arXiv subjects

Subham Sahoo

Publications and source records attributed to Subham Sahoo.

At least 19 recordsLinked to original sources

arXiv2026

From Grid to Chip: Power Architecture, Stability, and Flexibility of AI Data Centers

The rapid growth of artificial intelligence (AI) computing is transforming data centers into large, dynamic electrical loads. Their deployment is primarily constrained by energy availability and grid-connection capacity, which is further aggravated by the ability of power-delivery architectures, control systems, and computing workloads to operate reliably during fast grid disturbances. This article presents a technological perspective on AI data centers as grid-interactive computing systems. First, it reviews grid-integration bottlenecks, evolving connection policies, grid-code requirements, which has fostered new technological trends via spatio-temporal flexibility available through workload orchestration, cooling systems, on-site resources, and energy storage. Second, it maps the evolution of power-delivery architectures from medium-voltage grid interfaces to chip-level, discussing higher-voltage DC distribution, solid-state transformers, wide-bandgap devices, advanced chip-level power delivery, and liquid cooling. Third, it establishes a three-level stability framework spanning rack-level DC-bus dynamics, facility-level converter interactions, and system-level grid-coupled behavior. The framework connects dominant instability mechanisms, including constant power load effects, impedance interactions, forced oscillations, and operating-mode transitions, with suitable modeling, assessment, and mitigation approaches. Synthesizing these topics, this article highlights grid-to-chip co-design as a central requirement for scalable AI infrastructure, linking computing workloads, power-delivery systems, energy buffers, and grid operation.

cs.ET↗

arXiv2026

Unlocking Embodied Probabilistic Computational Features in Motor Drives

Artificial intelligence (AI)-driven fault diagnosis in motor drives often requires significant computational efforts and time for re-training, in addition to the limited knowledge behind the model and suitability of training and learning mechanisms. This work bridges this gap by proposing a structured mechanism of transforming untapped labeled fault data into AI parameters to leverage probabilistic data-driven learning. This novel AI reservoir modeling framework for power electronics not only eliminates exogenous efforts behind learning data patterns and its optimization, but also provides intuitive guidelines for power electronics engineers behind sizing of AI models. This alignment between data and system physics makes the proposed model transparent and interpretable, bridging practical understanding with data-driven learning. Its computational efficiency is demonstrated using experimental data that structured, physics-aware reservoirs achieve higher diagnostic accuracy and clearer explanations than conventional black-box AI methods.

eess.SY↗

arXiv2026

Quantized Probabilistic AI for Gear Fault Diagnosis in Motor Drives

Deploying large artificial intelligence (AI) models in power electronics often demands high computational resources. Driven by the quantization paradigm, this digest proposes a quantization-aware training (QAT) principle to substantially minimize the number of bits required and simultaneously maximize the accuracy of computations in pre-trained AI models. Considering a pre-trained probabilistic Bayesian Neural Network (BNN) for gear fault diagnosis in motor drives as an example, we quantize its weights and activation functions from floating-point FP32 to low-precision INT8 values, which enhances the computational efficiency by a significant margin of 30-45% (for different model versions) without any compromise in the accuracy and uncertainty estimates. This substantiates a sustainable mechanism of deploying most quantized light-weight AI models into low-cost edge processors for power electronic applications.

eess.SY↗

arXiv2026

Synapse-Inspired Energy Networks: A Neuromorphic Approach to Microgrid Protection without Communication Links

Traditional protection systems for microgrids, which rely on high fault currents and continuous communication, struggle to keep up with the changing dynamics and cybersecurity concerns of decentralized networks. In this study, we introduce a novel biologically inspired protection system based on neuromorphic principles, where each distributed energy resource (DER) functions as a simple neuron. These neurons process local changes in voltage, current signals, and converting them into spike patterns that represent the severity of disturbances. Just as neurons communicate via synapses in biological systems, we exploit transmission cables to coordinate between DERs, enabling them to share information and respond to faults collectively. Fault detection and circuit breaker activation are driven by a First-To-Spike (FTTS) mechanism, similar to the concept of traveling wave protection, but without needing GPS synchronization or communication links. A key innovation is the ability to use the timing of spikes to locally determine the nature of a fault, offering an intelligent, adaptive response to disturbances. Performance shows tripping latency of 10-58 ms, surpassing conventional relays and even traveling-wave methods (60 ms), while maintaining detection accuracy above 98% and spatial selectivity over 97%, enabling real-time, communication-free, scalable protection for plug-and-play microgrids.

