Negative inductance is not a traditional physical componentwhich must be implemented through active circuits. Its equivalent implementation schemes include low-power negative inductance based on operational amplifiers and high-power negative inductance based on power electronic converters. Negative inductance exhibits impedance phase characteristics opposite to those of traditional positive inductance, enabling its use of impedance matching or transformation. Currently, its typical applications are concentrated in the microwave field and power systems, with relatively limited applications in power electronics. The definition, characteristics, stability and distinctions from capacitance of negative inductance are first discussed. Then focuses on introducing the principles and control strategies of existing methods for implementing high-power negative inductance based on inverters, comparing these approaches and validating them through simulations. Subsequently, analyzes the application of negative inductance in active common-mode chokes and in simplifying compensation for wireless power transfer. Finally, discusses critical issues requiring urgent resolution in negative inductance research.

Under abusive conditions, the battery cell will experience severe thermal runaway. The heat is then transmitted to the adjacent battery through thermal radiation and convection, which may cause a battery pack and entire electric vehicle to explode. Therefore, the effects of heat abuse of battery cells under various state of charge (SOC) and heating powers, as well as the effects of various electrical connection modes and battery pack arrangement shapes, are investigated on thermal runaway propagation based on an experimental platform that was created by the user. It is of great significance to the safety protection and early warning of electric vehicles. According to the findings, there are four stages of thermal abuse for single cells: heating, secondary heating, thermal runaway, cooling and extinguishing. The battery voltage drop time, exhaust time, and maximum temperature all exhibit a certain linear relationship as SOC increases. Voltage drop, exhaust, and maximum temperature will all advance as heating power increases. The propagation speed of thermal runaway in different electrical connection modes from fast to slow is parallel, series and non-electrical connection. When the shape factors are triangles and diamonds, the propagation speed is the fastest and close, followed by squares.

Solid oxide fuel cell (SOFC) has attracted increasing attention due to its environmental friendliness, high efficiency and broad prospects. The life prediction of SOFC system is critical to the long-term operation of power generation system. However, the accuracy in predicting the life of SOFC using long short-term memory (LSTM) neural network model with numerous parameters is still insufficient. To address this problem, the orthogonal experiment method was employed to systematically optimize the parameters for LSTM. The results indicate that batch size has the greatest influence on the prediction accuracy and calculation time of LSTM. And the root mean square error (RMSE) of optimized model is reduced by 72.36%、18.18%、35.05% and 32.97% when compared with that of recurrent neural network (RNN), gated recurrent unit (GRU), whale optimization algorithm (WOA) optimizates LSTM (WOA-LSTM) and grey wolf optimization (GWO) optimizates LSTM (GWO-LSTM), respectively. The systematically optimized LSTM model exhibits improved performance, which may mitigate fuel cell degradation with other control strategies.

In order to address the shortcomings of existing detection methods, which are unable to effectively capture the complex nonlinear features of photovolatic time series data, a distributed photovolatic anomaly data detection method considering time series clustering is proposed. This method is designed to facilitate the accurate identification and location of abnormal data. Firstly, the similarity of photovolatic power plant time series data is calculated using the dynamic time warping (DTW) method, and a similarity matrix is constructed. Subsequently, spectral clustering is applied to obtain the feature matrix, and the intrinsic connection between photovolatic data points in the feature matrix is analysed. This enables similar photovolatic data to be clustered into a collection. Then, a joint detection model of sub-isolated forest is established based on the collection of photovolatic data clusters. The abnormal score geatures are obtained using the local outlier factor algorithm. The features are used as inputs to the joint detection model to obtain the anomaly score matrix of photovolatic data for each similar clustering set. Finally, the anomaly score matrix is selected and reconstructed, and the anomalies of photovolatic data are identified and located after decision analysis. The experimental results demonstrate that the method is an effective means of capturing photovolatic anomaly data. The precision and recall rates of extracted anomaly samples reach 0.9, thereby enhancing the accuracy of photovolatic anomaly data identification.

Accurate state of health (SOH) estimation is crucial for the safe operation of lithium-ion batteries. However, the complex cycling conditions and random initial charging state of lithium-ion batteries make the inputs of most existing methods unable to be satisfied. To address this problem, we propose an end-to-end SOH estimation method based on relaxation voltage. This method directly establishes the nonlinear relationship between relaxation voltage and SOH using a Bayesian optimization-assisted deep feedforward neural network (BO-DNN), eliminating the laborious feature extraction process. The performance of the estimation method was verified on three different battery aging datasets. The results show that the model achieves a root mean square error (RMSE) of 0.011 0 on the dataset used for model building. A transfer learning model developed by fine-tuning the basic model only relies on less than 3.5% training data, and it achieves an RMSE of less than 0.016 3 on the datasets used for model validation.

