Detection and localization of false data injection attacks based on multi-scale feature fusion and attention enhancement network in smart grid

Assessing the performance of the smart-grid system (SGS) under uncertainty is essential for ensuring a reliable energy supply from the perspective of the grid operator. In this study, a multistate smart-grid network (MSGN) is developed to evaluate the delivery capability of the SGS. An MSGN consists of multiple interconnected facilities, where nodes represent energy sources or converters and arcs denote feeders. The output of each facility in the MSGN is modeled as multistate, as maintenance activities and partial failures can result in multiple possible output levels. During power delivery, transmission losses may arise due to heat dissipation and feeder aging, potentially resulting in insufficient power supply at the demand side. From a smart-grid management perspective, delivery reliability, defined as the probability that the MSGN can successfully deliver sufficient power from energy sources to the destination under transmission loss, is adopted as a performance index for evaluating SGS capability. To compute delivery reliability, a minimal-path-based algorithm is developed. A practical SGS is presented to demonstrate the applicability of the proposed model and to provide managerial insights into smart-grid performance and operational decision-making.

Machine learning-based real-time outage fault detection for distribution networks in smart grid

Cooperative trading in smart-grid networks

The growing adoption of Electric Vehicles (EVs) has brought new challenges in energy management, charging infrastructure, and efficiency. Traditional plug-in charging systems require manual connection, long charging times, and significant stationary downtime, which limit EV convenience and scalability. To overcome these challenges, this project — “Wireless EV Charging with V2V Support” — proposes an innovative solution that combines wireless power transfer with vehicle-to-vehicle (V2V) energy sharing capabilities. The system employs inductive-coupling-based wireless charging technology, enabling efficient energy transfer without physical connectors. The transmitter coil embedded in the charging pad and the receiver coil installed in the EV operate based on the principle of electromagnetic induction, allowing safe and contactless charging. A microcontroller-based control unit regulates the power flow, monitors charging parameters such as voltage, current, and coil alignment, and ensures system safety through feedback and protection mechanisms. In addition to stationary wireless charging, the system introduces V2V support, which enables one EV with sufficient charge to transfer energy wirelessly to another vehicle in need. This peer-to-peer energy exchange not only extends the operational range of EVs but also minimizes dependency on fixed charging stations. The implementation of communication modules, such as WiFi or Bluetooth, allows vehicles to coordinate charging operations and monitor energy transfer status in real-time. The project demonstrates an effective prototype of a smart, flexible, and sustainable charging ecosystem, suitable for modern urban mobility. It addresses key limitations of wired charging by reducing wear and tear, enhancing user convenience, and improving charging safety under all weather conditions. Moreover, its modular architecture supports future integration with IoT-based monitoring systems, renewable energy sources, and smart grid networks, making it scalable and environmentally friendly.

ABSTRACT Electrical power systems are critical to modern society but are vulnerable to faults that can cause interruptions and hazards. Timely and reliable fault detection is essential for minimizing losses and ensuring efficient operation. This paper presents TCN-TransNet, a hybrid deep learning architecture for intelligent fault detection in electrical power networks. The model integrates Temporal Convolutional Networks (TCNs) and Transformer-based modules to capture both local and global dependencies in multivariate time-series signals (current and voltage) indicative of various faults. TCNs employ causal and dilated convolutions to model short-term dependencies, while the Transformer module uses self-attention mechanisms to capture long-range dependencies and interphase relationships. Fault types including single-line-to-ground, line-to-line, and three-phase faults are classified via a Softmax layer. Hyperparameter tuning with the Rectified Adam (RAdam) optimizer ensures stable convergence and effective training. Experimental results demonstrate that TCN-TransNet achieves 98.75% accuracy, 98.15% precision, 98.22% recall, and 98.21% F1-score, outperforming traditional fault detection methods. The framework is particularly suitable for smart grid networks and online monitoring systems, enabling fault prevention, reduced response times, and enhanced reliability. These results confirm the scalability and robustness of TCN-TransNet in improving the performance and stability of modern electrical power systems.

