Accurate estimation of the State of Charge (SoC) is essential for reliable and energy-efficient operation of electric vehicles (EVs), particularly under dynamic driving conditions and sensor degradation. Traditional SoC estimation approaches often rely on battery-internal measurements and electrochemical models, which limit their transferability across heterogeneous vehicles and restrict applicability in large-scale fleet deployments. This study proposes a battery-agnostic SoC estimation framework that integrates Global False Nearest Neighbor (GFNN)-based temporal embedding with a Convolutional Long Short-Term Memory (Conv-LSTM) network using real-world driving-behavior inputs (velocity, acceleration, and distance). Unlike many existing methods that rely on laboratory drive cycles or empirically selected window lengths, the proposed approach determines the sliding-window length systematically and is trained and validated using heterogeneous real-world driving data from BluSmart Mobility EV fleets collected across campus, market, city, and ring routes. To enhance practical relevance, the model is further evaluated under missing-data and multi-level noise scenarios to emulate realistic sensor degradation. Across four test trips, the proposed method achieves approximately 99.96%, 98.89%, 99.3%, 99.23%, and 98.72% accuracies under normal, missing, and low, medium, and high noise conditions, respectively, demonstrating reliable SoC estimation under real-world operating conditions.

The increased penetration of inverter-based distributed generation has led to the coexistence of diverse control strategies, such as droop control and virtual synchronous generator (VSG) control, for proportional power sharing. Although most existing studies are confined to homogeneous systems with identical controlled schemes, this paper addresses the emerging need to analyze heterogeneous microgrids incorporating droop and VSG-controlled inverters. A key challenge in such systems is the deviation in frequency and voltage under dynamic conditions. To address this, a decentralized washout filter-based control strategy is proposed, eliminating the extensive communication requirement. Further, the paper presents a small-signal stability analysis of the heterogeneous microgrid and investigates the influence of washout filter parameters on system behavior. The effectiveness of the proposed method is demonstrated through simulations and experimental results carried out on a CIGRE low-voltage distribution network with various combinations of droop and VSG-based inverters.

This paper proposes a Physics-Informed Neural Network (PINN)-based distributed control framework to enhance the reliability and cyber-resilience of DC microgrids (DCMGs). Unlike conventional proportional-integral (PI) or purely data-driven controllers, the proposed approach embeds converter governing equations directly into the neural network training process, enforcing volt-second and charge balance constraints within the control law. This physics-informed formulation enables improved dynamic response, robustness to nonlinear load dynamics, and resilience against measurement corruption. A distributed secondary control strategy is integrated to ensure accurate bus voltage regulation and proportional current sharing in multi-bus DCMGs. The framework is validated through detailed MATLAB/Simulink simulations and real-time hardware experiments under source variations, constant power load (CPL) disturbances, continuous/discontinuous conduction mode (CCM/DCM) transitions, and false data injection attacks (FDIAs). Experimental results demonstrate less than 1% steady-state voltage error, 4.2% peak overshoot, and approximately 5 ms settling time, while maintaining regulation under corrupted sensor measurements. The controller is deployed on a low-cost ARM Cortex-M3 microcontroller with 80 KB memory footprint and 1.5 ms execution time, confirming practical feasibility for embedded implementation.

With the growing reliance on DC microgrids (DC MGs) in critical infrastructure, securing them against sophisticated cyberattacks is essential. This study presents a hybrid cyber-defense (HCD) framework that detects and mitigates false data injection (FDI) and replay attacks through a combination of bidirectional LSTM (Bi-LSTM) autoencoders for anomaly detection, cross-correlation analysis for replay attack identification, and a physics-informed neural network (PINN) for adaptive control. A Kalman filter-based estimator maintains control stability when sensor measurements are compromised. The framework operates in a fully decentralized manner at the local controller level, enabling fast, low-latency responses. Validation on a 4-bus MATLAB/Simulink model and a 3-bus DC MG experimental testbed demonstrates over 95% detection accuracy and effective mitigation, preserving voltage stability under attack. The proposed control algorithm is deployed in the AT19SAM3X8E microcontroller, acting as the local controller at each node, occupying 40 $KB$ of memory with an execution time of 0.4 $ms$. This real-time deployment confirms practical applicability for lightweight, standalone operation. These results demonstrate the proposed method's robustness and scalability, advancing intelligent, real-time cyber-resilience for future DC MG.

