Renewable power generators typically participate as price-takers in current electricity markets, and market clearing prices are determined by cost curves of conventional generators while treating renewable generation as negative load. In this paper, we analyze a new design of electricity markets where renewable power generators are able to bid their expectation-adjusted cost curves into the market, just like the cost curves of conventional generators. We formally define the new market model, and mathematically show how the renewable cost curves should be determined for the new market to be equivalent to the current market in terms of overall system cost. We further establish that on the average, the total revenue of both conventional and renewable generators, as well as the total cost to consumers, under the proposed market is the same as the current practice. While equivalent in terms of aggregate cost and revenue measures, the new market design allows renewables to be direct participants in the bidding process and bid according to their risk-return tolerance. We demonstrate our analytical results numerically using data obtained from the the Texas grid.
An increasing number of corporations and individuals are interested in minimizing the carbon footprint of their electricity use. In order to do this, they must have access to a ’carbon signal’ that quantifies the real time emissions intensity of their supply of electricity. One promising carbon signal is Locational Marginal Carbon Emissions (LMCE). However, as LMCE is often not available, Locational Marginal Price (LMP) is often suggested as an proxy for LMCE. This paper seeks to contribute to the evaluation of LMP as a potential signal for informing carbon-aware load-shifting. We evaluate the connection between LMCE and LMP via statistical analysis on real data from the PJM system operator. Additionally, we conduct data-center load-shifting case studies using each of the signals. Our results show that while LMP and LMCE have a positive relationship for many samples, there are also extremely high-magnitude exceptions to this pattern. Driven by such examples of ’misalignment’ between LMP and LMCE, our load-shifting case study shows the potential for large emissions increases when shifting based on LMP.
Special Protection Schemes (SPS) have been widely adopted to help power grids operate closer to their stability limits, maximizing efficiency and reducing costs. While SPS improves system stability, its heavy reliance on communication networks introduces significant cybersecurity challenges. A cyberattack against the cyber layer of SPS threatens the availability and integrity of the physical layer of power systems. This paper explores the impact of a Denial-of-Service (DoS) attack on a communication router within the Wide Area Network (WAN), focusing on how such an attack can disrupt SPS operations and compromise overall system stability. The study models three key components of SPS: the control center, MPLS-based WAN, and field devices. Two attack scenarios are simulated on the IEEE 9-bus test system. Simulation results and voltage stability analysis reveal that a successful DoS attack, by preventing the SPS from executing predefined protective actions, can lead to severe voltage drops and push the system into instability. These findings highlight the urgent need to address cyber-physical vulnerabilities when designing and deploying special protection schemes.
The energy market has undergone significant evolution, driven by modern energy systems. Nonetheless, the underlying infrastructure remains vulnerable to stealthy attack vectors and exploits, particularly False Data Injection Attacks (FDIAs). Profit-oriented FDIAs strategically manipulate Locational Marginal Prices (LMPs) to achieve long-term financial gains. The stealthy nature of such manipulations makes them difficult to distinguish from normal price fluctuations, posing severe threats to market stability. In parallel, the dynamic nature of energy markets shaped by seasonal variations, demand shifts, and generation changes adds complexity to the detection of these attacks. Given the aforementioned constraints, we introduce a novel market-level unsupervised model designed to detect stealthy LMP manipulations. We thus leverage non-parametric change-point detection principles to identify anomalous pricing behaviors without prior knowledge of attack patterns. In order to enhance robustness and higher confidence on our anomaly diagnosis, we propose a drift adaptive mechanism by adapting to evolving data distributions. Our extensive experiments on synthetic and real-world LMP datasets, showcase the effectiveness of our approach in reducing false alarm rates compared to state-of-the-art methods. Hence, we justify through our experimental outcomes the potential of adaptive unsupervised models in safeguarding modern energy markets against sophisticated and financially motivated cyber threat vectors.
