Voltage control in modern power systems has become increasingly complex due to the high penetration of renewable generation. Numerous solutions have been proposed from both the transmission and distribution sides, involving generators and system operators. However, the contribution of loads has remained limited, mainly to demand shifting and basic demand response strategies. This work introduces a novel approach that leverages digital twins to enhance the active participation of loads in supporting voltage control. Unlike traditional methods, the proposed framework builds digital twins exclusively from measurable data, enabling virtually any converter-interfaced load connected to a grid, regardless of whether the network is fully known or not, to contribute effectively to voltage regulation. The methodology is first demonstrated through a parametric study, which evaluates the impact of different load behaviors and control strategies on network voltage stability. To further validate the approach, hardware-in-the-loop (HIL) experiments are conducted, confirming the feasibility of real-time implementation. Four voltage control use-cases are developed and tested for a controllable thermal load, showing that even individual loads can provide meaningful support to grid voltage regulation. The results highlight the potential of data-driven digital twins to unlock new, scalable, and flexible contributions from loads, reinforcing the stability of future power systems with high renewable penetration.

Choosing the right HVAC system or the right algorithm of implementing demand response could create significant energy and environmental gains while maintaining resident comfort. However, these choices are closely related to the concept of user comfort, which in turn requires a reasonable fit between user preferences and temperature setpoints. While setting the temperature right is a well-researched question, systems in transition from one setpoint to another are currently not thoroughly addressed in research. But how tolerant the residents really are if a system spends a large share of time outside of the comfort setpoint? This study gives some early insights on how the deviation of temperature from the setpoint affect perceived resident comfort. We use two weeks of data for a smart neighborhood located in Atlanta, GA. We find that the system spends 20% - 50% of time deviating from the setpoint by more than 1°. However, we do not find that increasing deviations cause resident complaints or increasing overrides.

A major factor behind the success of machine learning (ML) models in multiple domains is the availability and accessibility of large, labeled, and well-organized datasets for training and benchmarking. In comparison, power grid datasets face three major challenges: (i) real-world data is often restricted by regulatory constraints, privacy reasons, or security concerns, making it difficult to obtain and work with; (ii) synthetic datasets, which are created to address these limitations, often have incomplete information and are released using specialized tools, making them inaccessible to the broader community; and, (iii) input-output datasets are difficult to generate through simulation for non-experts because open-source simulators are not known outside the power system community. This survey addresses these challenges by serving as an entry point to publicly available datasets and simulators for researchers venturing in this area. We review the current landscape of open-source power network data, machine models, consumer demand profiles, renewable generation data, and inverter models. We also examine open-source power system simulators, which are crucial for generating high-quality, high-fidelity power grid datasets. We aim to provide a foundation for overcoming data scarcity and advance towards a structured web of datasets and simulators to support the development of ML for power systems.

In this research, we present a multi-objective optimization model for efficient energy management of greenhouse (GH) microclimate control systems. The study aims to minimize energy costs and peak demand while controlling the GH temperature, humidity, and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CO</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> levels within a predefined range. A mathematical model is developed to quantify the power consumption of GH components, including lighting, heating, cooling, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CO</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> injection, and ventilation systems, considering electricity costs based on time-of-use (ToU) tariffs distinguishing peak and offpeak periods. Multi-objective convex optimization, implemented in MATLAB, optimizes control decisions for these systems over 24 hours in 15-minute intervals. The decision variables, ranging from 0 (off) to 1 (fully on), determine the operational states of each system. The model uses historical weather data from Regina, Canada, on July 1st (as a sample day), applied to a 1000 <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</i><sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> GH in the simulation. To verify the feasibility and test the effectiveness of the proposed control and optimization method, also to provide a sensitivity analysis based on changes in weather elements, including temperature, solar radiation, and wind speed, we perform the proposed strategy over the first day of each month for every 12 months of the year. Tomato cultivation is chosen for GH due to its compatibility with the controlled environment (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CO</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub>, relative humidity (RH), and temperature). The findings reveal a remarkable up to 43.13% energy savings a day with the optimized strategy compared to the base scenario of the GH control system. The results illustrate optimal activation patterns for GH systems throughout the day, offering a practical tool for smart GH energy management.

False Data Injection Attacks (FDIAs) pose a substantial risk to the reliability and stability of Cyber-Physical Power Systems (CPPS). While federated learning (FL) has emerged as a promising approach to detect such attacks without exposing sensitive data, security concerns remain in FL, including untrusted central aggregators and potential malicious client updates. This research integrate a private Ethereum blockchain layer and homomorphic encryption into a secure FL framework for FDIA detection to verify model updates and authenticate participating nodes. We design smart contracts to immutably log model update hashes and enforce client authentication, enhancing traceability and tamper-resistance. A prototype implementation uses Ethereum smart contracts for model update verification and client identity management. We simulate the blockchain-integrated FL on a cyber-physical power system dataset using three detection models – XGBoost, LSTM, and a Transformer – and analyze the blockchain-induced latency and communication overhead under a specific network configuration. Results show that the blockchain layer has negligible impact on detection accuracy (global AUC ∼ 0.94-0.96 across models) while introducing a moderate training time overhead (∼ 13–40% increase in training duration due to block confirmation delays). The proposed research demonstrates a viable approach to blockchain-aided federated learning for critical infrastructure security, combining data privacy, model integrity, and participant trust in a unified framework.

