Centralised vs. decentralised federated load forecasting in smart buildings: Who holds the key to adversarial attack robustness?

Energy and Buildings, Journal Year: 2024, Volume and Issue: 324, P. 114871 - 114871

Published: Oct. 4, 2024

Language: Английский

---

Towards an Explainable Artificial Intelligence approach for smart grid systems

Discover Artificial Intelligence, Journal Year: 2025, Volume and Issue: 5(1)

Published: April 26, 2025

Language: Английский

---

A review on the mathematical models of thermostatically controlled load

Architectural Intelligence, Journal Year: 2024, Volume and Issue: 3(1)

Published: Sept. 29, 2024

Language: Английский

---

Multi agent framework for consumer demand response in electricity market: Applications and recent advancement

Sustainable Energy Grids and Networks, Journal Year: 2024, Volume and Issue: unknown, P. 101550 - 101550

Published: Nov. 1, 2024

Language: Английский

---

Securing IoT Devices Using Generative AI Techniques

Advances in information security, privacy, and ethics book series, Journal Year: 2024, Volume and Issue: unknown, P. 219 - 264

Published: July 26, 2024

Generative artificial intelligence (GenAI) is a part of which has the ability to generate content in various formats ranging from text videos and images audio formats. GenAI inherently learn large datasets can produce results that be optimal use case cybersecurity. In current digital landscape, we see plethora electronic gadgets connected this seamless network devices online. These were earlier unable connect due lack ip addresses are now able improving quality human life home appliances health domain. From here emergence smart networks at one side boon but same time they have risk exploitation with unexpected cyberattacks. Hence, chapter an effort highlight issues concerning cyberthreats advice on how utilized mitigate these risks. This focused applying generative AI secured IoT devices. By discussing core concepts security, such as device authentication access control, demonstrated next-generation models, including GANs VAEs, boost anomaly detection for security. The also provided examples real-life cases illustrate used optimize energy grid, protect data privacy, strengthen cybersecurity efforts. Additionally, presented key related ethical considerations pertaining bias, accountability development deployment responsible AI. Moreover, it introduced legal aspects privacy legislation, protection, compliance. Finally, outlined some future trends security name few enhanced threat detection, privacy-preserving multimedia processing, secure communications. then encourages organizations start using enable systems become proactive about reduce massive onslaught cyber threats while navigating ever-evolving landscape.

