With the advancement of digitalization and intelligence in substation protection and monitoring systems， the volume of multi-source heterogeneous data generated by secondary systems has increased sharply， with continuously rising dimensionality. This has led to issues such as frequent anomalies， increased storage and computational costs. Therefore， cleaning monitoring data， consolidating redundant information， and establishing an efficient framework for data quality improvement have become critical to supporting the development of modern power systems. In light of this， this paper reviews the research progress in substation monitoring data cleaning and， in combination with the development trends in large language models （LLMs）， explores the application potential of multimodal large language models （MLLMs） for data quality improvement. First， the fundamental characteristics of substation monitoring data and the urgent need for improving data quality are analyzed. Then， it examines the principles， advantages， disadvantages， and applicable scenarios of existing data cleaning techniques around three key aspects： anomaly data detection， anomaly type identification， and data restoration. Further， it discusses the potential application pathways of MLLMs for these tasks. Finally， it outlines future development directions， with particular emphasis on the key issues that must be addressed in applying LLMs technologies.

With the increasing integration of renewable energy sources and power electronic devices into power systems, subsynchronous oscillation (SSO) has become a prominent stability issue. Accurate identification of oscillation sources is therefore essential for ensuring the secure and stable operation of power systems. This paper proposes a new source identification method for wind farm SSO by integrating the generalized multisynchrosqueezing transform (GMSST) with an improved impedance-based method. First, GMSST is employed to extract subsynchronous oscillation components from PMU measurements. Then, an improved impedance-based method is developed by combining the least-squares method with the impedance model. Finally, the proposed method is validated through simulation studies on a multi-wind-farm system, with the dissipating energy flow (DEF) method adopted as a reference for comparison. The results show that the proposed method yields results consistent with those of the DEF method, while improving the completeness and physical interpretability of the analysis from the perspective of impedance characteristics, thereby enhancing the reliability of oscillation source identification.

With the increasing proportion of the interconnection of renewable energy in electric power systems, challenges related to frequency stability have become increasingly prominent. This paper focuses on the system frequency support characteristics of grid-following (GFL) wind power and proposes a control parameter configuration method based on a two-dimensional frequency support safety domain framework, considering both penetration level and disturbance factors. First, a system frequency response model incorporating synchronous generators and GFL wind power is established. The influence of equipment control parameters on key frequency stability indicators—namely the frequency nadir(fnaidr), the rate of change of frequency (RoCoF), and system damping—is analyzed. Then, based on the frequency support safety domain defined with constraints on the fnaidr and RoCoF, and combined with damping constraints, the feasible range of control parameters is determined. Furthermore, a control parameter configuration method aimed at enhancing the frequency support margin of new energy equipment is proposed. Finally, the effectiveness and applicability of the proposed method are validated with simulations.

Driven by the 'Dual-Carbon' goals, power systems characterized by a 'high proportion of renewable energy and high proportion of power electronic equipment' have become increasingly prominent. Consequently, frequency instability, particularly localized issues, occurs frequently. The conventional System Frequency Response model often suffers from poor accuracy and insufficient robustness in these complex scenarios, necessitating urgent improvements. This paper establishes an improved SFR model considering primary frequency regulation that accounts for nodal inertia and the spatio-temporal distribution of frequency, and proposes a multi-dimensional assessment method for frequency stability adequacy. First, an improved SFR model incorporating spatio-temporal characteristics is developed based on nodal inertia. Second, the spatial divergence of frequency responses across different nodes is validated through both theoretical derivation and simulation analysis. Finally, leveraging the improved SFR model and spatiotemporal frequency characteristics, a frequency security adequacy assessment method based on the violation area is proposed.

Sub-synchronous oscillation (SSO) threatens the safety and stability of the power grid, and is one of the main facts that affects the consumption of renewable energy, so it is important to accurately analyze and identify the source of SSOs. This paper proposes a subsynchronous oscillation source method based on Multisynchrosqueezing transform (MSST) and cross power spectral density (CPSD). Firstly, combined with MSST, the time-frequency analysis results of the compression energy are obtained; Secondly, the ridge line tracking is utilized to identify and reconstruct the sub-synchronization component signals; Again, the flow direction of subsynchronous oscillations in the branch is given by applying the cross power spectral density method. Furtherly, combined with the inflow and outflow properties of different branches, the oscillation source is determined. Finally, the effectiveness and practicality of the proposed approach is verified with simulated data in a Four-machine two-zone system (4M2A system) and field measured data in China.

