State Grid Electric Power Research Institute

Shunt reactors are widely adopted in low-voltage-level stations to solve the reactive power reversal in urban power grid. Compared with traditional iron-core shunt reactor, toroidal air-core superconducting shunt reactor comprehensively exhibits the advantages of high compactness, low magnetic field leakage and low noise. However, relatively high transport AC loss may pose unexpected huge burden to cryogenic system, hindering the stability and feasibility of superconducting shunt reactor. In this paper, a partial-size model with appropriate geometrical simplifications is built to analyze the electromagnetic characteristics of toroidal superconducting shunt reactor with racetrack double-pancake (DP) coil as the least unit. Result shows the effect of toroidal structure on the amelioration of the uniformity of magnetic field distribution, leading to the suppression of the transport AC loss. This improvement will give practical engineering guidance to the future application of toroidal superconducting shunt reactor.

A high temperature superconductor (HTS) magnet is an attractive solution for reducing energy consumption to generate a major-diameter, high-quality single silicon crystal. However, quench has been an urgent problem for the application of HTS coils. Rapidly discharging the current in the coil is a crucial part of quench protection. Previous studies have shown that the copper plates initially intended for support and cooling can accelerate the coil current discharging rate during fast energy extraction. However, it's unclear how the copper plates influence the discharging process of the large scale HTS magnet with a low magnetic field for a single silicon crystal growth system. This paper designs a HTS magnet for a single silicon crystal growth system and a detailed study of the electromagnetic thermal characteristics of the magnet coupled with copper plates and rings has been conducted through simulations. The results show that both the copper plates and the copper rings can absorb energy from the HTS coil through electromagnetic coupling and the copper plates perform better than the copper rings. The copper frame combined with the copper plates and rings shows the best performance in accelerating the current discharging rate as well as decreasing the hot spot temperature of the HTS coil. This copper frame is a promising structure for the quench protection of HTS magnets for a single silicon crystal growth system.

The thunderstorm hour (TH) is a valuable indicator for assessing the impact of climate change on thunderstorm frequency. Using the cloud-to-ground (CG) lightning stroke data detected by the lightning location network (LLN) from 2013 to 2021, the annual spatial distribution of stroke density (SD) and THs in China is analyzed. The monthly and diurnal variations of THs on different landforms were examined over central and eastern China as the region of interest. Considering the differences in geomorphic types, we also explored the relationship between THs and land features and monthly average convective parameters. In space, the overall pattern of THs is similar to that of SD; however, there are differences in local regions. In time, the maximum value of THs occurs in July for annual variation and around 1600 Beijing time for daily variation. The average thunderstorm hours in mountainous areas are higher than in flat areas. In addition, the TH index (THI) on mountainous landforms is usually greater than 1, meaning these landform types have a “preference” for thunderstorm occurrence. Due to the higher coverage area of flat terrain compared to other terrains, THs tend to accumulate in low-elevation areas. In mountainous areas, the sloping terrain is a factor in the increase in thunderstorm hours in all seasons. The THs generally increase with increasing spring, summer, and autumn vegetation cover. However, in summer, there is an opposite trend when the vegetation index exceeds 0.8. Instability, humidity, and composite parameters can be used as proxies to predict the monthly thunderstorm number. Furthermore, the correlation relationship between thunderstorm hours and land features and thermodynamic parameters varies with geomorphic types. Our research provides a theoretical foundation for constructing a long-term historical dataset of thunderstorm hours in the future.

Fire water systems serve as a core component of the city's lifeline program. Its safety and reliability are directly related to the safety of people's lives and property and the operational order of the city. However, firefighting pipelines are often at risk of leakage due to aging, mechanical damage, corrosion, and circumferential weld cracking. Therefore, effective methods for detecting and locating pipeline leaks are essential. Existing methods are often complex, time-consuming, and unreliable. This paper proposes an integrated approach for leak localization in fire-distance fire protection pipelines by constructing a light gradient boosting machine-based model. The model simultaneously addresses the challenges of multi-categorical leak localization with unbalanced data and enhances the accuracy and reliability of leakage detection in long-distance pipelines. The model identifies leak locations through a two-step process and demonstrates high performance, achieving an accuracy rate of over 90%. Furthermore, it exhibits strong generalization capability and robustness. This method significantly enhances the accuracy of leak detection in long-distance fire protection pipelines, improves the operational reliability of fire water networks, and enables staff to make prompt maintenance decisions.