Language: Английский

---

Revolutionizing smart grid security: a holistic cyber defence strategy

Frontiers in Artificial Intelligence, Journal Year: 2024, Volume and Issue: 7

Published: Dec. 3, 2024

The country's energy infrastructure is a national asset inextricably linked to progress [1]. Old grids are stiff, fail balance loads, and have significant risk of cascading failures, making them unsuitable for current times [2]. Other difficulties include interoperability scalability, high costs, data privacy, security. They also face legacy system dependencies, regulatory compliance issues due outmoded capabilities [3]. Transitioning smart grid enables dynamic solutions load management, self-healing capabilities, decentralized decision-making.As help us move away from issues, the inclusion new-generation technology makes prone cyberattacks. In many developed developing countries, bring hope strengthening sector by providing clean that meets future goals both political economic classes. However, when negligence occurs while introducing these futuristic systems, they usually result in inheriting along with vulnerabilities arising cyberspace. context Indian institutions, often demonstrate weak approach inefficient environments, susceptible attacks adversaries on their [4].Espionage, an ancient form warfare, becomes particularly lethal individual citizen country as whole, combined individually motivated attackers. Between 2021 2022, there were reports Chinese government-linked hackers attempting infiltrate steal government well major players within power 2019, Venezuela struggled attack not only its technical aspects but through cyberattacks, leaving prolonged blackouts [4]. Similarly, 2015, Russian targeted Ukraine's [5], pattern continued during 2023-24 conflict between two nations.These well-coordinated, synchronized, executed level professionalism, leading day-long outages. effects go beyond losses, which range millions billions dollars, imperil lives.All precautions necessitate understanding most prevalent security face, can serious consequences operation integrity: Network Attacks: These mostly target network operators, plants, utility businesses. This disrupt delivery causing disruption potentially obtaining ransom payments.  Breaching Sensitive Customer Data: client may be compromised adversaries, so posing privacy concerns. Malware Propagation: readily permeate affecting operations widespread disruption. Distributed Control Devices: According reports, attackers exploit distributed control devices take over or impair without authorization.Smart varying degrees vulnerability; however, types attacks: At consumer access level, meters disadvantage serving gateway collecting transmitting about consumption. meters, if infiltrated, would represent breaches illegal access, allowing tamper even services. Another highly sought-after site communication supports communication, whether wireless networks [9]. Here, might transit jeopardize control, resulting operational pandemonium. Such SCADA crucial ability destabilize manipulate functionalities. Decisions made at company operator levels. A company's fail, disruptions electricity distribution customers type multi-layered vulnerability framework has been shown comprehensive designs protect numerous changing threats. It indicates deeper knowledge threats critical improving resilience ensuring dependable preventing potential disruptions.With introduction new methods appropriate procedures, more complex safeguard itself. it understand proper implementation always necessary. PSU large organization requires recurring multiple permissions, such raising tickets accessing ports common channels, assuring across organizations. But processes steps, assessments, evaluations, managerial approvals. complexity length results delay irritation frustration towards among developers, architects, other team working project. implies actors will need ask faster doesn't bureaucratic red tape inadvertently introduce vulnerabilities.The Holistic Cyber Defence Interaction (HCDI) technique solve creating collaborative environment entire business works together develop best answers. HCDI aims combine human-AI interaction powerful Deep Learning (DL) graph-based algorithms ensure measures resilient, comprehensive, efficient, streamlined. would, extent, uniformity process decrease number approvals assessment review automated. claims basis Policy Mechanisms written. Thus, less human error omission. enable organizations sustain robust cyber strengths still maintaining pace efficiencies multi-dimensional concerted effort.The represents pioneering synergistically combines advanced methodologies semisupervised anomaly detection, deep representation learning, specification analysis, adaptive real-time learning ensembles attention mechanisms. designed way enhancement cybersecurity simplification terms Policies mechanisms implement any leaks.The Technique like wrapper around available Frameworks today. focuses base upliftment pillar. includes takes motivation several starts collection development scalability.Implementing involves key steps effective deployment integration frameworks:To train detection models effectively grids, datasets gathered must diverse collected various components grid, substations systems. Some points considered include:1. Diverse Operational Scenarios: Datasets should scenarios, including peak off-peak hours, maintenance periods, different weather conditions, few mentioned. important diversity features dataset create model handle real-world variabilities. 2. Historical Having historical necessary capture longterm trends patterns. Exploratory Data Analysis (or commonly abbreviated EDA) behaviour identify outliers deviation anomalies limitation 3. Formulating Mechanisms: Before moving ahead formulate policies what allowed. built implements gaps left out. step phase maintain matrix. 4. Real-Time Integration: Build gathering integration. continuous updating dataset. helpful bringing newly detected anomalies, thus enhancing adaptability 5. Anomaly Authorization: Use mechanism approving verifying recently discovered before adding guarantee training uses pertinent validated abnormalities, accuracy dependability system. 6. Sources: Making use sources inside grid-such sensors, networks-allows one multi-source method give full picture situation grid. We consider additional factors could lead source corruption consequent loss relevance development. Integrating into given frameworks information uniform harmonious responding framework. Advanced AI ML techniques threat identification prediction clustering, learning.TensorFlow PyTorch used strong based Scikit-learn. Implement computing platforms Apache Kafka Spark analyze efficiently. Additionally, SIEM systems Splunk ELK Stack centralized further analysis optimized response. From tools techniques, support overall responsiveness against compliant skillful employees make successful sustained secured resilient operation.Design scalable large-scale deployments. Plan adaptation evolving technological advancements cybersecurity. Here's example outlines how AWS service open-source here - procedures undertaken companies businesses, posture increase, reduce approval process, let collaboration prevail threats.This encapsulates approach, hybrid combined, calls unite driven elastic solution grids. strategy assures improvement.This actually simultaneously threatens this regard, emphasized study practice extensive measures, particular technique, obligatory. integrating realtime stands ensures timely, incident response expediting encouraging cross-functional cooperation. When implemented, become reliable protecting infrastructures

Language: Английский

---

Quantum tomographic reconstruction: a Bayesian approach using the extended Kalman filter

Published: April 19, 2024

Language: Английский

---

Advancing energy efficiency: Harnessing machine learning for smart grid management

E3S Web of Conferences, Journal Year: 2024, Volume and Issue: 524, P. 01003 - 01003

Published: Jan. 1, 2024

The concept of Smart Grids (SG) has emerged as a solution to address challenges in traditional power systems, including resource inefficiency, reliability issues, and instability. Since its inception the early 21st century, Grid technology undergone significant development, integrating advanced information communication automation technologies with conventional infrastructure. This integration enhances efficiency, reliability, sustainability, while enabling renewable energy sources optimizing distribution consumption. Machine learning algorithms play pivotal role development Grids, facilitating consumption prediction, optimization, anomaly detection, fault diagnosis. paper explores methodologies for developing improving machine efficient prediction management within Grids. It discusses application deep techniques, reinforcement learning, Internet Things (IoT) enhance systems. study highlights potential impact convolutional neural networks (CNNs) on regulation emphasizes need further research associated model complexity data requirements contexts.

Language: Английский

---

Green Electricity and Green Hydrogen Coupled Coal Chemical System: Optimization Scheduling with Multi-Operating Conditions in Load Modeling Considering Flexibility

Published: Jan. 1, 2024

Language: Английский

---

An effective ensemble electricity theft detection algorithm for smart grid

IET Networks, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 2, 2024

Abstract Several machine learning and deep algorithms have been presented to detect the criminal behaviours in a smart grid environment recent studies because of many successful results. However, most for electricity theft detection their pros cons; hence, critical research issue nowadays has how develop an effective algorithm that leverages strengths different algorithms. To demonstrate performance such integrated model, proposed first builds on neural networks, meta‐learner determining weights models construction ensemble then uses promising metaheuristic named search economics optimise hyperparameters meta‐learner. Experimental results show is able find better outperforms all other state‐of‐the‐art compared terms accuracy, F1‐score, area under curve precision‐recall (AUC‐PR), receiver operating characteristic (AUC‐ROC). Since can improve accuracy algorithms, authors expect it will be used learning‐based applications.

Language: Английский

---