The high penetration of renewables in power systems is responsible for the increased localisation of system dynamics. These changes undermine the reliability of traditional Single Machine Infinite Bus (SMIB)-based methods, highlighting the need for higher granularity in modelling these dynamics to prevent unforeseen locational violations. Existing alternatives are mainly analytical, often computationally intensive, and not predictive — only suitable for system monitoring and post-mortem analyses. We address this challenge by proposing a physics-informed learning approach to predict spatially distributed dynamics with minimal computational overhead. By embedding a multi-machine swing equation within its loss function, the proposed approach tracks the physics governing the dynamic system, facilitating convergence to reliable solutions that align with the physical characteristics of the network. The approach accounts for realistic system voltages, rather than assuming nominal values, by initialising steady-state conditions using the AC Optimal Power Flow (OPF), thereby demonstrating its potential for practical applications. Simulation studies on the IEEE 9-bus and IEEE 39-bus networks demonstrate that the proposed approach is both accurate and efficient in capturing rotor angle dynamics across the network, achieving over 10× computational speed-ups compared to traditional numerical solvers.
Microgrids represent a promising paradigm within smart grid systems, enabling efficient energy management through the integration of renewable energy sources via distributed generators (DGs). These DGs are interconnected through communication networks and managed by distributed, cooperative control systems that synchronize set points via information exchange. A critical challenge in these systems is the control network reconfiguration in response to synchronization attacks targeting communication links. In this paper, we propose a Deep Reinforcement Learning (DRL)-based reconfiguration approach that autonomously adjusts the control network among DGs, considering the microgrid’s stability constraints. The main idea is to enhance synchronization in our microgrid by connecting synchronized nodes to unsynchronized ones. Our objective is to construct a minimum spanning tree (MST) that enables the distributed control system to exchange synchronization information efficiently and in a timely manner, while avoiding compromised links and minimizing disruption to microgrid stability after reconfiguration. Our experimental results demonstrate that the DRL-based strategy outperforms a traditional greedy algorithm by achieving a more optimal reconfiguration of the control network.
In high-latitude regions, the frequent occurrences of snowfall and snow coverage during winter can lead to significant fluctuations in photovoltaic power output, thereby triggering large-magnitude and long-duration ramp events, which pose risks to power grid dispatching. To address the insufficient consideration of snow conditions in existing studies, this paper proposes a two-stage indirect forecasting framework of "explicit snow condition modeling – power forecasting – event warning." First, a "snowfall–snow coverage mask" is constructed: the mask is set to 1 during snowfall periods and then linearly decays after snowfall to generate a continuous snow coverage score, characterizing the degree of residual snow obstruction. Then, the mask, along with meteorological features selected by mRMR feature selection considering Granger causality and historical power data, is input into a conditional attention Transformer model to complete day-ahead power forecasting. Subsequently, three ramp indicators are designed (15-minute slope R1, 4-hour maximum ramp rate R2, and 4-hour net variation D3). Within a recent 30-day sliding window, the quantile thresholds of the three ramp indicators are adaptively estimated. The binarized 0/1 sequences of these indicators are then combined using inverse-weighted soft voting, and continuous "1" segments are merged to form the predicted ramp events, resulting in the final ramp warning outcomes. Finally, predicted and actual events are matched using temporal intersection-over-union, and performance is evaluated. Validation based on real photovoltaic power plant data shows that the proposed method significantly improves power forecasting accuracy and ramp event warning performance under snowfall and snow coverage scenarios, verifying the effectiveness of the proposed approach.
AI-based methods have advanced electricity market trading, yet most time-series and deep reinforcement learning (DRL) models struggle to incorporate unstructured information such as news and policies. This paper introduces a large language model (LLM)-driven trading framework that integrates multi-level memory—short-, mid-, long-term, and reflective—to emulate human decision processes under complex market signals. The framework dynamically evaluates the semantic relevance of news across temporal scales and models delayed effects in relation to market behavior. Experiments on the Australian electricity market show that our approach outperforms traditional strategies in both interpretability and predictive performance. These findings highlight the critical role of textual data in improving market foresight and adaptive trading under information-rich conditions.