Anomaly detection has proven effective in detecting cyber-attacks in Industrial Control Systems (ICS). However, most existing anomaly detection methods suffer from low accuracy because they ignore the effects of packet loss and network delay on time features, the sequential nature of transition time, masquerade transitions, and system recovery. Meanwhile, current Cyber-Physical Model (CPM) construction methods struggle to effectively address the state explosion problem and properly balance the removal and retention of low frequency states (LFS). In this paper, we propose a novel baseline model for ICS to detect anomalies through learning device-level polling time patterns and system-level CPM. The polling time pattern learning method reduces the effects of packet loss and network delay on time features by extracting only matching packets and replacing outliers. The CPM construction method mitigates state explosion through mixed-event discretisation, reduces the effects of network delay on transition/action times through outlier replacement, and captures the sequential nature of transition times with circular permutation sets. CPM model optimisation uses a post-pruning algorithm to balance the removal and retention of LFSs, and a CPM periodicity detection method that mitigates the effects of network delay to ensure that all industrial process periods are detected. A real-time anomaly detection method with a two-layer defence mechanism is proposed using the baseline model. Experimental results from two lab-scale ICSs with six process-related attacks confirm the effectiveness and superiority of the proposed method. It achieves average F1 scores of 98.81% and accuracy of 99.24%, outperforming the state-of-the-art work by 18.51% and 13.96% respectively.

This study investigates vibration changes due to transformer core deterioration by monitoring core vibration, compression force and shell wall vibrations simultaneously. The transformer core operates in a compound environment of mechanical vibration and thermal ageing for extended periods, and the correlation mechanism between core structural deterioration and shell vibration changes remains unclear. This study first derives and analyses the propagation mechanism of core vibration in oil. The experiments simulate the internal deterioration of a 10 kV transformer using pressure sensors to monitor the compression force on the core and windings and vibration sensors on the internal upper yoke and the enclosure to capture full vibration measurements. Analysis of the vibration data during the experiment, using two quantitative indicators—vibrational entropy and fundamental frequency weight—reveals that measurement point #2 (on the outer case wall corresponding to the internal upper yoke) shows a value approximately 1.2 times that of the internal upper yoke. However, measurement point #5 (located away from the upper yoke near the windings) demonstrates a value about 2.3 times that of the internal upper yoke. The results indicate that measurement point #2 has high vibration consistency with the internal upper yoke, whereas it exhibits significant variability compared to measurement point #5. To validate these findings, researchers collected 24‐h vibration data from 105 in‐service 220 kV transformers and the results aligned with those from the experimental platform. This study quantitatively addresses the changes in case vibration characteristics caused by core degradation and proposes a novel method for detecting the mechanical state of transformer cores through vibration analysis.

Grid‐interactive efficient buildings (GEBs) have garnered global attention for their ability to achieve flexible, resilient, and environmentally friendly objectives. However, the increasing integration of renewable energy sources (RESs) introduces challenges which can compromise power system stability. Traditional robust energy management approaches fall short as they fail to address the adverse impacts on small‐signal stability. Additionally, the complexity of coordinating diverse devices and their intricate interactions leave the concept of co‐optimization in GEBs in its nascent stages. To address these challenges, this paper proposes a robust optimization model for GEBs that minimizes costs while ensuring system stability. The model integrates adjustable droop gains in inverters connected to distributed energy resources (DERs). First, dynamic models for various GEB devices are developed. Next, an hourly optimal power flow problem is formulated using interval predictions for RESs to ensure robustness against uncertainties. Leveraging a polyhedral uncertainty set, the model is solved via a Benders decomposition‐based method, incorporating analytical stability sensitivity cuts. Simulations on a 33‐bus GEB demonstrate that the proposed model significantly enhances small‐signal stability at a relatively low cost, outperforming benchmark models in handling uncertainties. This approach marks a significant step forward in advancing the co‐optimization of energy management and stability in GEBs.

Thermally responsive microspheres (TRMs) have received increasing attention due to their unique temperature-dependent expansion performance. In this work, the preparation of a non-toxic, environmentally friendly, and low-cost TRM was investigated using non-nitrile monomers in suspension-emulsion polymerization. Various TRMs were prepared with different dispersant and emulsifier concentrations, various core materials and polymerization times to study the thermal expansion behavior, thermal response and the mechanism of microsphere polymerization, respectively. The optimized conditions, involving nano-SiO2 as a dispersant, sodium dodecyl sulfate as an emulsifier, azobisisobutyronitrile as an initiator, as well as styrene, methyl acrylate, methacrylic acid, and methyl methacrylate at a certain mass ratio with a 16-hour reaction time at 65 °C, enabled the production of TRMs with stable spherical structures, high expansion ratios, and smooth, defect-free surfaces. This work provides a facile and environmentally friendly strategy for preparing TRM, which has the potential to be used in industrial applications.