In the world today the distribution of electricity has evolved to the point where electricity is been delivered through smart grid network, in other to harness the full power and efficiently too, smart equipment is required, the 100A smart distribution board is one of this equipment. The aim of this work is to develop a smart, modern, and intelligent distribution board with high efficiency capable of handling current up to 100A, the advance circuit protection mechanism ensures safety to the end users and the electrical system, the system reduces down-time and improves reliability by using real-time fault and notification systems and also by the incorporation of internet of things (IOT) the use of mobile application or web interface can be used for remote control and monitoring of the distribution board is achieved. With the use of cutting-edge components, state of the art materials and advanced control the resulting smart distribution board has achieved a much more efficient, user friendly and safer electrical power distribution to end users. This work presents a step forward in modernizing electrical power distribution systems.

Abstract The proliferation of connected devices has led to a paradigm shift in cellular standards, typified by the long-term evolution (LTE) standard. The fifth-generation (5G) standard supportsnumerouspromising mobile technologies, including Machine-to-Machine (M2M) and Device-to-Device (D2D) communication, which enable the communication of a large number of intelligent devices and ubiquitous devices. The deployment of M2M devices in the application of smart grid (SG), specifically in power systems, has introduced new compromising challenges in the areas of resource allocation and interference management. The 5G cellular network’s qualityofservice (QoS) and performance deteriorate due to interference brought on by the reduced inter-cell distance and the smooth integration of heterogeneous devices.In this paper, an interference-aware multi-objective particle swarm optimisation(MOPSO)scheme is proposed for M2M communication in SG to mitigate the interference generated as a result of the localisation of M2M devices on the grid. In order to evaluate performance, the MOPSO approach was developed and implemented for smart grid situations based on pre-fault, during-fault, and post-optimisation conditions. The initial step was to use the multi-objective particle swarm optimisation (MOPSO) algorithm to optimise the smart grid network in order to decrease grid interference. According to simulation results, under different pre-fault and post-optimisation settings, the system throughput and signal-to-interference-to-noise ratio (SINR) were greatly increased by 32.69 and 21.94%, respectively. Furthermore, by using MOPSO, the fault clearance time was reduced by 106.06%, reducing the amount of time needed to clear an impending fault with interference. Additionally, the smart grid network’s power loss was improved and maintained at levels comparable to those of the pre-fault conditions. In the subsequent step, the performance of the developed MOPSO technique was compared to that of the non-dominated sorting genetic algorithm (NSGA-II) in terms of convergence in fault clearance time, SINR, and throughput. Simulation results indicated that, in comparison to NSGA-II’s performance, MOPSO throughput and SINR were enhanced by 5.93%, 4.65%, and 0.96%, respectively. In comparison to NSGA-II, the proposed MOPSO converges to the optimal solution more quickly for the various objective functions. The findings provided by the developed MOPSO demonstrate that it can efficiently compete with similar algorithms when tackling problems involving interference optimisation algorithms.

A Generalized Stealth Attack Localization Framework for Smart Grid Network Under Real-Time Test Environment

ABSTRACT Smarter grids depend on Cyber‐Physical Systems (CPS) to merge physical energy distribution networks with computational intelligence, because these systems optimize reliability and sustainability and power delivery efficiency. CPS in smart grids present both enhanced interconnectivity and complexity which creates substantial security challenges because they become exposed to complex cyber‐attacks that harm operational processes and degrade data integrity criteria. The current intrusion detection systems in smart grid environments encounter multiple obstacles when they attempt to detect and counteract security threats effectively. This paper develops an innovative security solution by integrating Zero‐DAGNet with POCO as a solution to combat smart grid security challenges. The Zero‐DAGNet employs domain‐adversarial learning techniques that operate inside a graph‐based deep‐learning structure for identifying complex network entity relationships. The designed structure helps the model adapt to unidentified attack patterns which resolves domain shift problems encountered during intrusion detection operations. The POCO brings forth an innovative optimization technique based on primate cognitive operations that optimizes network parameter settings efficiently. Through this well‐merged structure, the model demonstrates enhanced flexibility and operational performance when operating in dynamic smart grid networks. Results from empirical tests confirm that the combination of Zero‐DAGNet and POCO produces effective outcomes. On ICS and SWaT and CICIDS17 benchmark data sets, the proposed model demonstrates superior performance than traditional and deep‐learning machine‐learning algorithms. Using the ICS data set allows the framework to reach a 99.10% accuracy and a precision of 98.89% while producing a recall of 98.88%, which results in an F 1‐score of 99.08%. It shows improved performance compared to previous solutions. The Zero‐DAGNet + POCO approach demonstrates its capability to deliver resilient and efficient intrusion detection solutions which strengthen the security features of smart grid networks.