The reliability of relay protection systems is crucial to the safe and stable operation of substations and power grids. To overcome the limitation of conventional reliability block diagram (RBD) methods in describing dynamic fault detection and repair processes, this paper proposes a multi-state Markov reliability evaluation method for a 110 kV station-level protection system considering self-checking mechanisms. A three-state model is first established for a single protection device by distinguishing selfdetected faults from undetected faults. Then, line, bus-section, and transformer protection subsystems are modeled in a unified framework, and a state aggregation strategy is adopted to control state-space growth. Based on this approach, a 28 -state Markov model is constructed for the studied station-level protection system, and steady-state unavailability is calculated. Results show that the self-checking rate has little effect on steady-state unavailability when different fault-detection modes share the same repair efficiency. However, when self-detected faults can be repaired faster, increasing the self-checking rate significantly reduces system unavailability. The proposed method provides a practical tool for reliability evaluation and maintenance optimization of modern substation protection systems.

Under slow power-flow dynamics, a voltage instability phenomenon induced by low-voltage ride-through (LVRT) control switching rather than by saddle-node bifurcation (SNB) has been observed at a renewable energy station (RES) in China. The phenomenon leads to a large-scale disconnection of RESs. However, the understanding of the patterns and mechanisms of this LVRT-control-switching-induced (LCSI) voltage instability is still insufficient, resulting in potential operating risks for grid-connected RESs. This paper investigates the potential LCSI voltage instability patterns and develops a graphical discrimination method. First, three scenarios involving the triggering of LVRT control switching are identified. Second, a mathematical model of the grid-connected RES system is established. Then, based on the accessibility of equilibrium points, the LCSI voltage instability pattern and the corresponding bifurcation properties in each scenario are revealed. Furthermore, a graphical discrimination method that accounts for both SNB and LCSI voltage instability is proposed. Finally, the existence of three LCSI voltage instability patterns is verified, and the effectiveness and robustness of the proposed graphical discrimination method are validated.

After fault clearance, load center power grids with large-scale distributed renewable energy integration exhibit complex voltage instability phenomena, driven by the coupling effects between the Low-Voltage Ride-Through (LVRT) control switching of renewable energy stations and the load characteristics of induction motors. Notably, there is currently a lack of systematic analysis and understanding regarding the voltage evolution patterns and their underlying physical mechanisms under different proportions of induction motor loads. This paper aims to explore potential post-fault voltage recovery scenarios and conduct an analysis based on switching system theory. Specifically, a switching system model of the receiving-end power grid is first established, incorporating the continuous dynamics of induction motors and the discrete control logic of renewable energy sources. Secondly, the transient responses of the system under different induction motor proportions are simulated using the PSASP platform, revealing three typical evolution scenarios: normal recovery, voltage oscillation, and sustained voltage drop. Finally, these different evolution scenarios are analyzed by combining the $U_{\text {pcc-s }}$ phase plane trajectories with the system operational trajectories. Furthermore, a qualitative discussion on the coupling effect of renewable energy penetration and induction motor proportion is introduced, highlighting how their synergy amplifies the system's sensitivity to LVRT control switching and exacerbates instability risks.

Abnormal functions of the substation monitoring system may lead to unsuccessful remote operations and even cause power grid accidents. Therefore, it is necessary to scientifically and accurately assess the functional status of the substation monitoring system. This paper, in combination with the integrity theory, proposes an evaluation method for functional integrity of the substation monitoring system. Firstly, based on various standards and norms of the monitoring system and the message transmission logic, the functional mechanism model of the substation monitoring system is established; Secondly, real-time monitoring information is utilized to evaluate the functional availability and equipment health, and the comprehensive functional integrity is obtained. The evaluation results can provide a reference for the maintenance and debugging of operation and maintenance personnel. Finally, the feasibility of the proposed method is verified through an example.

