For new power systems, stable operation depends on a safe and reliable grounding system, especially in complex terrains such as mountainous areas with high soil resistivity. Traditional grounding designs often face challenges like excessive grounding resistance, insufficient lightning protection performance, and high economic costs, which severely threaten the safety of transmission grids. To address these issues, this study proposes a novel composite grounding system based on conductive concrete materials. The core objective is to optimize the design parameters of the composite grounding electrode for electrical performance and economic efficiency, thereby providing a feasible technical solution for grounding system construction in high-resistivity mountainous areas.

First, a numerical model of the composite grounding system suitable for high-resistivity areas was constructed to clarify the quantitative mapping relationships between structural dimensions, laying methods, material parameters, and key indicators such as grounding resistance and project cost. Second, taking electrical performance indicators and economic cost as optimization objectives, a multi-objective optimization model was established, with grounding electrode length, spacing, burial depth, and conductive concrete mix ratio as optimization variables. To solve this model, an improved particle swarm optimization algorithm was designed to enhance the global search ability and convergence speed by introducing adaptive inertia weights and mutation operators. Finally, a 110 kV transmission line project in the mountainous areas of Zhejiang Province was selected as a case study for simulation analysis and on-site comparative verification.

Simulation results show that the optimized design effectively reduces the grounding resistance to comply with relevant electrical standards. Compared with traditional grounding schemes, the proposed composite grounding system lowers construction costs while maintaining excellent electrical performance. Furthermore, the conductive concrete material exhibits good corrosion resistance and stability, with a predicted service life exceeding 30 years.

This study develops an optimized design method for composite grounding systems based on conductive concrete, effectively balancing electrical performance and economic efficiency. The proposed method addresses the challenges of excessive grounding resistance and high costs in high-resistivity mountainous regions, providing both a theoretical foundation and an engineering reference for grounding system design in new power systems. Future research will focus on integrating intelligent monitoring technologies to enable real-time performance detection and early warning, further enhancing the reliability and intelligence of grounding systems.

Due to the short operational period of the C919 aircraft, relatively limited operational data are available, and data collection in the event of faults is even more difficult. In addition, large aircraft systems have a complex structure, and the research on fault mechanisms remains undeveloped. Moreover, aircraft operation tasks and environmental conditions change frequently, and fault propagation characteristics remain unclear. Furthermore, the experimental and operational timeframes are relatively short, and a complete and effective fault diagnosis method has not yet been established. Therefore, fault diagnosis in large aircraft energy systems faces a series of challenges. Consequently, the flight safety of C919 hinges on developing efficient fault diagnosis techniques and promptly detecting system anomalies.

Because of low accuracy and poor real-time performance in fault diagnosis caused by insufficient fault data and the long-distance dependence of operating data features in the energy system of the C919 aircraft, a fault diagnosis technology combining an auxiliary classifier generative adversarial network (ACGAN) module and a bidirectional long short-term memory (BiLSTM) network is proposed. First, given the problem of limited operating time and scarce fault data in energy systems, a data augmentation module with an ACGAN network as the core is constructed to effectively generate operating data and balance data of different categories. Second, owing to the long-term dependence of fault data, a feature extraction network based on BiLSTM as its core is constructed to accurately characterize the feature dependency relationship during fault evolution and improve the accuracy of fault diagnosis. Concurrently, because it is difficult to efficiently process key features affecting fault diagnosis, an accelerated calculation module based on an attention mechanism is designed to focus on critical fault feature information and greatly reduce information overload. This module achieves rapid processing of fault features and improves the real-time performance of fault diagnosis.

To verify the effectiveness of the proposed method, comparative experiments were conducted with commonly used diagnostic models, including support vector machine, long short-term memory (LSTM) network, empirical mode decomposition, convolutional neural network, and the wavelet transform (WAVELET) model. To reduce statistical errors, each model was tested 30 times, and the average values were calculated to obtain diagnostic accuracy, recall, F1 score, and diagnostic time. The experiments showed that the proposed method achieved the best results in fault diagnosis accuracy, recall, and F1 score. Although the diagnostic time was lower than that of the LSTM and WAVELET methods due to model complexity, the proposed method still maintains high applicability in practical applications.

The proposed method achieves good diagnostic performance across various fault modes, with higher diagnostic accuracy and real-time performance than existing fault diagnosis methods. In-depth analysis of fault modes based on operational data enables more accurate and rapid identification of fault types and locations, providing useful references for fault diagnosis and offering strong support for the flight safety of large aircraft.

Modern power systems are rapidly evolving, characterized by an exponential increase in equipment volume, highly complex network topologies, and continuously changing operating conditions. Traditional fault diagnosis methods, relying on expert knowledge and threshold-based judgment, now face technical bottlenecks, including limited real-time performance, high misjudgment rates, and difficulties in multimodal data fusion. Developing intelligent diagnostic technologies based on machine learning, artificial intelligence, big data mining, and multisource information fusion has therefore become essential for achieving accurate fault identification and rapid fault location.