This paper presents a novel deep learning scheme for power load prediction in smart grid networks, combining temporal modeling with adaptive feature integration to tackle the complex dynamics of electricity consumption. The proposed scheme features a hybrid architecture that merges recurrent neural networks with attention mechanisms, enabling simultaneous capture of long-term load patterns and dynamic weighting of external influences like weather conditions and temporal features. Moreover, the model incorporates specialized preprocessing to decompose load data into periodic and volatile components while employing robust normalization techniques to handle non-stationary behavior. Then, a dual-objective loss function is used to enhance both prediction accuracy and resilience to outliers, supported by adaptive optimization with regularization. Simulation results are provided to demonstrate the proposed scheme’s superior performance, achieving 96.1% prediction accuracy with 5 hidden layers. The attention mechanism proves particularly effective, reducing weather-related prediction errors by 22% while maintaining faster convergence rates than conventional methods. This comprehensive solution offers grid operators a reliable tool for demand-side management, renewable integration, and operational planning in modern power systems.

ABSTRACT Smart Grid facilitates two-way communication and promises features, such as self-healing, efficiency, and resilience. All the advancements in Smart Grid technology require frequent communication with the smart meter over a public channel, exposing it to several security challenges. Additionally, as real-time power consumption data is transmitted periodically over Smart Grid networks, it becomes crucial to protect consumers’ privacy and confidentiality. Authentication, key management, and encryption techniques are crucial in Smart Grid communications. This paper provides an extensive survey of Authentication and Key Agreement (AKA) and data aggregation techniques used in Smart Grid communications. An overview of Smart Grid architecture and potential security vulnerabilities and privacy threats are also discussed. Furthermore, this paper outlines our perspective on security challenges and lesser-explored security technologies in the Smart Grid.

A Smart grid is generally an electrical grid combined with the best of digital technologies. The benefit of this shift, though very dependent, efficient, and sustainable, also has a lot of cybersecurity risks. Modern energy systems are based on the smart grid, which is vulnerable to hackers and hence threatens the safety of public and economic activities. In this study, we provide a thorough analysis of Smart Grid cybersecurity concerns. One of the major issues with a smart grid network has been security; up until now, cybersecurity has been the main factor to be taken into account. Again, it took a lot of investigation to uncover those security flaws. Cybersecurity issues are continuously changing, especially those related to do with privacy, connectivity, and security management. Modern cybersecurity technology and best practices are mostly borrowed from the traditional telecommunications sector due to its laxer availability and safety standards. The oil and gas industry can provide very valuable inputs on how to handle all operational integration security issues; however, the smart grid faces a very different reality, with an extremely high number of end-users and very high geographic dispersion. Global growth in electricity demand necessitates the need to preserve the reliability, robustness, and safety of energy infrastructure, hence shielding smart grids against cyberattacks. Whoever is responsible will be able to adequately protect this vital piece of modern civilization's infrastructure against smart grid cybersecurity and, moving forward, implement remedial strategies before the threats. Only with strong partnerships working and striving in the right direction will we be able to address all aspects of smart grids associated maintenance challenges, delivering on that promise. A brief discussion on major cybersecurity concerns to a smart grid and some strategies for risk reduction is covered in the forthcoming section of this research.