A porous carbon material with high specific surface area was prepared from squid ink as carbon source, and a composite material Fe₃O₄/HPBC of Fe₃O₄ and highly porous biomass carbon (HPBC) was prepared by immersion reduction method. The material was used in lithium-ion batteries. At the same time, the effects of different iron ion content and calcination temperature on the electrochemical properties of the composites were also studied. When the prepared Fe₃O₄/HPBC and carbon composite material is used as the negative electrode of lithium-ion batteries, the specific capacities of initial discharge and charge at 0.2 A/g current density are 1 590.0 and 961.0 mAh/g respectively, the coulombic efficiency is 60%, after 300 cycles the capacity without attenuation. After 400 cycles of 5.0 A/g current density, the specific discharge capacity can still reach 484.0 mAh/g, which is higher than the specific discharge capacity of the prepared carbon material electrode, indicating that the Fe₃O₄/HPBC electrode still has good lithium storage performance and cycle stability under large current densities.

The performance degradation of lithium-ion batteries at low temperature is a main bottleneck that restricts the electric vehicle development in cold regions. Low temperature heating (LTH) is an effective solution. A study is conducted on LTH method based on short-term high-current pulse excitation, and the heating characteristics are analyzed. Electric and thermal co-simulations are conducted and verified by experiments. The results show that the presented method can achieve a heating rate of 10.90 °C/min and consumes 4.35% of nominal capacity for raising 20 °C. The temperature consistency and its influencing factors of single cells during the heating process of battery module is investigated by simulation analysis. The results indicate that the difference of resistance is the main cause of inconsistent temperature rise while the inlet wind speed only has a minor impact. When the standard deviation of internal resistance is 4.3 mΩ, the maximum temperature difference can reach over 6 ℃.

With increasing penetration rates of power electronic devices into grid continuously, the threat of new-type power system oscillations becomes much serious. On this basis, the mechanism of a particular form of oscillation, i.e., power-frequency oscillation (or called as synchronous-frequency resonance, synchronous-frequency oscillation) is focused. First, by comparative studies on different models of grid-forming converter tied systems by considering or ignoring current-control loop and/or line dynamics, it is found that the power-frequency oscillation can appear in the single-loop grid-forming converter tied systems. Therefore, for the occurrence of power-frequency oscillation, converter should show the characteristics of a voltage source, and meanwhile, line dynamics should be considered. Then, the mechanism of the power-frequency oscillation is studied under a simple model of two voltage sources connected by a dynamical line. Finally, by time-domain analyses of current and power waveforms under the abc three-phase stationary and dq two-phase synchronous-rotating-frame coordinates, it is found that the power-frequency oscillation comes from the un-damping (or weak-damping) transient component (or free component), whereas the inherent power-frequency sinusoidal oscillation comes from the force component. In addition, it is found that the power-frequency oscillation does not show the frequency-coupling phenomenon. The occurrence conditions and physical origin of power-frequency oscillation are revealed directly in time domain. Clearly, all these findings could help understand the power-frequency oscillation and even use power electronic devices better.

In order to improve the stability and reliability of energy storage auxiliary thermal power units participating in primary frequency regulation, a control strategy for primary frequency regulation of energy storage auxiliary thermal power units considering ultra-short-term load prediction is proposed. The two-layer control strategy is divided into a load prediction layer and a primary frequency modulation layer. The load prediction layer uses a combination prediction model of graph convolutional network-bidirectional gated recurrent units (GCN-BiGRU) and improved temporal convolutional network (iTCN) to predict ultra-short-term load and obtain frequency modulation demand. The primary frequency modulation layer used an improved dragonfly algorithm to dynamically adjust the unit power of virtual droop control and virtual inertial control based on the frequency modulation requirements, with the goal of minimizing frequency deviation. Finally, a simulation model was built on the MATLAB platform to compare and analyze the frequency deviation data and the state of charge (SOC) curves of different control strategies under the working condition of continuous load perturbation, and verify that the control strategy proposed can effectively improve the effect of frequency modulation and maintain the state of charge to a certain extent.

In the inverter grid-connected system with weak grid, the existence of grid impedance and voltage harmonic perturbation will cause the distortion of grid-connected current, and the traditional voltage full feedforward method reduces the steady-state error but affects the quality of current, and even threatens the instability of the system. To address the above problems, it is proposed to improve the grid voltage feedforward, i.e., to add resonant feedforward in the feedforward channel to offset the influence of grid impedance, and at the same time ensure the system open-loop gain, increase the system margin, and improve the stability; and secondly to introduce a proportional multi-resonance controller instead of the traditional proportional integral controller, which is used to suppress the influence of grid harmonics on the grid-connected currents, and then to lower the harmonic content of the grid-connected currents and improve the quality of grid-connected currents. Finally, the effectiveness of the control strategy is verified by simulation.