Recent advances in machine learning and large language models (LLMs) have opened new paths for fault diagnosis. Deep-learning-based architectures can integrate time-series operational data with multisource monitoring signals, extract cross-modal fault correlation features, and analyze fault propagation paths under dynamic coupling relationships. Transfer learning and federated learning help mitigate data silos and improve cross-regional diagnostic generalization. Supervised learning technology enhances anomaly detection robustness in small-sample scenarios. Meanwhile, LLMs can incorporate domain knowledge graphs through knowledge distillation, perform semantic reasoning and multicriteria collaborative decision-making, and support the development of diagnostic systems that combine data-driven learning with knowledge-informed causal inference.

This article systematically reviews recent research progress in machine learning and LLM-based fault diagnosis technologies for new-type power systems, compares the characteristics of representative approaches, summarizes the key challenges in their practical application, and finally, outlines future research directions.

The work provides theoretical insights and technical pathways for building efficient, robust, and interpretable intelligent fault diagnosis systems for new power systems through systematic technical analysis and sorting.

The fault diagnosis technology of power systems is a core means to ensure the safe and stable operation of new power systems. As these systems evolve toward a high proportion of renewable energy, multi-energy coupling, and power electronicization, fault diagnosis technology faces challenges such as complex dynamic characteristics, diversified data dimensions, and increased uncertainty. Therefore, a comprehensive analysis of fault diagnosis technologies and their adaptation to new power systems is urgently needed.

Currently, fault analysis methods based on mechanism modeling usually analyze the operating characteristics of system mechanisms, construct corresponding physical equations, and infer fault states. These methods focus on building physically driven mathematical models to achieve fault localization through dynamic system analysis. They offer clear physical interpretability but face challenges such as high modeling complexity and limited adaptability in strongly nonlinear or multistage scenarios. Conversely, diagnostic methods based on signal analysis focus on extracting features from monitoring data, using time-domain or frequency-domain techniques to analyze fault signals, and combining these with feature engineering to complete fault classification. These methods employ machine learning algorithms to establish warning models for trend analysis and fault prediction. The application of big data technology further enhances the coupling between historical data and machine learning models, improving fault recognition rates and prediction robustness. However, these data-driven methods depend heavily on data quality and the completeness of feature engineering, and their generalization to new types of faults is limited. Mechanism modeling and signal analysis methods represent the traditional mainstream approaches, constructing diagnostic systems from complementary perspectives.

This article systematically reviews research progress in both approaches, analyzes their principles, advantages, and limitations, and summarizes the challenges and potential research directions based on the requirements of fault diagnosis in new power systems. In particular, future research should focus on the deep integration of the two approaches: (1) incorporating prior knowledge from mechanistic models to optimize the design of signal analysis–based methods and reduce dependence on annotated data, and (2) applying signal analysis techniques to address unmodeled dynamics in mechanistic models. These complementary strategies lay a solid technical foundation for constructing effective fault diagnosis methods.

By clarifying the characteristics and limitations of mechanism-based and signal analysis–based methods, this study provides valuable guidance for developing more robust and effective fault diagnosis strategies, ultimately supporting reliable analysis and control of faults in new power systems.

Medium-voltage DC full-electric propulsion technology in vessels offers high energy density, small space occupation, and flexible system layouts, making it the main direction for the development of integrated power systems in ships worldwide. However, medium-voltage DC full-electric propulsion systems operate in harsh environments with high temperatures, humidity, limited space, and extreme power shock changes. These conditions lead to complex system structures, high component coupling, and frequent operating state changes, increasing the probability and impact of failures and causing propagation of failures, thereby endangering the stable operation of the overall vessel system.

A fault diagnosis technology based on convolutional neural network (CNN)–bidirectional long short-term memory (BiLSTM)–Transformer is developed. To improve the effectiveness of fault feature extraction, a CNN-based single-point feature cascade network is constructed to achieve in-depth extraction of spatial details of fault signals at a specific moment. Furthermore, to improve the accuracy of fault diagnosis, a BiLSTM-based multipoint feature dependency network is designed to effectively learn the feature dependency relationship of fault signals between multiple time points through gate control and bidirectional temporal learning mechanisms. Next, a Transformer-based sequence-feature parallel processing network is established and a self-attention mechanism is used to characterize contextual relationships in fault evolution, improving diagnostic accuracy. Multihead attention design enables parallel processing of input sequence features, enhancing real-time fault diagnosis performance.

