Study on Multi-source Data-Driven Static Security Risk Assessment of Power Grids

Due to the high probability of power flow violation during the N-1 static safety check under maintenance, and the risk assessment indicators are single and subjective, a method of static security risk assessment based on multi-source heterogeneous information for planned maintenance is proposed. By regularizing dispatcher experience, the static security factors are analyzed from equipment failure rate, electrical characteristics and grid topology, etc. Then, the selected indicators are preprocessed. Finally, Deep Belief Neural Network (DBNN) is used to evaluate risk. Through deep mining of the membership and mapping relationship between multiple indicators, the risk self-assessment is realized, and dispatcher is given the assistant decision when orderly adjusting equipment that causes limit. Simulation results of IEEE 39-bus show that the method proposed in this paper can quickly assess risk level without analysis of dispatcher; compared with DNN algorithm, DBNN has higher accuracy. The validity and feasibility of this research are proved.

Due to the lack of the support from the main grid, the islanded hybrid ac/dc microgrid will more likely cause the power flow severe fluctuation and even the microgrid to be split once a serious contingency occurs. To maintain the security and stability of the system under such a condition, this paper proposes a new static security risk assessment of the islanded hybrid ac/dc microgrid considering that the hybrid ac/dc microgrid was split. First, this paper presents a static security analysis criterion to determine the post-contingencies splitting scheme and a new load-shedding strategy to achieve power balance of subsystems. On this basis, a probabilistic power flow calculation method based on Levenberg-Marquardt with a nonmonotone line search (LMNL) algorithm is applied to calculate the probabilistic power flow of the islanded subsystems. Finally, the probability distribution of static security risk assessment indexes, including the dc loss load risk, ac loss load risk, and interlinking converter (ILC) active power limit violation risk, are obtained. The simulation results on the test system demonstrate the necessity of considering the hybrid ac/dc microgrid was split, and prove the feasibility and effectiveness of the proposed method.

Static risk factors have been viewed as relatively fixed aspects of offenders’ histories, such as age and the extent of previous offending, which increase the risk of reoffending but do not change and cannot be changed through intervention. Given that they are “fixed” historical characteristics, static risk factors are not typically seen as meaningful targets for intervention. In contrast, dynamic risk factors (DRFs) have been viewed as possible psychological or behavioral features of the offender that raise the risk of reoffending but maybe potentially changeable. The notion that some dynamic risk factors may be relatively stable for sexual offenders in that change occurs over extended periods of time, while other DRFs are less stable and may change acutely based on immediate contexts or affective experiences. Some authors have suggested that both proposed static and dynamic risk factors each are markers of underlying psychological vulnerabilities to commit sexual offenses and others have argued that such vulnerabilities or needs reflect propensities to act out sexually. In this chapter, two efforts to identify and quantify dynamic risk factors are reviewed that have culminated in two risk assessment instruments, the Stable-2007 and the SRA-FV. The results of the research efforts suggest that each of these instruments provides a useful measure of characteristics of sexual offenders that are related to sexual offense recidivism; they have been shown to have incremental validity relative to so-called static risk assessments. However, instead of being truly “dynamic” and changeable, such risk factors may simply reflect a more general or broader dispositional characterization of sexual offenders and one that surprisingly captures less variance than the more concrete manifestation of underlying psychological vulnerabilities as expressed in static, historical behaviors.

Methodologically-sound security assessments are crucial for understanding a system in fulfilling the requirements, realizing its behavior, and identifying implications. A system is made resilient if and only if there is enduring confidence that it will function as expected. It is cyber secure if it displays this property in the face of an adversary. This chapter provides an explanation of various static security risk assessment methodologies (SSRAM) having long epochs for assessing and revisiting the risks and explains their strengths and weaknesses. The SSRAM will form the basis for elaborating on dynamic security risk assessment methodologies that must have very short epochs depending on the emergent threats and vulnerabilities to combat new or evolving threats. This is essential for handling cyber security of dynamic complex systems including future networks having a number of self-managing properties including self-protection for defending against malicious attacks.

Visualization in scientific computing has been a developing computer technology in recent years. It has been used in many fields because of its strong ability in large data management and display, while its applications in power systems are still in the study stage. The method introduced in this paper combines geographic images and power system information, and gives a more accessible and realistic visual expression of static security assessment. This method applied ArcViewGIS software (a product of ESRI) to the traditional static security assessment algorithm. Using the function of image overlapping, automatic image making and large-scale data processing, the method turns the results of security assessment into outstanding pictures and tables on a geographic map. In this way, one can obtain a more efficient power system security assessment method.

According to the growth rate of Machine Learning (ML) application in some power system subjects, this paper introduce a comprehensive survey of Artificial Neural Network (ANN) in Static Security Assessment (SSA). Advantages and disadvantages of using ANN in above mentioned subjects and the main challenges in these fields have been explained, too. We explore the links between the fields of SSA and NN in a unified presentation and identify key areas for future research. Recent developments in the solution methods for SSA are reviewed. Hybrid techniques in SSA are also discussed and reviewed and future directions for research are suggested.

