Low frequency transmission system is a new type of AC transmission technology that combines traditional high voltage AC transmission and high voltage DC transmission, and adopts low frequency below power frequency. The core power converter of the system, the modular multilevel matrix converter (M3C), is a new AC/AC converter topology suitable for high voltage and high power. The fault-tolerant control of M3C is of great significance for improving the reliability of M3C system, but there are still problems with its fault-tolerant control strategy at present. The fault-tolerant control strategy with an optimal common mode voltage (CMV) injection method is proposed in this paper, which is able to maximize the input and output voltage ranges of the M3C after the failure of the non-redundant submodules. By calculating the relationship between the fundamental reference voltage and the maximum available voltage of each bridge arm branch in real time, the method can dynamically adjust the optimal CMV for injection to adapt to different fault situations. Finally, experiments are conducted in both MATLAB/Simulink and RT-LAB platforms to verify the correctness and effectiveness of the method proposed in this paper under single and multiple bridge arm branches fault conditions.

A strong electromagnetic pulse (EMP) has a short rise time and high peak field strength, propagating to electrical equipment through strong coupling with cables, thereby causing damage. To investigate the coupling effect of strong EMPs on primary equipment, a conductive EMP environment is established, and equivalent circuit models for zinc oxide surge arresters, glass insulators, and distribution transformers are constructed. Through simulation, the port voltages and currents of each device under strong EMP conditions are obtained. The results indicate that when the terminal voltage waveform generated by the coupling of strong EMPs with overhead lines enters typical primary equipment, the voltages and currents of the surge arrester, insulator, and transformer remain as pulse waveforms when a protective resistor is connected at the port. However, the leading edge becomes slower, and the pulse width increases. Without a protective resistor, streamers form on the surface of the insulator string after flashover, causing the equivalent impedance to shift from capacitive to inductive, leading to sustained oscillations in the ground current. During flashover, the voltage and current amplitudes rapidly decrease. The transformer′s port voltage and ground current without a protective resistor both exhibit oscillatory waveforms, gradually decaying to zero.

Power capacitor is one of the main noise sources in high-voltage direct current converter stations. The coherent sound field of power capacitors exhibit significant directivity in the far field, and directional correction must be applied when calculating the noise from capacitor towers. This study investigates the methods for obtaining directional parameters of sound sources, defining directivity, and reconstructing three-dimensional directivity in the simulation of capacitor towers. The proposed method, along with the Cadna/A software, is used to conduct comparative simulation calculations on an AC filter capacitor tower in a converter station. The results demonstrate that this method effectively addresses the issue of directional correction in capacitor tower noise calculations.

Under the backdrop of the digital economy, the data center industry has rapidly developed, with its energy consumption accounting for a significant proportion of societal electricity usage. Data centers have become a crucial component of the power distribution network. To fully exploit the flexibility resources of data centers, a data load scheduling model is established based on the spatiotemporal adjustable characteristics of data loads, which is then integrated into integrated energy stations to achieve efficient energy utilization. An in-depth analysis is conducted on the complex relationship between data load processing capacity and electricity consumption during data center operations. Dynamic voltage and frequency scaling （DVFS） technology is employed to model the data centers, and demand models for different types of data loads are developed based on a mixed queuing theory model. To balance data load fluctuations and improve overall efficiency and power consumption ratios, a data center alliance model for shared data loads is proposed, and a data load scheduling model under the alliance is established. Subsequently, with the goal of minimizing comprehensive costs, a multi-regional energy station coordinated scheduling model is developed, considering the complementary characteristics of multiple energy sources and spatiotemporal adjustability. Finally, simulation analysis is conducted using energy stations integrated with data centers to analyze the positive effects of multi-energy complementarity and spatiotemporal adjustability on the economic and environmental performance of data center operations, validating the economic benefits of the proposed scheduling strategy.

In response to the current practical demand of timely and high-frequency data updates of energy consumption and carbon emissions in relevant government departments and enterprises, based on power data as the core and integrating multiple sources of data, a ridge⁃multivariable linear regression model for energy consumption calculation is proposed, by comparing and analyzing the performance of various models under various fitting methodologies. Based on this, a model for carbon emission monitoring is proposed. The superposition principle of linear models is applied to achieve the functions of annual data modeling and monthly data prediction, solving the key difficulty of modelling with inconsistent frequency of statistical data and avoiding the errors introduced by traditional data splitting methods at the data source. The application of the ridge regression weighting method solves the multicollinearity problem of the input data. The proposed model has a wide range of applicability and realizes the functions of monitoring and predicting energy and carbon emissions at different spatial scales such as regions and industries and at different time scales such as annual, quarterly and monthly. It has the effect of "solving complexity with simplicity" and fills the gap of high-frequency statistics data in energy and carbon emissions and provides theoretical basis and data support for relevant government departments, industry associations, etc. to monitor energy consumption and carbon emissions and formulate relevant policies.