With the rapid increase in renewable energy penetration, renewable energy power plants, characterized by low inertia and weak grid support capability, have a growing impact on transient voltage stability in power systems. However, existing transient voltage stability margin indices based on multi-binary tables are not well suited for renewable energy power plants. To overcome this limitation, this paper proposes an improved transient voltage stability margin index tailored for renewable energy power plants. The deficiencies of the conventional index in transient stability assessment are first analyzed. Subsequently, incorporating the low-voltage ride-through (LVRT) characteristics of renewable energy plants, an enhanced assessment index is developed. Simulation studies are carried out on a modified IEEE $\mathbf{1 0}$-machine $\mathbf{3 9}$-bus system, and the performance of the proposed index is evaluated under different LVRT control strategies. The results indicate that the proposed index is computationally robust and physically interpretable, enabling effective quantification of transient voltage stability margin for renewable energy power plants, and demonstrating its validity and engineering applicability.

Focusing on power systems with high renewable penetration, this study conducts renewable capacity planning based on system hosting capacity indicators. First, wind and photovoltaic output samples are used to analyze their probability and cumulative energy distributions, identifying the patterns of low output with high probability and progressive saturation. Then, considering system load, export scale, peak regulation capability of conventional units, and reserve constraints, the accommodation coefficient method is applied to characterize the mapping between renewable energy utilization and system hosting capacity. Furthermore, for the Qinghai grid, an explicit approximate model relating utilization to output level, along with a capacity planning model under target utilization constraints, is developed. The impacts of total load, load factor, peak regulation capability, and reserve level on the acceptable renewable capacity are also analyzed. The proposed approach unifies the statistical characteristics of renewable output with grid hosting capacity analysis, providing a concise and intuitive framework for determining planning scale and evaluating utilization in the Qinghai grid.

To address the issues of computational complexity and poor real-time performance in traditional low-frequency oscillation monitoring methods for power systems, this paper proposes a rapid identification method based on impedance trajectories. First, the characteristics and influencing factors of the constant active power impedance circle and the constant reactive power impedance circle are derived and analyzed. By incorporating the Thevenin equivalent circuit, an oscillation discrimination criterion is formulated utilizing the distribution patterns of the measured impedance trajectories at the generator terminals. Simulation results demonstrate that the impedance trajectory lies on the constant active power impedance circle during normal operation, but deviates significantly when oscillations occur, thereby verifying the correctness and effectiveness of the proposed method. The proposed method is computationally simple and highly time-efficient, enabling the effective identification of low-frequency oscillations.

Energy storage system has been adopted in power system to enhance its frequency stability. However, little research has been done on the types, installation locations and capacities configurations of energy storage. This paper proposes a method for determining the locations and capacities of multi type energy storage installations considering frequency stability requirements for a certain system. Firstly, it introduces a combined offshore wind power - thermal power - energy storage output system, along with its frequency stability equivalent model. Secondly, it presents frequency stability requirements. Thirdly, it proposes the optimization problem for configuration of a multi-type energy storages with the objective of minimizing total cost and frequency constraints, and utilizes particle swarm optimization algorithm to solve it. Finally, simulation results with battery energy storage and hydrogen energy storage show the effectiveness of the proposed methods in different scenarios.

The substation monitoring system(SMS) is the eye of the substation to ensure the safe and stable operation of power system. The correctness of the SMS function is important to meet the reliable operation of the substation. However, the underlying logic and processing mechanisms of SMS functionality are complex and difficult to monitor. Therefore, a method to construct the functional mechanism model for SMS is proposed. Firstly, the sub-functions involved in the three devices of the monitoring system are sorted out. According to the execution process of the function, it is disassembled into an ordered set of key nodes, and the monitoring elements related to node status are screened. Finally, the same type nodes and monitoring elements are integrated to form a three-level functional mechanism model of “element-node-function” which is suitable for the whole monitoring system.

Hydrological modeling is essential for effective water resource management, especially in complex regions such as the Tibetan Plateau, where conventional models often struggle to capture dynamic glacio-hydrological processes. This study introduces HydroTrace, an algorithm-driven model with custom attention mechanisms that consistently outperforms existing models across hydrological benchmarks at two observational sites on the Tibetan Plateau. HydroTrace achieves high predictive accuracy, with a calibrated Nash-Sutcliffe Efficiency (NSE) of 0.98, enables direct mapping of input–output contributions, identifies key physical drivers, and uses learned attention to guide water-balance regression. The approach reveals a previously overlooked year-round glacier influence on river discharge across the central and southeastern Plateau: in the downstream southeast, glaciers, melt, and baseflow co-vary with the monsoon yet glaciers remain a meaningful contributor outside the ablation peak; at higher central elevations, hydrology remains glacier-dominated throughout the year. By bridging algorithm-driven and physics-based modeling, with attention serving as a data-informed proxy, HydroTrace offers a practical pathway toward high-performing, interpretable, process-aware hydrological AI.