In recent years, with a rapid growth of the generation of hydropower, wind and solar power, renewable energy has occupied more than 60% of the total installed capacity of power grids in Qinghai, Yunnan, Sichuan, Tibet and other areas. Along with the yearly increased power transmission from these provinces, the ultra-high voltage AC/DC hybrid power grids will be evolving to be the multi-DC asynchronous grid, which is a trend of future represented by the work done in Yunnan. Taking Yunnan province as an example, this article studies the operational problems related to the multi-DC asynchronous power-sending grid with high proportion of renewable energy, and describes a new secure and stable operational technology that has been applied into the grid in Yunnan. We case-study the effects of the operation after the application of such technology and evaluate its practical influences.

Compared with other conventional energy sources, renewable energy, which results in less pollution and has numerous resources, is a significant factor in addressing the current issues of the serious environmental pollution and the resource depletion. Large-scale renewable energy connected to the grid could lead to a the change of the morphological structure and the operation modes of energy transmission. While current power grid evolution models did not consider the characteristics of renewable energy in the evolution strategy. The expansion mechanism of the future transmission network with a high proportion of renewable energy is studied in this paper. The operation cost including renewable energy curtailment are optimized using the production simulations, where the stochastic and intermittent of wind and solar energy are modeled. And the dual variable information is adopted to drive the grid upgrading. The expansion strategies for power grid of Zhejiang province in China are analyzed to balance the construction cost and curtailment cost of renewable energy.

The secure integration of renewable generation into modern power systems requires an appropriate assessment of the security of the system in real-time. The uncertainty associated with renewable power makes it impossible to tackle this problem via a brute-force approach, i.e. it is not possible to run detailed online static or dynamic simulations for all possible security problems and realizations of load and renewable power. Intelligent approaches for online security assessment with forecast uncertainty modeling are being sought to better handle contingency events. This paper reports the platform developed within the iTesla project for online static and dynamic security assessment. This innovative and open-source computational platform is composed of several modules such as detailed static and dynamic simulation, machine learning, forecast uncertainty representation and optimization tools to not only filter contingencies but also to provide the best control actions to avoid possible unsecure situations. Based on High Performance Computing (HPC), the iTesla platform was tested in the French network for a specific security problem: overload of transmission circuits. The results obtained show that forecast uncertainty representation is of the utmost importance, since from apparently secure forecast network states, it is possible to obtain unsecure situations that need to be tackled in advance by the system operator.

The high proportion of renewable energy into the power grid, say, the large-scale wind power generation connected to the power grid, is the key approach to achieve the clean and low-carbon energy transition. Under the premise of ensuring power grid's safe and stable operation, it is necessary to study the influence of wind power capacity expansion on the transient angle stability (TAS). Aiming at several typical dominant modes in Qinghai power system, which holds many new energy resources, the sensitivity of wind power capacity expansion in different locations to the TAS, as well as its key influence factors and feasible explanations, are explored. It may lay the foundation of the location optimization of wind power capacity expansion.

In this paper, a knowledge-based system for static security assessment of power system is proposed and investigated. An expert system (ES) for screening the consequences of transformers' and electrical lines' single outage was developed. The rules derived from a simulation model, irrespective of the size of the power system, are thus obtained. A dedicated artificial intelligence language (Prolog) for rules implementation was used to build the ES named CONTIG. The results show that such ES can be very effective for contingencies' selection. The benefits for a moderate size power system (25 buses and 30 branches) were analyzed using CONTIG. This approach has a great potential for online application

This study proposes a new paradigm for real-time, predictive, and multidimensional risk assessment in industrial environments by leveraging multimodal data (video, audio, environmental sensors). Existing risk assessment systems typically rely on single data sources or subjective judgment, making it difficult to adapt swiftly to complex workplace changes and achieve real-time responsiveness. To address these limitations, we constructed a large-scale multimodal dataset of approximately 10TB, employing real-time streaming-based preprocessing and data synchronization. This approach integrates various data sources—high-resolution cameras, high-sensitivity microphones, and environmental sensors (temperature, humidity, vibration)—and applies data augmentation techniques such as AutoAugment and MixUp to build robust models capable of handling diverse environmental conditions. We adopted a hybrid analytical algorithm combining Vision Transformer (ViT) and YOLOv8, achieving high accuracy (over 95%) and real-time processing (average response time under one second). Additionally, we utilized machine learning algorithms such as SVM, Random Forest, and K-Means to detect anomalies in audio and environmental sensor data, thus identifying latent risk factors. Experimental results demonstrate multifaceted performance improvements compared to conventional approaches, including over a 15% increase in accuracy, approximately 30% reduction in response time, about 20% reduction in power consumption, and user satisfaction exceeding 90%. These achievements were verified across various industrial settings—chemical, manufacturing, logistics—highlighting the system’s capacity to detect complex risk factors and respond proactively. By seamlessly integrating multimodal data analysis, state-of-the-art deep learning models (ViT, YOLOv8), and reinforcement learning-based response strategies, we have demonstrated a transition from traditional, static, and retrospective risk assessment to an intelligent, real-time, and predictive safety management framework.