Recent research has shown promise in using machine learning for cyberattack detection in power systems. However, current studies face limitations: (a) dependence on either physical or cyber features, overlooking multi-modal cyber-physical (CP) correlations; (b) unrealistic full observability assumptions; (c) focus on detecting basic attacks instead of advanced threats such as ransomware (RW); and (d) use of deep learning (DL) models built for 2D data, despite the graph-structured nature of power systems. To address these gaps, we develop a CP testbed using OPAL-RT and a cyber range to simulate both physical and cyber layers under full and partial observability. The testbed produces a realistic multi-modal dataset covering normal operations and various cyberattacks, including RW, brute force, false data injection, reverse shell, and backdoor. Using this dataset, we design graph neural network (GNN)-based multi-modal intrusion detection systems (IDSs) that fuse CP features and capture spatio-temporal dependencies. Results show that CP fusion improves detection rates (DRs) by up to 16% compared to single-modal inputs. The proposed GNN-based IDSs outperform benchmarks by up to 26% in DR, remain effective under partial observability, and demonstrate up to 6% improvement in scalability when applied to larger system topologies.

Developing innovative technologies is considered a critical enabler to successfully integrate large amounts of renewable generation and ensure the stable operation of power systems. In this paper, we propose a novel modular static synchronous series compensator (M-SSSC) that can adjust transmission line reactance in real-time in order to change network power flows. For the first time, a detailed demonstration in a real transmission system is presented that shows how such an innovative technology can maximize overall network utilization and manage operational constraints. Usually, only power flow simulations studies are presented that inevitably contain model inaccuracies and design assumptions. By exploiting the concepts of correlation coefficient and linear regression we demonstrate how the M-SSSC can facilitate renewable generation and cross-border power flows. A comparison with other innovative technologies such as phase shifting transformers, unified power flow controllers and dynamic line rating systems is presented. It is showcased that the proposed technology can be a suitable economical solution when distributed impedance control is needed, has a smaller footprint per ohm impedance, and depicts various advantages, such as rapid deployability, scalability and redeployability compared to other technologies. Furthermore, the M-SSSC can offer real-time, granular control and coordination between multiple deployments is possible, to optimize system power flows.

Deep electrification by 2050 is expected to increase distribution systems by three to five times and include innumerable distributed energy resources (DERs). Robust methods for operations will be required. Reconfigurations, well researched for 50+ years, are created given the size and importance of present distribution systems. This paper proposes a network configuration method which is significantly dense, heavily loaded, societally important, and has innumerable loads and DERs. This method reduces sections of feeders with DERs to equivalent reduced Pi-Model representations. It then uses a regression model to correlate loading scenarios of the distribution to reduced Pi-Model parameters feeder sections. A regression model yields reduced Pi-Models of feeder sections, and they are used to construct a complete distribution system representation, with this reduced model used for reconfiguration. The proposed method was tested on modified 33-, 69- and 123-Bus data networks and reduced the number of buses to around 50%. Computing time was reduced by 26.30%, 58.54% and 67.33%, respectively while providing accuracy of 97.35%, 97.30%, and 99.05%, respectively. The computation time was lowered by 46.45% when the methodology was expanded to the North Dakota 880-Bus network. As the method scales for larger distribution systems, it should increasingly perform better.

The increasing integration of renewable energy sources (RESs), particularly distributed PV systems, poses significant challenges to voltage stability in modern distribution systems. Existing studies use reactive power control to address voltage deviations but incur high losses, with no systematic solution achieving both voltage regulation and loss minimization. This paper proposes a novel voltage control strategy based on multi-agent deep reinforcement learning (MADRL), leveraging decentralized agent coordination to maintain voltage levels while minimizing PV inverter and system losses. Also, a new framework is formulated based on a Markov game, wherein each PV inverter operates as an autonomous agent that adjusts its reactive power output via a centralized training process. The agents, defined as PV inverters, employ the multi-agent twin-delayed deep deterministic policy gradient algorithm to collaboratively minimize voltage deviations. Through the use of local observations and shared global information during training, agents learn robust control policies that generalize to varying conditions and enable decentralized execution without ongoing coordination. Performance of the proposed control strategy is validated on a modified IEEE 33-node distribution system under high variability in PV generation and load demand. Results show that the proposed control strategy significantly improves voltage regulation and reduces power losses compared to state-of-the-art MADRL techniques.