eess.SP↗

arXiv2025

Unlocking Innate Computing Abilities in Electric Grids

Electric power grids are engineered energy systems whose forward electrical responses embody high-dimensional and memory-bearing transformations of input signals. In this work, we reveal that these transformations-inherent in electric circuit elements, power flows and network topologies-can be conveniently harnessed for computation without modifying physical grid architectures. By encoding structured input data into the operational setpoints of power electronic converters inside grids, we demonstrate how forward grid dynamics are interpreted into physical representations comprising system variables by showcasing through an affine transformation example implemented on a direct-current (DC) grid, which justifies the capability of grids performing information processing tasks concurrently alongside normal power flows. Our work not only underscores the computation capability intrinsic to grid physics, but also opens a new perspective on how energy networks can function as sustainable computational substrate. This positions them as flexible assets where several computing tasks from data centers can be sustainably outsourced.

eess.SY↗

arXiv2025

Physics-Informed Neural Network for Parameter Identification: a Buck Converter Case Study

System-level condition monitoring methods estimate the electrical parameters of multiple components in a converter to assess their health status. The estimation accuracy and variation can differ significantly across parameters. For instance, inductance estimations are generally more accurate and stable than inductor resistance in a buck converter. However, these performance differences remain to be analyzed with a more systematic approach otherwise the condition monitoring results can be unreliable. Therefore, this paper analyzes the training loss landscape against multiple parameters of a buck converter to provide a systematic explanation of different performances. If the training loss is high and smooth, the estimated circuit parameter typically is accurate and has low variation. Furthermore, a novel physics-informed neural network (PINN) is proposed, offering faster convergence and lower computation requirements compared to an existing PINN method. The proposed method is validated through simulations, where the loss landscape identifies the unreliable parameter estimations, and the PINN can estimate the remaining parameters.

eess.SP↗

arXiv2025

Carbon and Reliability-Aware Computing for Heterogeneous Data Centers

The rapid expansion of data centers (DCs) has intensified energy and carbon footprint, incurring a massive environmental computing cost. While carbon-aware workload migration strategies have been examined, existing approaches often overlook reliability metrics such as server lifetime degradation, and quality-of-service (QoS) that substantially affects both carbon and operational efficiency of DCs. Hence, this paper proposes a comprehensive optimization framework for spatio-temporal workload migration across distributed DCs that jointly minimizes operational and embodied carbon emissions while complying with service-level agreements (SLA). A key contribution is the development of an embodied carbon emission model based on servers' expected lifetime analysis, which explicitly considers server heterogeneity resulting from aging and utilization conditions. These issues are accommodated using new server dispatch strategies, and backup resource allocation model, accounting hardware, software and workload-induced failure. The overall model is formulated as a mixed-integer optimization problem with multiple linearization techniques to ensure computational tractability. Numerical case studies demonstrate that the proposed method reduces total carbon emissions by up to 21%, offering a pragmatic approach to sustainable DC operations.

eess.SY↗

arXiv2024

Uncertainty-Aware Artificial Intelligence for Gear Fault Diagnosis in Motor Drives

This paper introduces a novel approach to quantify the uncertainties in fault diagnosis of motor drives using Bayesian neural networks (BNN). Conventional data-driven approaches used for fault diagnosis often rely on point-estimate neural networks, which merely provide deterministic outputs and fail to capture the uncertainty associated with the inference process. In contrast, BNNs offer a principled framework to model uncertainty by treating network weights as probability distributions rather than fixed values. It offers several advantages: (a) improved robustness to noisy data, (b) enhanced interpretability of model predictions, and (c) the ability to quantify uncertainty in the decision-making processes. To test the robustness of the proposed BNN, it has been tested under a conservative dataset of gear fault data from an experimental prototype of three fault types at first, and is then incrementally trained on new fault classes and datasets to explore its uncertainty quantification features and model interpretability under noisy data and unseen fault scenarios.

eess.SY↗

arXiv2024

Spike Talk in Power Electronic Grids -- Leveraging Post Moore's Computing Laws

Emerging distributed generation demands highly reliable and resilient coordinating control in microgrids. To improve on these aspects, spiking neural network is leveraged, as a grid-edge intelligence tool to establish a talkative infrastructure, Spike Talk, expediting coordination in next-generation microgrids without the need of communication at all. This paper unravels the physics behind Spike Talk from the perspective of its distributed infrastructure, which aims to address the Von Neumann Bottleneck. Relying on inferring information via power flows in tie lines, Spike Talk allows adaptive and flexible control and coordination itself, and features in synaptic plasticity facilitating online and local training functionality. Preliminary case studies are demonstrated with results, while more extensive validations are to be included as future scopes of work.

