retain less than 10% of their strength after failure, compared to over 75% for normal units. Finite element simulations confirmed that electric field distributions stay within safe limits, supporting the decoupling concept. The presence of multiple defective units in the same production batch indicates a quality control problem. Current acceptance tests are unable to detect this hidden defect. Mechanical tests further revealed an unusual failure mode where the steel pin pulls out, rather than the glass breaking. This defect cannot be detected by current factory acceptance procedures, indicating a need to revise mechanical verification protocols and implement enhanced quality control measures for batch-produced glass insulators.

ABSTRACT Multi‐chamber arc‐extinguishing devices have a simple structure and excellent arc‐extinguishing performance; thus, they have attracted considerable attention as lightning protection devices. However, the structural parameters of different models of the aforementioned devices vary, leading to inconsistent protection levels; therefore, the arc‐extinguishing characteristics of their chambers require further investigation. This study employed simulation and experimental methods to analyze the process through which devices with a semi‐closed short‐gap structure extinguish alternating‐current (AC) power frequency (50 Hz) arcs. First, a numerical model of AC arcs was established on the basis of magnetohydrodynamics theory using COMSOL Multiphysics software. This model was then used to simulate the dynamic arc plasma velocity, pressure and temperature within the extinguishing chamber of a device with a semi‐closed short‐gap structure. Subsequently, an experimental platform was constructed to conduct experiments involving extinguishing various magnitudes of continuous AC arcs in a fabricated semi‐closed short‐gap structure. The experimental and simulation results agreed with each other, demonstrating the effectiveness of the developed model. These results indicate that as the magnitude of an AC arc increases, the arc extinction time increases but remains within the time corresponding to half a cycle of the AC waveform.

The reliability of the generator circuit breaker (GCB) switching coil affects the safe and stable operation of the power system, in which the faults of abnormal voltage, poor contact, and mechanical jamming of the switching coil can easily lead to the refusal of the circuit breaker, which threatens the safety of the power grid. In order to study the fault characteristics of the GCB switching coil, this paper combines multi-physical field simulation and experimental testing, establishes the electromagnetic field simulation model of the switching coil, and analyzes the characteristics of current waveforms under typical faults such as voltage abnormality, poor contact, and core jamming. Through simulation and testing to verify the mechanism of current waveform distortion under different fault states, demonstrated the change rule of characteristic parameters when the fault occurs, and provided a basis for the diagnosis of the operation status of the switching coil based on current waveform.

This paper proposes a novel permanent magnet traction machine with a non-uniform air-gap rotor topology specifically designed for high-speed rail applications exceeding 400 km/h. The innovative rotor configuration could effectively suppress air-gap magnetic field distortion during high-speed operation, thereby reducing harmonic components in the magnetic field and minimizing core losses while enhancing overall efficiency. Firstly, the basic topology of the proposed non-uniform air-gap rotor is studied and the main design parameters of the traction machine used for high-speed rail train with speed over than 400km/h are analyzed. Secondly, the influence of main design parameters of the proposed rotor with non-uniform air gap on the key electromagnetic characteristics such as no-load characteristics, rated operating characteristics, losses, and efficiency is analyzed. Then, the no-load back electromotive force (EMF), rated torque, losses, efficiency and inductance of the proposed traction machine with non-uniform air gap are comprehensively analyzed based on the optimized structure. Finally, experimental validation through prototype testing confirms that the proposed non-uniform air-gap permanent magnet traction machine could successfully meet the operational requirements of high-speed rail train exceeding 400 km/h.

The problem of insulator pollution flashover is a serious threat to the safe operation of the power system, and insulators coated with RTV paint are widely used in the power system because of their superior pollution resistance. In order to study the influence of the distribution of water droplets on the surface of RTV glass insulators on the pollution flashover voltage and resistance, the article obtained different water droplet distributions by preparing samples with different hydrophobicity, saturated with moisture. The experimental results show that, under the condition that the surface water droplets are all independent of the water droplet distributions, the flashover voltage of the polluted sample decreases by about 30% compared with that of the unpolluted sample, and the resistance decreases by about 92.4%. Under the same degree of pollution, the flashover voltage decreased by about 13%, and resistance decreased by about 88.1% when the water droplets on the surface were distributed as a water film compared with when the surface was distributed as independent water droplets after being damped.

In this paper, the cloud-to-ground (CG) lightning data and topography data in a region of China are collected and processed respectively, and the variation trend of CG lightning density and CG lightning strength with single topography factor is quantitatively analyzed. Then the Apriori association rule mining algorithm is improved to identify the typical terrain scene of lightning activity, and the operation efficiency is improved. It is found that the confidence of the association rules between CG lightning strength and topography is up to 75.07 %, which shows obvious correlation. Finally, three typical terrain feature scenarios of strong lightning current are identified and summarized, including 1) high-altitude forest areas (>605 m) on gentle slopes (3.19-10.04 degrees), 2) high-altitude forests (>605 m) with south, southwest, west, or northwest slope orientations, and 3) elevated forested zones (>605 m) on windward slopes near valleys. Their causes are summarized, related to the high altitude atmospheric pressure reduction, the towering objects in flat and open areas triggering lightning discharges, and the local airflow in the valley. Through the research of this paper, the distribution law of lightning current and other parameters can be mastered, which can provide some reference for reducing the lightning risk of power system.

The ionic conductivity of stator cooling water is a key parameter influencing the safety and efficiency of watercooled generators. However, the widely adopted upper limit of 2.0 mu S/cm lacks a unified scientific basis and has been shown to impose unnecessary restrictions in practical applications. This study addresses this gap by systematically evaluating the feasibility of increasing the threshold to 5.0 mu S/cm through thermodynamic modeling and performance comparison. Results demonstrate that the higher limit does not compromise the safety of grounding insulation and significantly improves resistance to copper conductor corrosion and plugging. Moreover, it enhances the flexibility of water chemistry control by expanding the adjustment ranges of alkalizing agent dosage, ionic conductivity, and pH by 171.0 %, 172.7 %, and 43.8 %, respectively, under standard operating conditions. Under CO2 inleakage saturation, the 2.0 mu S/cm threshold severely limits these ranges by 38.6 %, 41.8 %, and 34.8 %, while the 5.0 mu S/cm threshold leads to much smaller reductions of 13.3 %, 14.9 %, and 14.8 %. This work establishes the first quantitative framework for evaluating ionic conductivity limits in generator stator cooling systems and provides practical guidance for optimizing water chemistry management. The proposed strategy holds broad potential for improving system reliability and simplifying field operations in the power generation industry.

Due to the complexity of lightning formation and development process, the lightning proximity warning based on single data is difficult to meet the requirements in terms of prediction accuracy. In this paper, we conduct a lightning activity warning study based on lightning localization data and meteorological data. The combination of ERA5 dataset and lightning localization dataset is used as the model training and testing data, and the format unification of 1 km spatial resolution and 10 min temporal resolution is achieved with the help of cubic spline interpolation. On this basis, this paper compares the good and bad prediction performance of decision tree, support vector machine and bp neural network models for the target dataset, respectively, in which the support vector machine model performs better in terms of probability of detection (POD), while the bp neural network performs better in terms of false alarm rate (FAR).