Traditional equation-driven hydrological models often struggle to accurately predict streamflow in challenging regional Earth systems like the Tibetan Plateau, while hybrid and existing algorithm-driven models face difficulties in interpreting hydrological behaviors. This work introduces HydroTrace, an algorithm-driven, data-agnostic model that substantially outperforms these approaches, achieving a Nash-Sutcliffe Efficiency of 98 strong generalization on unseen data. Moreover, HydroTrace leverages advanced attention mechanisms to capture spatial-temporal variations and feature-specific impacts, enabling the quantification and spatial resolution of streamflow partitioning as well as the interpretation of hydrological behaviors such as glacier-snow-streamflow interactions and monsoon dynamics. Additionally, a large language model (LLM)-based application allows users to easily understand and apply HydroTrace's insights for practical purposes. These advancements position HydroTrace as a transformative tool in hydrological and broader Earth system modeling, offering enhanced prediction accuracy and interpretability.

Currently, the scheduling model using DC power flow, thus it result in indirect considering of the voltage stability with checking. At the same time, there is lack of scheduling strategy for power system with multiple energy storage. A multiple time-scales scheduling strategy for power system with multiple energy storage which can directly considering voltage stability is proposed in this paper. Specifically, firstly, a linear AC power flow model that can directly reflect the voltage magnitude is presented; secondly, in order to consume new energy (NE), a multiple time-scales scheduling model combined with the linearized voltage stability constraint is constructed. Finally, the simulation results with the improved IEEE30 bus by GUROBI, show that the methods could meet the voltage stability requirements and effectively reduce the renewable energy abandonment rate.

Grid following battery energy storage system (GFL-BESS) can cope with the fluctuation and uncertainty of the output of direct-drive permanent magnet synchronous generator (D-PMSG). However, during the suppression of D-PMSG output fluctuations, the impact of GFL-BESS on the small signal stability of its control link is unknown. Based on the eigenvalue analysis method, this paper analyzes the influence of GFL-BESS on the dominant eigenvalues of D-PMSG during power response . In details, firstly, the state space model of D-PMSG is presented. Secondly, the mathematical model of the GFL-BESS with its control strategy is established. Finally, the influence of GFL-BESS on the small signal stability of D-PMSG is analyzed with the eigenvalue analysis method.

The phasor measurement units (PMUs) are the eyes for the dynamic of the power system. However, the measured phase angle data from PMUs for a transmission line (TL) may coexist with different deviations, i.e., the deviations in voltage/current phase angle difference (DVPAD/DCPAD) between different ends of the TL, and the deviations in power factor angle (DPFA, i.e., the deviations in phase angle difference between voltage and current at the same end), which threaten the advanced application with PMUs. This article proposes a correction method for the phase angle data when the DVPAD/DCPAD and DPFA coexist for the TL with unknown line parameters. Specifically, first, the abnormal voltage phase angle differences and power factor angle anomalies are discussed, and the corresponding illustrative examples are presented. Second, the mathematical model for estimating phase angle deviations in a general case is presented. Third, a "two-step" method to correct the phase angle data of PMU for the TLs with unknown parameters is proposed, in which the DVPAD/DCPAD between two ends is estimated in advance, and then, the DPFA is estimated. Finally, simulation and examples with field data verify the feasibility and engineering applicability of the proposed method.

The rapid development of intelligent substation makes the monitoring data of substation monitoring system constantly increase, and data anomalies occur frequently, which seriously affects the safe and stable operation of substation. Therefore, it is urgent to clean the monitoring data and improve the data quality. In view of this, the research status of data cleaning technology is summarized. Specifically, firstly, the substation monitoring system and its data problems are described. Then, the common data cleaning methods are summarized and compared from three aspects: correlation analysis, abnormal data detection and data repair. Finally, the problems encountered in the research are analyzed and the outlook is made.