cs.ET↗

arXiv2024

Data-Driven Graph Switching for Cyber-Resilient Control in Microgrids

Distributed microgrids are conventionally dependent on communication networks to achieve secondary control objectives. This dependence makes them vulnerable to stealth data integrity attacks (DIAs) where adversaries may perform manipulations via infected transmitters and repeaters to jeopardize stability. This paper presents a physics-guided, supervised Artificial Neural Network (ANN)-based framework that identifies communication-level cyberattacks in microgrids by analyzing whether incoming measurements will cause abnormal behavior of the secondary control layer. If abnormalities are detected, an iteration through possible spanning tree graph topologies that can be used to fulfill secondary control objectives is done. Then, a communication network topology that would not create secondary control abnormalities is identified and enforced for maximum stability. By altering the communication graph topology, the framework eliminates the dependence of the secondary control layer on inputs from compromised cyber devices helping it achieve resilience without instability. Several case studies are provided showcasing the robustness of the framework against False Data Injections and repeater-level Man-in-the-Middle attacks. To understand practical feasibility, robustness is also verified against larger microgrid sizes and in the presence of varying noise levels. Our findings indicate that performance can be affected when attempting scalability in the presence of noise. However, the framework operates robustly in low-noise settings.

eess.SY↗

arXiv2024

On Noise Resiliency of Neuromorphic Inferential Communication in Microgrids

Neuromorphic computing leveraging spiking neural network has emerged as a promising solution to tackle the security and reliability challenges with the conventional cyber-physical infrastructure of microgrids. Its event-driven paradigm facilitates promising prospect in resilient and energy-efficient coordination among power electronic converters. However, different from biological neurons that are focused in the literature, microgrids exhibit distinct architectures and features, implying potentially diverse adaptability in its capabilities to dismiss information transfer, which remains largely unrevealed. One of the biggest drawbacks in the information transfer theory is the impact of noise in the signaling accuracy. Hence, this article hereby explores the noise resiliency of neuromorphic inferential communication in microgrids through case studies and underlines potential challenges and solutions as extensions beyond the results, thus offering insights for its implementation in real-world scenarios.

cs.ET↗

arXiv2024

Inferring Ingrained Remote Information in AC Power Flows Using Neuromorphic Modality Regime

In this paper, we infer remote measurements such as remote voltages and currents online with change in AC power flows using spiking neural network (SNN) as grid-edge technology for efficient coordination of power electronic converters. This work unifies power and information as a means of data normalization using a multi-modal regime in the form of spikes using energy-efficient neuromorphic learning and event-driven asynchronous data collection. Firstly, we organize the synchronous real-valued measurements at each edge and translate them into asynchronous spike-based events to collect sparse data for training of SNN at each edge. Instead of relying on error-dependent supervised data-driven learning theory, we exploit the latency-driven unsupervised Hebbian learning rule to obtain modulation pulses for switching of power electronic converters that can now comprehend grid disturbances locally and adapt their operation without requiring explicit infrastructure for global coordination. Not only does this philosophy block exogenous path arrival for cyber attackers by dismissing the cyber layer, it also entails converter adaptation to system reconfiguration and parameter mismatch issues. We conclude this work by validating its energy-efficient and effective online learning performance under various scenarios in different system sizes, including modified IEEE 14-bus system and under experimental conditions.

eess.SY↗

arXiv2024

Spike Talk: Genesis and Neural Coding Scheme Translations

Although digitalization of future power grids offer several coordination incentives, the reliability and security of information and communication technologies (ICT) hinders its overall performance. In this paper, we introduce a novel architecture Spike Talk via a unified representation of power and information as a means of data normalization using spikes for coordinated control of microgrids. This grid-edge technology allows each distributed energy resource (DER) to execute decentralized secondary control philosophy independently by interacting among each other using power flow along the tie-lines. Inspired from the field of computational neuroscience, Spike Talk basically builds on a fine-grained parallelism on the information transfer theory in our brains, particularly when neurons (modeled as DERs) transmit information (inferred from power streams measurable at each DER) through synapses (modeled as tie-lines). Not only does Spike Talk simplify and address the current bottlenecks of the cyber-physical architectural operation by dismissing the ICT layer, it provides intrinsic operational and cost-effective opportunities in terms of infrastructure development, computations and modeling. Hence, this paper provides a pedagogic illustration of the key concepts and design theories. Since we focus on coordinated control of microgrids in this paper, the signaling accuracy and system performance is studied for several neural coding schemes responsible for converting the real-valued local measurements into spikes.

eess.SP↗

arXiv2024

Dynamic Quantum Key Distribution for Microgrids with Distributed Error Correction