Hybrid microgrids combine the advantages of both DC microgrids as well as AC microgrids for the renewable sources based energy generation. However, control of a hybrid microgrid becomes quite complex because of the multiple control loops involving both DC-DC converters as well as inverters. This manuscript presents comprehensive modelling of a hybrid microgrid in an autonomous mode. Two inverters in AC microgrid, two converters in DC microgrid connected through an interlinking converter have been considered for the modelling process. Furthermore, the methodology is extendable to any number of converters/inverters. Each sub-part is modeled in state-space form including phase locked loop of interlinking converter. The complete model is linearised about the operating point to obtain the complete system matrix. Sensitivity analysis of all the states have been performed rigorously and state variables corresponding to eigen values vulnerable to instability have been identified. Finally, a stability analysis has been performed by varying different parameters droop coefficients mp, and nq, PLL bandwidth, and coupling inductor of interlinking converter.

Accurate State of Charge (SoC) estimation is critical for mitigating range anxiety in Electric Vehicle (EV) users. Conventional methods relying on battery-specific parameters (e.g., voltage, current, and temperature) often lack robustness under real-world driving due to limited adaptability and sensitivity to data imperfections. This paper proposes a driving behavior-based SoC estimation framework leveraging real-time speed, acceleration, and distance measurements. A Convolutional Long Short-Term Memory (Conv-LSTM) model is developed with an optimal input window size determined via the Global False Nearest Neighbor (GFNN) method to capture temporal driving dynamics. The model is trained and validated on real-world data from BluSmart Mobility EV cabs operating across heterogeneous urban settings in Delhi, India. Robustness is further evaluated under imperfect data conditions, where real-world measurements deviate from ideal sensor readings due to sensor data missingness (Missing Completely at Random (MCAR) and Missing at Random (MAR)) and measurement corruption from Gaussian and structured noise. Results demonstrate high estimation accuracy and resilience in zero-shot settings, confirming reliable generalization to real-world EV operation. Unlike existing approaches that address data imperfections through retraining or condition-specific fine-tuning, the proposed framework evaluates robustness by directly applying the trained model to previously unseen degraded inputs without retraining.

This article proposes a resilient distributed secondary control (DSC) framework for the islanded ac microgrids to maintain the frequency and average microgrid voltage at a nominal value during cyber intrusions. In ac microgrids, any false data injection (FDI) in state variables used in DSC leads to frequency and voltage deviations. This affects the performance of connected loads and the undesired operations in the microgrids; at times, this may extend to disruption or destabilization of the ac microgrid. In this article, a steady-state FDI attack impact analysis on the conventional distributed controller is carried out to determine the deviations in terms of attack magnitude and controller parameters. Further, to mitigate the deviations, the auxiliary variables are incorporated into the conventional cooperative active and reactive power loops. The cooperative errors derived from the auxiliary variables are used to nullify the attack impacts on frequency and voltages. A steady state impact analysis is also presented for the proposed distributed control with an FDI attack to show the convergence of frequency and average voltage of the ac microgrid to its nominal values. An eigenvalue-based analysis is performed to optimally select the proposed controller gains. The proposed method is validated for various case studies in simulations and experimentally on a hardware setup for multiple communication link and local measurement FDI attacks.