Comparative experiments with commonly used diagnostic models, including GRU, LSTM, BiLSTM, and CNN, were conducted, analyzing performance indicators such as average diagnostic accuracy, recall, F1 score, and diagnostic time. The proposed method demonstrated high diagnostic accuracy and the shortest diagnostic time.

Compared with existing mainstream fault diagnosis methods, the proposed method can achieve excellent diagnostic performance across various fault modes, providing strong technical support for the safe operation of medium-voltage DC full-electric propulsion systems in vessels.

Due to the increasing system complexity, multielectric aircraft will experience various malfunctions, seriously affecting their safe and stable operation. However, traditional fault-detection methods are generally based on preset judgment rules and fixed detection thresholds, which are difficult to adapt to the dynamic changes in the operation status of aircraft. In addition, traditional fault-detection methods based on statistical data and expert experience lack learning ability and are, therefore, unable to detect unknown faults. Moreover, handling problems such as large data and low analysis efficiency using traditional methods is difficult, making it impossible to achieve real-time diagnosis. Therefore, there is an urgent need to develop an efficient fault-detection technology to achieve accurate and real-time fault detection in complex operating modes for multielectric aircraft.

First, a multilevel feature extraction module with ResNet as the core unit is built to obtain multilevel feature expressions about faults, improving the real-time performance of fault diagnosis. In addition, by constructing a multiscale feature extraction module with a BiLSTM network as the core unit, long-term-dependent features during fault evolution can be obtained, thereby ensuring fault diagnosis accuracy. Second, a Bayesian optimization algorithm-based hyperparameter optimization method with fault detection accuracy as the objective function is proposed, which helps improve the learning effect of fault features and further enhance the fault detection performance. Finally, a multielectric- aircraft fault-detection experiment is designed, and the detection performance under different fault modes is analyzed.

In the hyperparameter optimization method, the accuracy of the fault detection model on the training set is selected as the objective function, and an acquisition function is used to maximize the objective-function solution, thereby obtaining the optimal hyperparameter combination that affects the detection performance in the model. In a comparative experiment of fault detection, to verify the effectiveness of the proposed method, comparative experiments were conducted with commonly used methods, including ResNet-BiLSTM models without Bayesian optimization, CNNs, and BiLSTM. The results show that the proposed Bayesian optimization-based method can achieve effective fault-detection results in various fault modes and that compared with those of existing methods, the accuracy and real-time performance of this method are better.

The proposed method can effectively detect and classify different faults in various operation modes, which helps achieve accurate analysis of faults for multielectric aircraft and provides very powerful technical support for flight safety.

This project initiated a groundbreaking approach to overcome the inherent obstacles faced in traditional power system state analysis experiments, which are notably burdened by extensive construction, operational expenses, and significant safety risks due to the use of high-voltage equipment. The initiative to craft a virtual simulation experiment was driven by the urgent need to circumvent these issues, offering an innovative approach that ensures safety and cost-efficiency as well as diversifies the pedagogical methods employed in imparting critical electrical engineering knowledge. This endeavor aimed to provide an enriching, comprehensive, and interactive educational experience, enabling students to delve deep into the intricacies of power system state perception within a meticulously simulated environment.

This study adopted a thorough methodology that included the creation, deployment, and assessment of a virtual simulation experiment framework. This framework comprises three key modules, each designed to fulfill a specific role in the overall educational process: foundational principle and preparation module for basic cognition of the experiment , intelligent perception experiment module for hands-on experience, and assessment and evaluation module for gauging the comprehension and application skills of students. Emphasizing realistic and immersive learning, the framework especially highlights the intelligent perception experiment module, where students safely interact with simulated high-voltage equipment operations. Advanced simulation technologies are employed to recreate power system dynamics, promoting active exploration of state perception techniques. The evaluation strategy is based on outcome-based education principles, establishing clear and measurable learning objectives and outcomes, and employs qualitative and quantitative methods to evaluate the effectiveness of the virtual simulation in enhancing the educational experience.

Implementing the virtual simulation experiment eliminated the safety and financial concerns inherent in traditional high-voltage equipment experiments, thereby leading to significant improvements in electrical engineering education. It offered a safe, flexible platform for students to conduct experiments at their convenience, breaking time and location constraints. This flexibility encouraged deep engagement with the involved material, allowing students to conduct multiple experiments, explore different scenarios, and critically assess their findings in a risk-free setting. The experiment enriched the educational content and revolutionized teaching methodologies by integrating advanced simulation technologies with outcome-based educational principles, thus creating a dynamic and interactive learning environment that fosters innovative thinking and problem-solving skills.

The successful implementation of a virtual simulation experiment for power system state perception represents a significant step forward in electrical engineering education. By addressing the critical challenges associated with traditional experimental setups, this approach ensures a safe and cost-effective learning environment as well as enriches the educational experience through enhanced interactivity and accessibility.