Double fed power supply exhibits complex transient characteristics during short-circuit faults and significant vulnerability during low voltage ride through, posing challenges to its protection configuration. This article focuses on a short-circuit current protection setting algorithm for doubly fed power supply based on K-means clustering. A mathematical model of a doubly fed power supply in the dq coordinate system was constructed using the Park transformation principle. Based on this model, a deep analysis of the short-circuit current characteristics of the doubly fed power supply is conducted to determine the setting scheme for the short-circuit current. In addition, by introducing particle swarm optimization algorithm to optimize the K-means clustering process, the clustering results are used to identify short-circuit faults in doubly fed power supplies. Once a short circuit fault is detected, the protection device will activate to adjust the short circuit current. The experimental results show that the algorithm can effectively regulate the short-circuit current during the short-circuit fault of the doubly fed power supply, helping it quickly recover to normal operation.

Multi-label classification is a significant challenge in machine learning, especially as the dimensionality of the problem increases. As the number of dimensions grows, the performance of traditional classification algorithms often degrades substantially. Feature selection is a key technique for reducing dimensionality in multi-label scenarios, operating as a non-parametric process. Despite its importance, feature selection remains a complex issue without straightforward solutions, and various approaches using AI and evolutionary algorithms have been proposed to tackle it. However, these methods typically suffer from reduced efficiency and slower convergence as the dimensionality increases, due to the expanding search space. To address this issue and enhance convergence speed, this article introduces a hybrid AI solution that combines a binary particle swarm optimization algorithm with a local search strategy specifically designed for multi-label feature selection. Within this local search strategy, feature fusion plays a crucial role, where features are merged based on their relevance and correlation with the problem’s output. These fused features are divided into two categories: those directly associated with the problem class and those that are similar to the problem class but distinct from other feature fusions. By leveraging this categorization, the particle swarm optimization technique is augmented with a local operator that removes redundant feature fusions and refines each solution. By incorporating this operator, the proposed method achieves superior convergence speed compared to previous algorithms in the field. The performance of the proposed method was evaluated across several datasets against some of the most widely used feature fusion selection algorithms. The experimental results demonstrated the proposed method’s accuracy and efficiency, validating its effectiveness in multi-label feature selection.

A visible light‐driven trifluoromethylation protocol using SF6 as an oxidant has been developed. This method achieves regioselective C(sp²)–CF3 bond formation under mild conditions with Rose Bengal catalyst, blue LEDs and bench‐stable CF3SO2Na. The reaction affords moderate to excellent yields across various (hetero)arenes, which facilitates the efficient utilization and degradation of the potent greenhouse gas SF6. image © 2025 WILEY‐VCH GmbH

Background With the continuous increase in the penetration rate of distributed generation (DG) in distribution networks, the pressure on energy supply has been effectively alleviated. However, its impact on the voltage of distribution networks is gradually becoming apparent, and traditional reactive power voltage support methods are struggling to meet the requirements for fast and flexible voltage regulation in modern distribution networks with high DG penetration. Objective To accommodate the integration of DGs with high penetration rates and enhance the voltage support capability of the distribution network, an emergency voltage support scheme is proposed. The proposed scheme takes into account the impedance properties of the distribution network and focuses on low-voltage distribution networks. Method This paper proposes a voltage support scheme for low-voltage distribution networks, taking into account the resistive characteristics of distribution lines. Based on the distribution network impedance ratio (R/X), the optimal active and reactive current reference values are estimated. A voltage support strategy is planned according to the severity of the voltage drop, providing coordinated voltage support by simultaneously delivering active and reactive power during voltage sags and fault conditions. Result A digital-physical hybrid experimental platform and a simulation model of the low-voltage distribution network were established to validate the effectiveness of the proposed scheme. The results indicate that the voltage support capability of the proposed scheme is superior to that of traditional reactive power voltage support methods. Conclusion The low-voltage distribution network voltage support scheme proposed in this paper considers the distribution network impedance ratio to calculate the required injected active and reactive power, thereby providing maximum support to the voltage. In addition, the active and reactive currents are adjusted according to the severity of the voltage drop using a droop curve. The effectiveness of the proposed method on the operation of the distribution network is validated through simulations. The scheme can effectively provide voltage support during voltage dips in the distribution network, preventing voltage instability.