Quantum key distribution (QKD) has often been hailed as a reliable technology for secure communication in cyber-physical microgrids. Even though unauthorized key measurements are not possible in QKD, attempts to read them can disturb quantum states leading to mutations in the transmitted value. Further, inaccurate quantum keys can lead to erroneous decryption producing garbage values, destabilizing microgrid operation. QKD can also be vulnerable to node-level manipulations incorporating attack values into measurements before they are encrypted at the communication layer. To address these issues, this paper proposes a secure QKD protocol that can identify errors in keys and/or nodal measurements by observing violations in control dynamics. Additionally, the protocol uses a dynamic adjacency matrix-based formulation strategy enabling the affected nodes to reconstruct a trustworthy signal and replace it with the attacked signal in a multi-hop manner. This enables microgrids to perform nominal operations in the presence of adversaries who try to eavesdrop on the system causing an increase in the quantum bit error rate (QBER). We provide several case studies to showcase the robustness of the proposed strategy against eavesdroppers and node manipulations. The results demonstrate that it can resist unwanted observation and attack vectors that manipulate signals before encryption.

cs.CR↗

arXiv2024

Stability-Oriented Prediction Horizons Design of Generalized Predictive Control for DC/DC Boost Converter

This paper introduces a novel approach in designing prediction horizons on a generalized predictive control for a DC/DC boost converter. This method involves constructing a closed-loop system model and assessing the impact of different prediction horizons on system stability. In contrast to conventional design approaches that often rely on empirical prediction horizon selection or incorporate non-linear observers, the proposed method establishes a rigorous boundary for the prediction horizon to ensure system stability. This approach facilitates the selection of an appropriate prediction horizon while avoiding excessively short horizons that can lead to instability and preventing the adoption of unnecessarily long horizons that would burden the controller with high computational demands. Finally, the accuracy of the design method has been confirmed through experimental testing. Moreover, it has been demonstrated that the prediction horizon determined by this method reduces the computational burden by 10\%-20\% compared to the empirically selected prediction horizon.

eess.SY↗

arXiv2024

A Data-Driven Condition Monitoring Method for Capacitor in Modular Multilevel Converter (MMC)

The modular multilevel converter (MMC) is a topology that consists of a high number of capacitors, and degradation of capacitors can lead to converter malfunction, limiting the overall system lifetime. Condition monitoring methods can be applied to assess the health status of capacitors and realize predictive maintenance to improve reliability. Current research works for condition monitoring of capacitors in an MMC mainly monitor either capacitance or equivalent series resistance (ESR), while these two health indicators can shift at different speeds and lead to different end-of-life times. Hence, monitoring only one of these parameters may lead to unreliable health status evaluation. This paper proposes a data-driven method to estimate capacitance and ESR at the same time, in which particle swarm optimization (PSO) is leveraged to update the obtained estimations. Then, the results of the estimations are used to predict the sub-module voltage, which is based on a capacitor voltage equation. Furthermore, minimizing the mean square error between the predicted and actual measured voltage makes the estimations closer to the actual values. The effectiveness and feasibility of the proposed method are validated through simulations and experiments.

eess.SP↗

arXiv2024

A Gray-Box Stability Analysis Mechanism for Power Electronic Converters

This paper proposes a gray-box stability analysis mechanism based on data-driven dynamic mode decomposition (DMD) for commercial grid-tied power electronics converters with limited information on its control parameters and topology. By fusing the underlying physical constraints of the state equations into data snapshots, the system dynamic state matrix and input matrix are simultaneously approximated to identify the dominant system dynamic modes and eigenvalues using the DMD with control (DMDc) algorithm. While retaining the advantages of eliminating the need for intrinsic controller information, the proposed gray-box method establishes higher accuracy and interpretable outcomes over the conventional DMD method. Finally, under experimental conditions of a low-frequency oscillation scenario in electrified railways featuring a single-phase converter, the proposed gray-box DMDc is verified to identify the dominant eigenvalues more accurately.

eess.SY↗

arXiv2024

Data-Driven Stability Assessment of Power Electronic Converters with Multi-Resolution Dynamic Mode Decomposition

Harmonic instability occurs frequently in the power electronic converter system. This paper leverages multi-resolution dynamic mode decomposition (MR-DMD) as a data-driven diagnostic tool for the system stability of power electronic converters, not requiring complex modeling and detailed control information. By combining dynamic mode decomposition (DMD) with the multi-resolution analysis used in wavelet theory, dynamic modes and eigenvalues can be identified at different decomposition levels and time scales with the MR-DMD algorithm, thereby allowing for handling datasets with transient time behaviors, which is not achievable using conventional DMD. Further, the selection criteria for important parameters in MR-DMD are clearly defined through derivation, elucidating the reason for enabling it to extract eigenvalues within different frequency ranges. Finally, the analysis results are verified using the dataset collected from the experimental platform of a low-frequency oscillation scenario in electrified railways featuring a single-phase converter.

eess.SP↗