Modern industrial automation systems increasingly depend on network infrastructures for time-critical communication, driving the need for solutions that guarantee timely and reliable data delivery. IEEE 802.1 Time-Sensitive Networking (TSN) holds significant promise for converging Information Technology (IT) and Operational Technology (OT) networks, enabling interoperability and supporting the coexistence of mixed-critical traffic crucial for Industry 4.0 and IIoT. To achieve deterministic communication, TSN employs various traffic shapers such as the Time-Aware Shaper (TAS), Asynchronous Traffic Shaper (ATS), and Credit-Based Shaper (CBS). However, the effective deployment of TSN in industrial automation faces several challenges. These include the non-trivial mapping of diverse industrial traffic types to specific shapers, the complexity of optimizing shaper configurations. We present a model for effective traffic engineering within TSN enabled OT Network. Our experiments also demonstrate how shaping of certain traffic types get affected in absence of precise time synchronization and propose possible solutions based on experiment results. Based on our experimental results we provide recommendations on how traffic type assignments should be done and which traffic shaping mechanisms should be used for a particular traffic type.
Modern converter-dominated power grids require fast and adaptive frequency control while guaranteeing system stability and safety. However, existing reinforcement learning (RL) approaches for inverter-based control often sacrifice theoretical guarantees for adaptivity, making them unsuitable for safety-critical grid operations. In this paper, we propose a physics-guided control decomposition that decouples the stability and adaptation mechanisms within a provably stable RL framework. The controller comprises two components: (i) a strongly monotonic nominal control which is used as a baseline that guarantees exponential stability via Lyapunov energy decay, and (ii) a sign-controlled residual that enables robust adaptation to unknown and time-varying inertia and damping. We derive explicit design margins that certify local exponential stability under actuator saturation and parameter variations. The resulting method maintains linear computational complexity and is suitable for real-time deployment on grid-forming inverters. Simulation results on the IEEE 39-bus system show that our method achieves faster settling time, reduced frequency overshoot, and improved robustness to parameter drift compared to safe RL baselines such as Lyapunov RL techniques. As a result, this work provides a scalable and theoretically grounded framework for learning-based frequency control in low-inertia power systems.
Deep Neural Networks (DNNs) have achieved remarkable accuracy in various tasks, including their application in Cyber-Physical Systems (CPS) for detecting False Data Injection Attacks (FDIA) during critical operations. However, the unique infrastructure of CPS makes DNNs vulnerable to exploitation by attackers aiming to evade detection. Additionally, the distinct nature of CPS presents challenges for conventional defense mechanisms against FDIA. This paper proposes an innovative defense framework that strengthens DNNs against such attacks by introducing an additional input layer that performs padding in the input samples using pseudo-feature values derived from the input’s statistical distribution. This padding increases the input dimensionality in a randomized and data-aware manner, making adversarial attacks computationally infeasible due to the non-transferable nature of crafted perturbations and the unpredictability of the padded structure. Our method is lightweight, model-agnostic, and requires no modifications to the core architecture, making it highly deployable in real-world CPS settings. We evaluated our framework on critical power grid applications such as state estimation, using the IEEE 14-bus, 30-bus, 118-bus, and 300-bus systems. Experiments under adversarial settings demonstrate that our padding strategy significantly improves model robustness with negligible impact on performance, and effectively mitigates attacks that would otherwise bypass conventional defenses.
The massive integration of wind parks (WPs) in the modern power grid has raised significant security concerns. These challenges are particularly important when WPs with inherent stability issues, such as doubly-fed induction generators (DFIGs), become radially connected to series compensated transmission systems. On this basis, this paper presents a novel real-time data-driven security monitoring system designed to detect false data injection (FDI) and denial of service (DoS) cyberattacks targeting the subsynchronous damping controller (SSDC) in DFIG-based WPs. First, a detailed and realistic electromagnetic transient (EMT) model of the DFIG-based WP is developed, and an SSDC is designed to mitigate the subsynchronous control interaction (SSCI) phenomenon. Then, cyber vulnerabilities in the structure of the WP are analyzed, and several attack vectors based on IEC 61400 are identified. Next, it has been demonstrated that such attacks render the SSDC unable to dampen the oscillations and result in instability. To counter such an issue, a security monitoring system is developed based on a customized recurrent neural network (RNN)-long short-term memory (LSTM) network model to identify FDI and DoS attacks against the SSDC scheme of the WP. Finally, the performance of the designed monitoring tool is investigated under realistic operating scenarios.