The paper presents a novel AI-based simulation model that can be applied in the optimization of sustainable energy distribution in Smart Grids (SG). The framework integrates the energy demand forecasting, model of renewable energy generation, and online grid optimization relying on the features of the recent machine learning algorithms, including Random Forest Regressor. The model is trained on a synthetical dataset, which contains seasonal variability of demand, and renewable sources of power which comprise of solar and wind power. The performance of the demand forecasting model has also been evaluated in key measures with the coefficient of determination (R2) of 0.80 and mean squared error (MSE) of 7.25 as good forecasting performance of the model. Otherwise, a Watts-Strogatz graph is a simulated software that is designed to represent a smart grid network, with consumers, producers and storage units being the nodes. The allocation of resources is also efficient and the energy allocation within the network is optimized based on the nature of the nodes and their carrying capabilities. Real-time simulations suggest that the system maintains a balance between supply and demand of 100 percent, and renewable energy share of 46.5. The results emphasize the AI possibilities in enhancing Smart Grids performance with the integration of real-time monitoring, optimization, and detection of anomalies. This framework forms a platform onto which the Smart grid applications of the future can be designed with the focus on sustainability and efficiency of operation.

Fault detection is critical to the resilience and operational integrity of electrical power grids, particularly smart grids. In addition to requiring a lot of labeled data, traditional fault detection approaches have limited flexibility in handling unknown fault scenarios. In addition, since traditional machine learning models rely on historical data, they struggle to adapt to new fault patterns in dynamic grid environments. Due to these limitations, fault detection systems have limited resilience and scalability, necessitating more advanced approaches. This paper presents a hybrid technique that integrates supervised and unsupervised machine learning with Generative AI to generate artificial data to aid in fault identification. A number of machine learning algorithms were compared with regard to how they detect symmetrical and asymmetrical faults in varying conditions, with a particular focus on fault conditions that have not happened before. A key feature of this study is the application of the autoencoder, a new machine learning model, to compare different ML models. The autoencoder, an unsupervised model, performed better than other models in the detection of faults outside the learning dataset, pointing to its potential to enhance smart grid resilience and stability. Also, the study compared a generative AI-generated dataset (D2) with a conventionally prepared dataset (D1). When the two datasets were utilized to train various machine learning models, the synthetic dataset (D2) outperformed D1 in accuracy and scalability for fault detection applications. The strength of generative AI in improving the quality of data for machine learning is thus indicated by this discovery.By emphasizing the necessity of using advanced machine learning techniques and high-quality synthetic datasets, this research aims to increase the resilience of smart grid networks through improved fault detection and identification.

In smart grid infrastructure, where Smart Meters continuously send usage data to central servers for invoicing, monitoring, and optimization, data transmission reliability is crucial. However, transmission mistakes brought on by noise, interference, and signal attenuation can impair system performance. In this study, the Cyclic Redundancy Check (CRC) is used and tested as an error detection method in Smart Meter communication networks. In comparison to conventional parity approaches, CRC offers greater safety by using polynomial division to identify single-bit and burst mistakes with high accuracy. To investigate the operational efficacy, bandwidth effect, and error detection properties of CRC, a smart grid network was simulated using Cisco Packet Tracer. These findings confirm that CRC is a cost-effective, scalable, and safe method for ensuring dependable communication in wide-area smart grid configurations.