As the installed capacity of grid-following renewable power grows significantly, the risk of system oscillations is increasing. Compared to the past, where large-scale renewable power bases are sent out centrally leading to the instability of the farm-network mode of action, the decentralized integration of large-scale new energy into the receiving grid at multiple points may result in more pronounced interactions between different wind farms. Based on characteristic pattern analysis, the dominant oscillation mode of multi wind farms decentralized access multi region weak AC interconnection system is determined in both field- network interaction mode and inter fields interaction mode. Then, the impact of wind power control parameters and system operation mode on the stability of the dominant oscillation mode of the system is studied. Analysis shows that the control bandwidth of each wind farm tends to be synchronized and the coupling effect between wind farms is more obvious. On the other hand, the differences in electrical distance and operating conditions between wind farms can affect the interaction between them. Detailed electromagnetic transient simulations based on PSCAD/EMTDC verifies the correctness of the analytical conclusions. The research work has certain guiding significance for understanding the control interactions among multiple wind farms, guiding the site planning of wind farms in complex systems and proposing oscillation suppression strategies.

With the continuous expansion of modern power systems and the increasing complexity of their structures and operational modes, real-time and accurate state estimation of power systems is crucial. To address this, a power system state estimation method based on dual attention convolutional bidirectional long and short-term memory neural networks (DA-CNN-BiLSTM) is proposed. This model introduces a channel attention mechanism to adaptively adjust the weights of feature channels in convolutional neural networks (CNN), and incorporates a feature attention mechanism to dynamically allocate weights for individual features before inputing them into the bidirectional long short-term memory neural network (BiLSTM). Important features are filtered from spatiotemporal characteristics based on acquired importance metrics and the correlation between measurements and state variables is dynamically explored. By constructing a measurement dataset from historical data, the DA-CNN-BiLSTM state estimation model is established. Real-time measurement data is then fed into this model to obtain real-time state estimation results. Case studies on IEEE standard systems demonstrate that compared to WLS, SE-CNN, and FA-BiLSTM state estimation methods, the proposed method achieves superior estimation accuracy, robustness, and computational efficiency.

Wind power generation is significantly influenced by environmental factors, exhibiting high randomness and volatility, making it difficult to accurately predict the amount of electricity generated. To address this, an ultra-short-term wind power forecasting model combining multi-head self-attention, convolutional neural network, and long short-term memory (AM-CNN-LSTM) is proposed. The multi-head self-attention mechanism assigns weights based on the varying importance of historical wind power and meteorological data, generating key feature representations and mitigating interference. The convolutional neural network extracts complex spatial patterns and local dependencies between wind power and meteorological factors. The long short-term memory network captures long-term dependencies and dynamic changes in time series, enhancing forecasting accuracy. The results demonstrate that under the optimal time window of T=4 h, the model achieves mean absolute error, root mean square error, and mean absolute percentage error of 13.45 %, 19.48 % and 3.78 %, respectively, It outperforms comparative methods such as LSTM, CNN and CNN-LSTM. A comparison of forecasting errors for the optimal time window of different models over the next eight sample points reveals that the proposed model offers significant advantages in ultra-short-term forecasting accuracy and stability. Variable analysis indicates that the combined use of multiple meteorological variables significantly outperforms single-variable, uncovering the complex interactions and nonlinear relationships among meteorological factors.

The construction of virtual power plant (VPP） is now in the ascendant. It is necessary to comprehensively consider the economic requirements of VPP operators and the safety requirements of power grid for the participation of virtual power plants in grid dispatching. However, under the current power system information security requirements, the market-oriented virtual power plant operators have no aceess to the grid topology and real-time power flow information, and it may bring serious security risks to the power grid with the increase of aggregated resources in VPPs. The definition of virtual power unit is put forward based on the current situation of power dispatching system in China. Moreover, a dispatching architecture with partition aggregation and hierarchical scheduling for market-oriented VPPs is prospected from the aspects of partitional aggregation construction method of virtual power units, multi-level scheduling mode for VPPs and way of VPPs participating in the power market. Finally, an example shows that the dispatching architecture proposed can support the VPPs to participate in power grid dispatching and market transactions safely and economically without the grid topology information.