All nuclear power plants must consider external flooding hazards such as local intense precipitation, riverine flooding, flooding due to upstream dam failure, and coastal flooding due to storm surge or tsunami. While external flooding events could potentially interrupt off-site power, threaten plant structures, systems and components important to safety, or limit plant access, they have often been qualitatively assessed as risk insignificant and screened out from detailed evaluation and quantification. Recent lessons learned from the Fukushima seismic/tsunami initiated nuclear accident (2011), the Fort Calhoun (2011), Vermont Yankee (2013), Arkansas Nuclear One (2013), and St. Lucie (2014) flooding events have highlighted the need for more detailed risk analysis. However, incorporating an external flood model into the traditional event tree/fault tree approach used in static probabilistic risk assessment (PRA) models is challenging because it is difficult to accurately represent plant system and component behavior and reliability of manual actions during an ever-progressing flood event. The plant response to external flood may be highly spatial and time dependent, subject to the hydrological and hydraulic characteristics of the flood event. Such unique challenges prompt the investigation of using simulation-based dynamic analysis approaches for external flood risk assessment. Simulation methods can better model the performance of structures, systems, and components during an external flooding event. A general framework to perform a simulation-based dynamic flooding analysis is presented in this paper with the subtasks of flood hazard analysis, flood fragility analysis, plant response modeling, safety margin analysis or PRA quantification. A new type of PRA technique, State-based PRA Modeling, is introduced to incorporate time-related interactions such as those from both 3D physical simulations and random failures into traditional PRA logic models.

BACKGROUND Timely and accurate prediction of short-term mortality is critical in intensive care units (ICUs), where patients’ conditions change rapidly. Traditional scoring systems, such as the Simplified Acute Physiology Score and Acute Physiology and Chronic Health Evaluation, rely on static variables collected within the first 24 hours of admission and do not account for continuously evolving clinical states. These systems lack real-time adaptability, interpretability, and generalizability. With the increasing availability of high-frequency electronic medical record (EMR) data, machine learning (ML) approaches have emerged as powerful tools to model complex temporal patterns and support dynamic clinical decision-making. However, existing models are often limited by their inability to handle irregular sampling and missing values, and many lack rigorous external validation across institutions. OBJECTIVE We aimed to develop a real-time, interpretable risk prediction model that continuously assesses ICU patient mortality using irregular, longitudinal EMR data, with improved performance and generalizability over traditional static scoring systems. METHODS A time-aware bidirectional attention-based long short-term memory (TBAL) model was developed using EMR data from the MIMIC-IV (Medical Information Mart for Intensive Care) and eICU Collaborative Research Database (eICU-CRD) databases, comprising 176,344 ICU stays. The model incorporated dynamic variables, including vital signs, laboratory results, and medication data, updated hourly, to perform static and continuous mortality risk assessments. External cross-validation and subgroup sensitivity analyses were conducted to evaluate robustness and fairness. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), accuracy, and <i>F</i><sub>1</sub>-score. Interpretability was enhanced using integrated gradients to identify key predictors. RESULTS For the static 12-hour to 1-day mortality prediction task, the TBAL model achieved AUROCs of 95.9 (95% CI 94.2-97.5) and 93.3 (95% CI 91.5-95.3) and AUPRCs of 48.5 and 21.6 in MIMIC-IV and eICU-CRD, respectively. Accuracy and <i>F</i><sub>1</sub>-scores reached 94.1 and 46.7 in MIMIC-IV and 92.2 and 28.1 in eICU-CRD. In dynamic prediction tasks, AUROCs reached 93.6 (95% CI 93.2-93.9) and 91.9 (95% CI 91.6-92.1), with AUPRCs of 41.3 and 50, respectively. The model maintained high recall for positive cases (82.6% and 79.1% in MIMIC-IV and eICU-CRD). Cross-database validation yielded AUROCs of 81.3 and 76.1, confirming generalizability. Subgroup analysis showed stable performance across age, sex, and severity strata, with top predictors including lactate, vasopressor use, and Glasgow Coma Scale score. CONCLUSIONS The TBAL model offers a robust, interpretable, and generalizable solution for dynamic real-time mortality risk prediction in ICU patients. Its ability to adapt to irregular temporal patterns and to provide hourly updated predictions positions it as a promising decision-support tool. Future work should validate its utility in prospective clinical trials and investigate its integration into real-world ICU workflows to enhance patient outcomes. CLINICALTRIAL

Chapter 9 - Modeling flexibility in electric power market