The magneto‐mechano‐electric (MME) coupling principle plays an important role in synchronously harvesting magnetic energy and monitoring safe operation in the distribution grid and the Power Internet of Things (PIoT). In this study, an MME resonator is presented, containing two crossed U‐fingers made of piezoelectric phase‐elastic beams with different lengths and magnet masses, operating in symmetric, decoupling dual bending modes. One U‐finger resonating at 50 Hz serves as energy harvester (EH), while the other resonating at 185 Hz acts as current or magnetic sensor, enabling the resonator to simultaneously and wirelessly capture 50 Hz stray magnetic field (HAC,50 Hz) energy and ground fault message of power lines by injecting an additional non‐grid‐frequency (185 Hz) current. The EH U‐finger generates an output power of 1.53 mWRMS under a weak HAC,50 Hz of only 0.5 Oe, achieving a normalized power density surpassing current standards. While the sensing U‐finger shows a high sensitivity to HAC,185 Hz with a detectability of 710 pT, even the EH U‐finger is operating. The application test demonstrates the system's wireless monitoring and synchronous self‐powered functions, providing stable energy for self‐sensing data processing and transmission. This work introduces an efficient, wirelessly self‐powered, and self‐sensing method for advancements in PIoT.

This study delves into the vulnerability of the smart grid to infiltration by hackers and proposes methods to safeguard it by leveraging blockchain and artificial intelligence (AI). A categorization and analysis of cyberattacks against smart grids will be conducted, focusing on those targeting their communication layers. The main goal of the work is to address the challenges in this area by implementing novel detection and defense strategies. The authors categorize attacks on smart grid networks based on the communication classes they want to compromise. They propose novel taxonomies specifically designed to detect and implement defense strategies. The study investigates artificial intelligence and blockchain techniques to identify cyber-attacks that employ deceptive data injection. The study indicates that cyberattacks against smart grids are increasing in frequency and complexity. The paper proposes innovative strategies for defense, such as enhancing cybersecurity with artificial intelligence and blockchain technology. The research further enumerates several challenges, such as counterfeit topological data, imprecise data identification, and combining big data with blockchain technology. Given the increasing risks, the study emphasizes the crucial need for robust cybersecurity safeguards in smart grids. This work contributes to the protection of smart grid infrastructures by categorizing attacks, suggesting novel defenses, and exploring solutions integrating artificial intelligence and blockchain technology. Research should prioritize enhancing technology to maximize security and counter emerging attack methods. The intended audience of our paper comprises graduate-level academics and independent researchers.

With the rapid development of smart grid, power information network faces increasingly severe security threats. In smart grid, real-time security situation awareness is of great significance for preventing potential attacks. Most of the existing security situation assessment methods are based on rules or statistical analysis, lacking the deep understanding of time series data and the ability to dynamically adjust. This paper proposes a time series deep reinforcement learning (TDRL) algorithm to dynamically evaluate the security situation of power network. The TDRL algorithm constructs a reinforcement learning framework, regards the security state of power information network as the environment state, defines attack and defense strategies as actions, and combines historical data of security events for learning and optimization. Through training, the model can predict the changes in security situation in real time, and adjust according to the actual risks to improve the anti-attack capability of the power grid. The experimental part uses a real smart grid data set for simulation. The model performs better than the traditional rule-based evaluation method in multiple attack scenarios. In the comparison experiment, the TDRL algorithm has significantly improved in accuracy, detection rate and F1 value, with an accuracy rate of more than 95% and a false alarm rate of less than 5%.

The wide-ranging expansion of smart grid networks has resulted in insurmountable difficulties that must be overcome to ensure the security and reliability of crucial energy infrastructures. The information system can be subjected to threats such as cyber-attacks or hardware malfunctioning resulting in a data integrity compromise which implies that the system will consequently not operate correctly. Anomaly detection methods that are relying on centralized data aggregation are problematic to the issues of data privacy and scalability resulting from such approaches. In this paper, we present a completely distinct approach that is based on federated learning that is employed in anomaly detection of smart grid networks that makes it possible to learn collaboratively in a decentralized way and in the same time protecting user privacy through connections between many grid nodes. The method integrates multi-source information fusion, incorporating smart meter readings, IoT sensor logs, and substation performance metrics to enhance anomaly detection accuracy and robustness. Tests show that the system is among the top or the best systems that have successfully identified a wide range of anomalies, have required low communication overhead, and have exhibited scalability. These findings imply that the use of federated learning presents an attractive direction for future work on the enhancement of the security and resilience of smart grid networks amidst changing threats.