Accurate energy consumption (EnC) estimation is vital for enabling real-time energy-aware decisions in electric vehicles (EVs), particularly under dynamic and heterogeneous driving conditions. Existing models often rely on limited features such as gradient or average velocity, overlooking dynamic driving behavior critical to EnC. This work proposes a robust sequence-to-point EnC estimation framework using driving behavior parameters-velocity, acceleration, and distance, that collectively capture instantaneous power demand, propulsion variability, and trip-dependent energy usage. A Global False Nearest Neighbor (GFNN) method is employed to determine the optimal sliding window size, preserving dynamic dependencies without empirical tuning. The embedded sequences are processed by a 1D Convolutional Neural Network (CNN) for efficient and accurate inference. To ensure real-world robustness, a novel Field-Representative Composite Noise (FRCN) model is introduced, simulating sensor drift, dropout, and correlated noise artifacts. Evaluations on diverse real-world datasets from BluSmart Mobility EVs across varied urban routes demonstrate high generalization, transferability, and resilience under both normal (clean) and noisy (FRCN-perturbed) data conditions, confirming suitability for real-time deployment.

The increasing integration of distributed generations (DGs) in modern power systems has led to the proliferation of DC microgrids (DCMGs) due to their operational efficiency and reduced power conversion losses. However, their dependence on communication networks makes them susceptible to cyber threats, particularly false data injection attacks (FDIA), which can compromise system stability and control performance. This paper proposes a game theoretic neural consensus (GTNC) controller to enhance the cybersecurity and operational stability of DCMGs. By leveraging potential game theory, the controller ensures consensus even in the presence of cyberattacks. A dynamic average consensus-based control algorithm is employed to maintain global voltage regulation and proportional power sharing among DGs. Furthermore, an artificial neural network (ANN)-based attack mitigation strategy is integrated to detect and neutralize FDIAs before they propagate through the system. The proposed methodology is validated through MATLAB/Simulink-based simulations and real-time hardware implementation using an OPAL-RT controller. Experimental results demonstrate that the proposed control strategy effectively preserves system stability, restores voltage regulation, and maintains consensus even in the presence of adversarial attacks. This research contributes to the development of cyber-resilient distributed control strategies for future DC microgrids, enhancing their robustness, security, and reliability in smart power networks.

Identifying faulty lines and their accurate location is key to the rapid restoration of distribution systems. Fault identification and its location will become more challenging as power electronics penetration increases and contingencies are seen in larger areas. This paper proposes a single terminal fault location methodology (i.e., no communication involved) that is robust to variations of key parameters (e.g., sampling frequency, fault resistance, etc.) for low voltage DC systems. The proposed method uses local measurements to estimate the current caused by the other remote terminals affected by the contingency. This mimics the strategy followed by double terminal methods that require communications and decouple the accuracy of the methodology from the fault resistance. The algorithm takes consecutive voltage and current samples, including the estimated current of the other terminal. This mathematical approach results in better accuracy than other single-terminal approaches in the literature. The robustness of the proposed strategy against different fault resistances and locations is demonstrated using PSCAD/EMTDC and Real-Time Digital Simulator (RTDS).

Solar photovoltaic (SPV) arrays are subject to various electrical faults, such as line-to-line and line-to-ground. Quantifying the power injection during high impedance array faults and faults under low irradiance is challenging due to the maximum power point tracking control and the associated blocking and bypass diodes. Hence, global sensitivity analysis (GSA) of output power to random SPV array faults is imperative to develop efficient control, operation, and planning strategies for a renewable-integrated power system. Therefore, in this paper, a data-driven approach based on the polynomial chaos Kriging method is proposed for GSA. Four different state-of-the-art topologies of SPV array, namely, series-parallel, total-cross-tied, honey-comb, and bridge-linked, have been analyzed to find out the sensitivity of power to various electrical faults at different fault resistances. A sparse set of orthonormal polynomials approximate the global behavior, whereas analysis of variance kernel-based Kriging analyzes the local variability of the system output. This creates a hybrid metamodel that reflects the global relationship between the output power and random SPV array faults. With the developed metamodel, Sobol indices are calculated analytically to assess the sensitivity of outputs to the input variations, thus determining the severity of faults for array topology. The suggested methodology is less data intensive and is verified on a real-time hardware set-up of a grid-connected SPV system. Comparison results with the existing approaches substantiate the efficacy of the proposed method in terms of accuracy and scalability.

