Narayana Prasad Padhy

The power system operation is continuously being challenged due to high-intensity disasters and cyber-attacks, possibly leading to blackouts. Such systems can be prepared with efficient restoration plans, providing faster recovery for their resilient and reliable operation. Due to their ownership, operability, and data-sharing limitations, these composite systems separately prepare their plans for the transmission system (TS) and the active distribution system (ADS), restricting them from supporting each other. Therefore, this article proposes a distributed and parallel restoration plan following a blackout between TS and ADS using truncated diagonal quadratic approximation (TDQA). The decomposed TS–ADS are solved iteratively by sharing their boundary active and reactive powers to energize the network in multiple time steps, ensuring minimum losses and restoring maximum loads. The proposed multitime step restoration plan is validated on a T6D2 test system and CIGRE high-voltage and five medium-voltage system for various blackout scenarios and compared with different TS–ADS coordination strategies based on various performance metrics. The TDQA-based restoration, with its single loop and parallel modeling, frames a superior multitime step restoration plan and computational performance.

Accurate estimation of state of charge (SOC) and state of health (SOH) is critical for the safe and reliable operation of electric vehicle (EV) batteries. However, centralized approaches face challenges such as data privacy, communication overhead, and poor scalability under nonindependent and identically distributed (non-IID) data. Although some methods attempt joint estimation, most rely on centralized learning, lack computational efficiency for embedded systems, and overlook uncertainty quantification (UQ), limiting real-world applicability. To address these limitations, this article proposes a federated framework for joint SOC and SOH estimation that is privacy-preserving, lightweight, and uncertainty-aware. The core model is a temporal-degradation-aware informer (TDA-Informer), which incorporates an Informer-Lite encoder and a multitask prediction structure. A battery-focused multifeature attention (BFMFA) module emphasizes critical inputs such as voltage, current, and temperature. The stochastic controlled averaging for federated learning (SCAFFOLD) algorithm is employed to mitigate client drift under non-IID conditions, enhancing convergence across distributed EV clients. UQ is achieved via Monte Carlo Dropout to provide confidence intervals for SOC and SOH predictions. The framework is trained and validated on the NASA battery dataset and tested for generalization on the unseen Oxford dataset, with realistic sensor noise added to simulate practical conditions. Extensive experiments, including ablation studies and comparisons with baseline models, confirm its superior accuracy, robustness, and scalability. While physical deployment remains future work, the results confirm the framework’s potential for intelligent, scalable battery management systems (BMSs).

This paper presents a methodology for the aggregation of DC microgrid (DCMG) components for power system dynamic studies. The proposed methodology of the DCMG components preserves dynamics, dimension, and structure as that of the individual components of the DCMG. The aggregated model would be helpful to analyze the dynamics of the DCMG clusters connected to the grid through bidirectional power electronic interface and its effects during disturbances. With more penetration from solar photovoltaics, batteries, and loads, the aggregate model is computationally less intensive than the detailed model. The paper proposed different scaling parameters for the DCMG components and shows that the detailed and aggregate models have identical structures and dynamics, which have been mathematically proved via necessary and sufficient theorems. Further, extensive simulation studies have been carried out in MATLAB/Simulink to validate and compare the effectiveness in terms of computational time and applicability of the aggregate model with respect to the detailed model. The results have also been verified in real-time using an EMT-based Real-Time Digital Simulator (RTDS) on a standard IEEE 13-bus and IEEE 34-bus network with multiple DCMG clusters.

This paper addresses the computational challenges posed by integrating massive distributed energy resources (DERs) into evolving AC-DC distribution networks. As conventional centralized optimization approaches struggle to handle the increased complexity at the grid edge, we propose an Alternating Direction Method of Multipliers (ADMM)-based consensus approach for a fully distributed active power loss minimization strategy based on second-order conic programming (SOCP). Then, we extended the approach to another distributed approach based on block-coordinated descent to improve the convergence time by reducing the number of macro-iterations compared to ADMM for extensive networks. The original centralized problem is partitioned into multiple sub-problems, each further divided into three distinct areas corresponding to the AC, DC, and voltage source converter. Our approach efficiently determines optimal operating conditions by enabling parallel solutions of these sub-problems and facilitating the sharing of global information among different agents. Simulation results on 33-bus, 132-bus, 123-bus, and an extensive 8500-bus AC-DC radial test systems, under dynamic load and DER output scenarios, demonstrate that the proposed distributed strategy achieves optimal decisions significantly faster than centralized and state-of-the-art ADMM methods.

This article presents a reduced-sensor control framework for modular multilevel converters (MMCs), offering an alternative to conventional schemes that rely heavily on internal sensor measurements. The proposed approach employs a transformer-based Informer model to estimate submodule (SM) capacitor voltages and arm currents using a limited set of external signals, specifically the phase voltage, arm voltage, and dc-link voltage. A two-stage prediction structure is introduced, wherein the arm current is first estimated and subsequently utilized as an input for capacitor voltage prediction to improve accuracy. The estimated states are then processed by a hysteresis tolerance band balancing (HTB${}^{2}$) algorithm, which incorporates a voltage hysteresis band to limit unnecessary SM switching events. This contributes to a reduction in switching frequency and associated losses while supporting effective capacitor voltage regulation. The proposed framework is validated through detailed simulations and experiments under diverse operating conditions. With a hysteresis band set to 0.1% of the nominal capacitor voltage, the HTB2 method achieves an 86.3% reduction in average switching frequency, with only a 4.2% increase in peak voltage imbalance compared with the conventional sorting-based approach.

Distribution Use of Network (DUoN) charges are crucial for recovering the costs of installing, operating, and maintaining the distribution network (DN) assets. This article presents a pricing algorithm to charge consumers and distributed energy resources (DERs) based on their impact on the distribution network. This method utilizes a sensitivity approach to calculate the deviation in system annual chargeable cost based on the incremental change in a) the real and reactive power flow of the facilities, b) the MW-Miles and MVAr-Miles flow of the system, and c) the power factor of the system. This reflects the combined effect of all these parameters and either charges or incentivizes the consumers based on the amount of real and reactive power they are consuming, the distance used to support customers’ real and reactive power injection/withdrawal, the capacity of DN assets being utilized while power is being delivered to consumers and influence on system power factor.The main advantage of the proposed approach is that it is not dependent on the tracing algorithm to detect the power flow paths, hence capable of computing DUoN for decent-sized DNs embedded with DERs, where the contribution of power to the consumers by the DERs or from the grid is difficult to determine or trace. Additionally, due to the appraising and penalizing characteristics of the proposed DUoN charging mechanism, the proposed pricing algorithm may encourage users to act based on local economic signals. This algorithm has been tested on a modified Western network, the IEEE-15 bus network, the IEEE-69 bus network, and the IEEE-123 bus network.