In this work, we propose improved task mapping strategies for real-time electric power system simulations on heterogeneous computing clusters, considering both heterogeneous communication links and processing capacities, with a focus on bottleneck objectives. We approach the problem through two complementary models: the bottleneck quadratic semi-assignment problem (BQSAP), which optimizes task configuration for a fixed number of computing nodes while minimizing communication and computation costs; and the variable-size bin packing problem with quadratic communication constraints (Q-VSBPP), which minimizes the required number of computing nodes, valuable for resource provisioning scenarios. We extend the PuLP library to solve approximately both problems, explicitly including communication costs and processing constraints, and formalize the nomenclature and definitions for bottleneck objectives in graph partitioning. This formalization fills a gap in the existing literature and provides a framework for the rigorous analysis and application of task mapping techniques to real-time electric power system simulation. Finally, we provide a quantitative study and benchmark the extended PuLP library with the SCOTCH partitioning library in the context of real-time electromagnetic transient (EMT) simulation task mapping.
Power utility companies are increasingly interfacing with equipment and services they neither own nor directly control. These are typically provided by third-party entities and now form critical components of the power grid’s supply chain. Among them, aggregators collect and aggregate power usage data from prosumers, and report them for billing, and for monitoring usage and forecasting demand. However, many aggregators are small entities with limited security capabilities, introducing new vulnerabilities into the grid. In this work, we expose Hybrid False Data Injection Attacks (FDIA), whereby an adversary compromises an aggregator and subtly manipulates load reports in a topology-aware fashion to trigger blackouts or power losses. We develop a time-slotted model that abstracts low-level power flow dynamics while capturing interactions among key components in the power grid such as prosumers, aggregators, substations, and the control unit. To assess the impact of Hybrid FDIA, we introduce new performance metrics that quantify different dimensions of inflicted damage. Using realistic prosumer distributions in a regional smart grid, we examine how aggregator dominance can be exploited to deteriorate grid stability and demonstrate how safety margins and energy storage create tradeoffs between resiliency and operational cost.
This paper presents a control framework that enables medium-voltage (MV) grid operators to access flexibility from low-voltage (LV) networks through a structured, humanin-the-loop interface. The approach centers on a local LV agent that computes a multi-period Feasible Operating Region (FOR) and executes flexibility dispatches using Model Predictive Control (MPC). Unlike prior work, the framework combines AC-feasibility, real-time operability, and operator transparency. A simulation-based case study based on a realistic German LV topology illustrates the methodology. Flexible assets, including batteries, heat pumps, electric vehicles, and PV systems, are coordinated to follow a manually selected setpoint. The results demonstrate that the framework reliably delivers grid-relevant flexibility without violating local constraints. However, the analysis also reveals rebound effects caused by synchronized asset recovery after dispatch, which may affect MV-level stability. The study concludes that while local suboptimality is acceptable, post-dispatch coordination must be improved. Future extensions include proactive FOR shaping, transition-phase control, and the integration of multi-actor environments to reflect real-world flexibility access and control structures.
This paper investigates the energy efficiency of 5G Open Radio Access Network (RAN) architectures for smart grid applications. A comparative analysis is performed between a conventional x86-based cloud platform and a low-Size, Weight, and Power (SWaP), small-form-factor ARM-based platform. The study evaluates power consumption and energy efficiency across varying traffic loads, with a particular focus on uplink-dominant scenarios typical of smart grid communications. Results show that the low-SWaP ARM platform achieves significantly lower power consumption and higher energy efficiency per bit than the x86 alternative. The paper also discusses the trade-offs between energy efficiency and scalability when adopting SWaP-constrained platforms.