The integration of advanced communication technologies in smart grids enhances efficiency, reliability, and sustainability by enabling bidirectional information flow for energy management and demand response. However, this also increases cybersecurity risks, particularly Man in the Middle (MITM) attacks, which threaten data confidentiality and integrity. This study proposes a Data-driven Network Intrusion Detection System (D-NIDS) to detect MITM attacks in eavesdropping mode, ensuring secure communication. Three attack types-controller attack, router attack, and two-way attack-are analyzed using a diverse dataset with varying attack complexities, durations, and traffic conditions. Multi-class classification models, including K-Nearest Neighbors (KNN), Support Vector Classifiers (SVC), Decision Trees (DT), and Deep Neural Networks (DNN), are trained and evaluated on MATLAB using accuracy, F1-score, Receiver Operating Characteristic (ROC), and Area Under Curve (AUC) metrics. Additionally, the study examines smart grid communication protocols IEC 61850, and Modbus TCP/IP, highlighting vulnerabilities that enable indirect cyberattacks. The findings reinforce the importance of data-driven cybersecurity solutions for modern power grids.

False Data Injection (FDI) attacks pose a major cybersecurity risk to power electronic systems, particularly inverter-based microgrids. These attacks compromise signals like current, voltage, and phase, causing incorrect power flow calculations, inverter instability, and system outages. This study presents a data-driven approach to detect FDI attacks using an attack classifier trained on data generated from a Model Predictive Control (MPC)-based current control scheme for a three-phase, two-level inverter. The dataset includes normal and attack-induced current deviations, enabling robust model training. The study addresses multiple FDI scenarios, including amplitude, offset, and phase manipulations. A multi-class classification approach is adopted, utilizing machine learning classifiers such as Support Vector Classifier (SVC), Decision Tree, Naive Bayes, Ensemble Boosting, and Neural Networks (NN). Classifier performance is evaluated using accuracy, precision, recall, and F1 score. Among the classifiers, the Neural Network outperforms the rest, achieving an overall accuracy of 99.6%, demonstrating superior detection for both covert and stealthy attacks. The use of an MPC-based control framework to generate training data adds novelty to the approach. The proposed methodology is implemented using MATLAB Simulink, highlighting the potential of neural networks in enhancing the cybersecurity of inverter-based microgrids.

In a distributed direct current (DC) microgrid system, the networked communication architecture enhances the accessibility of data but introduces the risk of cyber attacks. Accurate and comprehensive attack detection and mitigation techniques are essential to ensure its reliable operation, effective control, and exposure to hidden dangers and security implications. This article proposes a Two-Fold Deep Neural Network (TFDNN) based control architecture for detecting and mitigating the False Data Injection Attack (FDIA) at the sensor level for a distributed DC microgrid system. TFDNN is a combination of two neural networks. The first neutral network predicts the converter's duty, and the second neural network detects the FDIA by producing the error value. The combination of two network outputs is the desired duty after eliminating the effect of an FDIA. Neural networks are trained with a wide range of data, including attack scenarios and system disturbances, to perform effectively for various FDIA and in-adverse conditions. Later the designed DC microgrid control is deployed into the microcontroller for standalone operation. The proposed scheme is implemented in real-time hardware, and the results are explored.

Typically, a PV-rich low voltage active distribution system (LV-ADS) suffers from voltage unbalance issue during high irradiance period. Lack of proper planning in deploying PVbased single-phase converters (SPCs) in the LV-ADS causes a high phase voltage unbalance factor (PVUF), leading to the tripping of three phase converters (TPCs). The IEEE 1547-2020 code does not mandate SPCs to undertake any preventive measure to reduce the PVUF of LV-ADS. The code recommends TPCs to participate in voltage regulation of the network. However, due to tripping of the TPCs, the voltage regulation degrades. Two new algorithms are proposed in this paper; coordinated leader-follower feedback-based algorithm and cluster selection algorithm which are combined together to resolve the abovementioned issues. The proposed approach is demonstrated in the European LV-ADS test bench to validate its superiority over IEEE 1547-2020 and SOCP based ACOPF methods.

