The integration of Distributed Energy Resources (DERs) such as photovoltaic (PV) systems, battery energy storage systems (BESSs), and electric vehicles (EVs) introduces new challenges to distribution networks despite offering opportunities for decarbonization and grid flexibility. This paper proposes a scalable simulation-based framework that combines deterministic nodal hosting capacity analysis with probabilistic Monte Carlo simulations to evaluate and optimize DER integration in diverse feeder types. The methodology is demonstrated using the IEEE 13-bus and 123-bus test systems under full-year time-series simulations. Deterministic hosting capacity analysis shows that individual nodes can accommodate up to 76% of base load from PV sources, while Monte Carlo analysis reveals a network-wide average hosting capacity of 62%. Uncoordinated DER deployment leads to increased system losses, overvoltages, and thermal overloads. In contrast, coordinated integration achieves up to 38.7% reduction in power losses, 25% peak demand shaving, and voltage improvements from 0.928 p.u. to 0.971 p.u. Additionally, congestion-centric optimization reduces thermal overload indices by up to 65%. This framework aids utilities and policymakers in making informed decisions on DER planning by capturing both spatial and stochastic constraints. Its modular design allows for flexible adaptation across feeder scales and configurations. The results highlight the need for node-specific deployment strategies and probabilistic validation to ensure reliable, efficient DER integration. Future work will incorporate optimization-driven control and hardware-in-the-loop testing to support real-time implementation.

Cyberattacks, especially data injection attacks, are becoming more common as smart grids are increasingly interconnected. In addition, accurate and unbiased high-quality data is required for model training. Most of the data we collect from the real world is sparse, incomplete, inconsistent, and skewed. To address these issues, we have proposed a framework to detect such attacks in this study. Using a stacked autoencoder architecture, synthetic instances of minority class data were generated to enhance the generalizability of the model and address diverse attack scenarios. Various machine learning algorithms were evaluated, and the Random Forest model consistently achieved superior accuracy, ranging from 99.32% to 95.89%. In particular, traditional algorithms such as Logistic Regression exhibited sensitivity to dimensionality reductions, experiencing a 16.96% accuracy drop when the principal components were reduced from all to 10. In contrast, RF demonstrated resilience, with only a 1.67% mean accuracy drop under similar conditions. Both RF and XGBoost emerged as standout models, showcasing high accuracy and robust performance even with dimensionality reduction via principal component analysis (PCA). However, reducing PCA components from 10 to 5 led to performance decreases in all models, with the Support Vector Machine Classifier showing the highest accuracy drop of 14.21%. This study underscores the importance of understanding algorithmic behavior and data features to strengthen cybersecurity in smart grids and emphasizes the critical need for careful feature selection and tuning, particularly for models sensitive to dimensionality reduction.

Distributed control is an effective strategy for synchronizing actions among numerous distributed generators (DGs) within a microgrid. This method adheres to specific communication network topology requirements, following the principles of Graph theory. Unexpected changes to this network structure can disrupt the Laplacian Matrix, resulting in unusual eigenvalues, and ultimately, disrupt the function of the controller, leading to operational disharmony due to a deviation from Graph theory principles. Denial-of-Service (DoS) attacks represent a significant cyber threat capable of altering the Laplacian matrix by blocking communication channels, resulting in structural changes to the communication network. Such unforeseen changes can leave the controller with incomplete neighbor information, which is pivotal for making control decisions. Decisions made without this vital neighbor data can significantly threaten grid stability. This paper demonstrates how the Laplacian matrix adapts its formation under normal conditions, maintaining stability through concurrent updates of the adjacency and diagonal matrix, thus preserving eigenvalues within the stability region. Conversely, it depicts the impact of DoS attacks on a cyber-physical microgrid. These attacks block communication links among DG units, leading to unusual Laplacian matrix formation due to the inability to update the diagonal matrix, causing eigenvalues to fall into an unstable region. The impacts of this unusual formation lead to abnormal control actions, disrupting voltage stability and the allocation of power among Distributed Energy Resources (DERs) has been validated by MATLAB simulation.

In the control system of a microgrid, the integrity of sensor data is crucial, and the potential for attacks on this data poses a significant threat. Such attacks can lead to system instability and, in the worst case, a blackout in the serviced area. This study focuses on detecting these attacks by employing a specialized type of neural network known as a recurrent neural network (RNN). The proposed approach involves using Deep learning techniques for attack detection, utilizing time-series sensor data collected under various conditions, including load changes and normal operation. The RNN functions as a state observer, closely monitoring the output of converters and aligning with actual measurements. Through effective error estimation, the AI-based solution successfully identifies anomalies in local measurements. By integrating this error information, the attack detection system gains the capability to categorize the estimated error, indicating the presence of an attack on the system. The effectiveness of the proposed system is validated using MATLAB, and realtime validation is conducted using OPAL-RT.