This paper investigates mean-field control problems for large populations of agents modeled by piecewise deterministic Markov processes. Motivated by the growing need for demand-side management in power systems, we develop a decentralized control framework that accounts for key operational grid constraints such as maximum aggregate power and ramping limits. These constraints are critical to ensure the feasibility and reliability of control strategies in real-world power grids. Furthermore, we address the limitation of classical mean-field approaches that assume agent homogeneity by extending the formulation to heterogeneous populations, where agent-specific characteristics (such as charging power or battery capacity) are encoded through fixed parameters. The resulting framework enables the scalable and feasible coordination of various flexible loads. Theoretical developments are illustrated through applications to electric vehicle charging, demonstrating the impact of the proposed constraints and heterogeneity modeling on the system’s aggregate behavior.
This paper proposes a novel security framework designed to protect smart grid systems from zero-day attacks that exploit vulnerabilities in the Manufacturing Message Specification (MMS) protocol, leveraging domain-specific knowledge. Existing machine learning-based IDS frameworks often overlook the contextual significance of domain-specification, which can provide valuable insights into structure or functionality of IED targeted in message. We propose two-layer feature scoring system that integrates semantic embedding and probabilistic modeling to enhance threat detection. The first layer employs a Word2vec embedding framework to encode itemId (domain-specification) text feature into dense vector representations, capturing their semantic relationships. The second layer utilizes a multivariate gaussian model to analyze these embeddings, enabling robust scoring based on statistical deviations. By incorporating this score along with the length of recently accessed messages around each domain-specification message, the IDS improves its ability to identify anomalous behavior in MMS message patterns, such as malware conducting IED structure scans. Experimental results demonstrate that leveraging these features enables one-class anomaly detection models—such as Isolation Forest, Autoencoder, and Gaussian Mixture Model—to achieve 80% to 99% accuracy in identifying anomalous traffic, such as IED scanning and invalid journal read attacks, highlighting the framework’s potential for advancing intrusion detection system. Additionally, we provide rules to detect malicious domain names.
Honeypots are valuable tools for collecting real-world attack data and threat intelligence, and a network of honeypots, also called honeynet, further allows us to observe attackers’ behavior after their penetration into the infrastructure. Yet, designing and implementing high-fidelity honeynets for smart grid systems remains challenging. Besides the scarcity of available implementations, most of the honeypots available for smart grid offer imitation of a single device (e.g., PLC), and the setup of such devices in a desired topology requires a significant amount of manual configuration efforts. To address the challenge, this paper presents the first-of-its-kind framework for automated instantiation of operational smart-grid honeynets based on user-provided configurations, by extending the automated cyber range generation toolchains, called SG-ML. The developed framework not only facilitates the development of high-fidelity smart grid honeynets but also operation of such honeypot, such as re-configuration and restoration. Our framework integrates and orchestrates configuration of multiple open-source tools, such as Honeyd and HoneyPLC, according to user preferences, for deception. Moreover, multiple communication protocols, Modbus, OPC UA, Siemens S7comm, in addition to IEC 61850, are supported for flexibility. We also tackle a challenge on systematic evaluation of smart grid honeynets. In this direction, in addition to conducting qualitative assessment based on the established taxonomy of fingerprinting tactics, we further develop a toolchain on MITRE Caldera platform for evaluating the deception and logging capabilities of smart grid honeypots/honeynets. The honeynet generation framework and evaluation toolchain will be open-sourced for smart grid R&D community.
Power line communication has achieved technological maturity and significant market penetration. Nonetheless, its continued evolution is crucial to support new applications and drive further market penetration. This paper addresses the question "what next in PLC?". It provides a vision of possible, and less conventional, applications, as well as technology directions with high potential but not prone to challenges and risks. The reasoning and ideas presented may foster research endeavors in the fascinating domain of communication over power lines.