Electric Vehicles (EVs) are gaining popularity due to reduced fuel consumption and emissions, yet challenges like limited battery range and charging infrastructure hinder wider adoption. Driving behaviors, such as speed variations, acceleration, and distance, significantly impact energy consumption. This research focuses on precise energy consumption estimation for routes like Ring, Campus, Urban-city, and Market, using velocity, acceleration, and distance as inputs. A convolutional neural network (CNN) is employed for efficient prediction, leveraging real-world data from Blu Smart Cab Mobility services in Delhi-NCR. The approach demonstrates high accuracy and generalization capabilities.

SOLAR PHOTOVOLTAIC (SPV) SYSTEMS HAVE ESTABLISHED their significance in the renewable energy sector by providing a promising alternative to conventional fossil fuel-based power sources. According to the International Energy Agency, SPV systems are estimated to generate 11% of global energy consumption by the end of 2050 [1]. Despite their potential, these systems are prone to various factors that can impede their performance [1]. These factors include low-efficiency levels that typically range from 15% to 22.9% due to material and manufacturing issues [2], limitations in maximum power extraction techniques [3], and various types of faults that can prevent the panel from delivering the maximum possible power [4].

This paper proposes a cyber-resilient cooperative control to operate autonomous AC microgrids efficiently and reliably. The communication medium is an inherent infrastructure to exhibit cooperative control in microgrids. The measurement states are communicated using the medium which forms consensus in the secondary layer. Thereby, the communicated states come under the realm of the attacker. Voltages of the converters in microgrids are one such state that can be vulnerable to cyber attacks. To overcome this issue, this paper proposes an event-driven resilient control strategy to detect and mitigate such attacks. The mitigation strategy can be exercised up to $N-1$ compromised agents/DERs. Moreover, unlike other prior art techniques, the proposed strategy does not isolate the compromised unit but rather drives it with an event-driven signal. The effectiveness of the proposed methodology is validated with different case studies.

Grid-forming (GFM) converters, designed to provide voltage and frequency stability, face limitations in handling overcurrent conditions and experience synchronization challenges during severe grid disturbances. This paper proposes an integrated control strategy combining current saturation control with supplementary adaptive power reference adjustments for enhancing the dynamic stability of GFM converters. A learning-based adaptive control system is developed to fine-tune the system’s parameters in real-time, offering rapid and robust adjustments to maintain stability during disturbances. In this paper it is demonstrated, through dynamic simulations, that the proposed strategy effectively improves system resilience by reducing oscillations and ensuring smoother synchronization in GFM converter-based grids.

This paper presents a rigorous technical evaluation of an integrated multi-energy microgrid system engineered to tackle the intricate challenge of ensuring dependable renewable energy provision in geographically isolated areas. The investigation is centered on the empirical case of Bilaput village situated in Dist.-Koraput, Odisha, India. The proposed microgrid seamlessly amalgamates solar photovoltaic (SPV) arrays, energy storage systems, and a biomass gasifier unit to establish a dynamic equilibrium between power generation and consumption, thus rendering an adaptive and robust energy supply mechanism for both residential and commercial loads. By orchestrating the interaction between electrical and thermal domains through a gasifier-based combined heat and power generation facility (CHP), the microgrid blueprint provides an important avenue for catering to the energy requisites of remote societies. The optimal orchestration of microgrid operations is realized through an intricate multi-energy dispatch framework and modeled as a complex mixed-integer non-linear programming (MINLP) paradigm. In recognition of the stochastic nature of pertinent variables, such as solar irradiation, ambient temperature, and load dynamics, a dual-tier stochastic programming framework is incorporated to encapsulate real-time unpredictabilities.