Multi-microgrid trading behavior not only achieves effective energy collaborative management but also brings economic benefits to entities in the distribution market. To minimize the operational costs of multi-microgrid while reducing power losses in the distribution network, a Nash bargaining method for multi-microgrid electricity trading considering distribution network reconfiguration is proposed. Firstly, an optimal operation model for multi-microgrid is established, accounting for their electricity trading. Secondly, based on Nash bargaining theory, a multi-microgrid electricity trading model is constructed, and the original Nash bargaining problem is equivalently transformed into two subproblems: minimizing multi-microgrid operational costs and maximizing payment benefits. Then, a distribution network reconfiguration model is developed using the mixed-integer second-order cone programming method to optimize power flow and reconfigure the network. The Dijkstra algorithm is employed to find the shortest electrical distance between microgrids under this network topology, updating the electricity transmission paths. Finally, case studies validate the effectiveness of the proposed method. Simulation results demonstrate that the Nash bargaining method for multi-microgrid electricity trading considering distribution network reconfiguration maximizes social benefits, reducing total social costs by 13.67 %.

In order to evaluate the maximum carrying capacity of new energy sources in the existing flexible interconnected distribution network, the original model is usually transformed into a mixed integer second-order cone optimization model. However, the second-order cone relaxation conditions are harsh, the generality is not high, the solution convergence is poor, and the model solution speed is slow, which cannot adapt to the fast evaluation of large-scale complex distribution networks. Therefore, a new energy carrying capacity evaluation model based on improved DC power flow algorithm is proposed, which transforms the original nonlinear model into an approximate linear model, improves the solving speed and accuracy of the model, and meets the requirements of rapid evaluation and analysis of new energy carrying capacity of flexible interconnected distribution network. The effectiveness of the proposed method is verified by IEEE 33-node distribution network and 1 000 random simulation examples, and the accuracy and numerical stability of the proposed improved DC power flow algorithm are compared and verified.

In the process of reversing the load of the distribution network, the loop current formed during the loop operation is large, resulting in the failure of the load transfer, and with the access of new energy, the load transfer is more and more complex. In order to quickly and efficiently conduct the load transfer of distribution network, a load transfer optimization model of source and load uncertainty is proposed. Firstly, the calculation method of measuring the coupling characteristics of the transmission element namely the transmission impedance electrical distance method is proposed. By describing the coupling degree between elements, the goal of reducing the loop current is achieved, forming an electrical distance constraint condition. Then, by integrating the safety constraints of load transfer and the objective function of minimizing the total number of switch operations, the uncertainties of distributed generation (DG) and load are analyzed to obtain the chance constraints of loop closing current, establishing an optimal load transfer model that accounts for the uncertainties of sources and loads. Finally,the wiring configuration of a regional distribution network is analyzed to verify the feasibility and effectiveness of the proposed model.

In response to the issues of low accuracy and poor real-time performance caused by the large scale of foreign objects and complex backgrounds in existing transmission line foreign object detection algorithms, a lightweight model based on improved YOLOv8 is proposed. Firstly, a modified feature extraction network is designed by integrating a dual-branch architecture with attention mechanisms into MobileNetV2, optimizing the model's parameters and enhancing the feature representation capability. Additionally, the CBAM attention mechanism is embedded into the SPPF and PANet modules to enhance the discrimination between foreign objects and the environment, to improve the model's detection capability in complex backgrounds and consequently enhance the detection accuracy. Furthermore, the WIoU loss function is introduced to properly allocate gradient gains, enhancing the model's generalization ability and detection box localization accuracy. Experimental results demonstrate that the improved YOLOv8 achieves a detection accuracy of 96.12% mAP and an inference speed of 60 frames per second. It outperforms YOLOv8 and five other mainstream object detection models in terms of overall performance. Moreover, it exhibits high stability and robustness under different lighting conditions, proving its practicality in complex environments. Further testing on embedded devices confirms that even under limited computational resources, the improved YOLOv8 can still achieve accurate detection.

In order to study the characteristics of zero sequence voltage signal change during tree line fault, an experimental platform for tree line fault is built to study the zero sequence voltage variation characteristics when single-phase tree line ground fault occurs in different vegetation under different residual currents in the resonant grounding system. The change stages and laws of the zero sequence voltage are summarized and the influence laws of the residual currents and vegetation's own parameters are analyzed on the change of the zero sequence voltage. Based on this, the prediction model for the peak value of the zero sequence voltage is built. The experimental results show that the development process of tree line fault can be expressed by the zero sequence voltage variation, which typically exhibit four stages: slow increase, rapid increase, decline and recovery. The trend of these changes follows three types of curves: U-shaped, J-shaped, and V-shaped. The increase of residual current will lead to a decrease in the amplitude of the zero sequence voltage change, which is not conducive to the detection of faults, and its value should be compensated to the range of 0~1 A as much as possible. For the same kind of vegetation, the peak zero-sequence voltage is positively correlated with the water content and diameter, and the peak time is negatively correlated with both of them. The structural difference of different kinds of vegetation is also the key to influence the peak zero-sequence voltage and peak time. The results of this study are of great significance for the detection and early warning of hill fires caused by tree line faults.