Aims The study aims to find effective approaches for accurate photovoltaic (PV) power prediction and offer support for grid scheduling and energy management. Background In the course of the global energy transition, PV power generation, being clean and sustainable, is essential for easing the energy crisis, reducing pollution, and promoting sustainable development. But achieving high-precision PV power prediction is difficult due to complex weather, unstable cell parameters, noise, and large power fluctuations. Objective The objective is to build a highly accurate and stable PV power prediction model. Methods This research innovatively combines the maximum information coefficient (MIC), neural network, and interpretable deep - learning (DL) methods. First, the MIC method is used to screen key driving features like tracking tilted irradiance (TTI), global tilted irradiance (GTI), and diffuse horizontal irradiance (DHI) that affect PV power. Then, the Dung Beetle Optimization (DBO) algorithm is applied to optimize the hyperparameters of the spatio-temporal convolutional neural network (TCN), aiming to enhance the model's prediction performance. Results Compared with traditional models such as LSTM and GRU, the proposed model shows greater accuracy and stability in both short- and medium- to long-term forecasting. In 24-hour forecasting, the mean absolute error (MAE) of the model drops to 3.92432, the root mean square error (RMSE) to 6.22452, and the R2 reaches 0.97984, which is much better than other comparative models. Besides, through the SHAP (SHapley Additive exPlanations) method for interpretive analysis of the model, the impacts of each driving feature on PV power generation are quantified. The results show that TTI has the most significant impact on PV power generation, followed by GTI and DHI. Conclusion The contribution of this study is to mine the key factors affecting the output of PV power generation by using the MIC, and apply the DBO to optimize the hyper-parameters of TCN, which simulates multiple behaviors of dung beetle to achieve optimization, and has a better ability to find the optimum compared with other algorithms. The SHAP method is introduced to perform explanatory analysis of the model to quantify the contribution of each input variable to the model output. The model proposed in this study effectively improves the accuracy of PV power generation prediction. It provides strong support for grid scheduling and energy management, which is of great significance for the development of the PV power industry.

The fault current of a voltage source converter-based high-voltage direct current (VSC-HVDC) system has the characteristics of high amplitude and fast-rising speed. Therefore, the available hybrid dc circuit breaker (HDCCB) faces great challenges in breaking capacity and breaking speed. A saturated iron-core superconducting fault current limiter-based hybrid dc current-limiting circuit (SI-SFCL-HDCCLC) is proposed, which uses the saturated iron-core superconducting current limiter and current-limiting resistor to realize the resistive–inductive compound current limiting, and uses the energy-absorbing resistor to accelerate the fault current clearing and relieve the breaking pressure of HDCCB. In this paper, we will introduce the working principle of the SI-SFCL-HDCCLC theoretically by stages in combination with the coordination strategy of the proposed SI-SFCL-HDCCLC and HDCCB based on the transient characteristics of a dc fault. Then, the principles of the parameter design of the SI-SFCL-HDCCLC are analyzed with mathematical derivation. Finally, the theoretical correctness and functional effectiveness of the proposed SI-SFCL-HDCCLC are validated in PSCAD/EMTDC-based environment and in lab-scaled experiment. Both the simulation and experimental results indicate that the proposed SI-SFCL-HDCCLC combined with HDCCB can effectively suppress fault current, shorten the isolation time of the fault line, reduce the energy absorption of the MOV and realize the rapid recovery of SI-SFCL.

Stretchable, self-powered strain sensors with high response and multifunction are of significant interest due to their potential in next-generation wearable technology. In this study, an approach was presented for design and fabrication of stretchable, self-powered strain sensors with enhanced-response and exteroception-visualizing. The sensors were fabricated by encapsulating a polyvinyl alcohol (PVA) electrode layer inside Ecoflex triboelectric layer. ZnS: Cu particles and NaCl were incorporated into triboelectric layer and electrode layer to enhance responses of sensors twice sequentially and further introduce exteroception-visualizing capability. The open-circuit voltage of the resulting sample with 0.048 g NaCl is about 3.6 times that of sample without NaCl and ZnS: Cu particles, exhibiting a substantial enhancement in the output voltage. The samples have a good stretchability, which is capable of detecting the human motions in a self-powered manner, and harvesting mechanical energy from these motions. Meanwhile, the as-prepared samples exhibit a clear luminescence response to external force stimuli, demonstrating an exteroception-visualizing capability, which could be used as a handwriting recognition system through their force-optical and force-electrical responses. Our approach could be a promising strategy to design and fabricate high-response, and multifunctional strain sensors, which could have a great potential for the next generation of wearable devices.