Research in robust estimation, uncertainty quantification, data fusion, and dynamical modeling. Each illustration explores the central idea behind a paper.
Elapsed0:00Remaining1:41Total1:41
01Typed sensor inputs02Attention and continuous fusion03Residual state correction
1 / 3
01Typed sensor encoders
02Sensor-typed attention
03Continuous graph ODE
04Residual state correction
2026NeurIPS 2026 Workshop: AI Foundations for Power Grids: From Models to Deployment at Scale
Distribution grids combine sensors with different measurement types, reporting rates, and missing data. Het-ODE uses sensor-specific encoders and attention to learn a shared continuous-time graph representation, then refines fused reconstructions with a graph-transformer residual estimator. Tested on three IEEE feeders, the approach estimates states at sparsely observed buses while retaining the distinction between measured inputs and inferred states.
How can we estimate a changing system when its sensors measure different quantities, report at different rates, and leave gaps? In this work, each sensor type has its own encoder to handle its available measurement channels. An observation mask records which values are present. The encoders produce shared latent features, a common representation that also retains the identity of each sensor type.
Sensor-typed attention
The next challenge is deciding which observations should inform one another. Sensor-typed attention learns these relationships using the measurement features, sensor identities, and time between observations. A query represents the information being sought; keys and values represent the information other observations can contribute. Combining this evidence across observation times gives each bus a probability distribution for its initial latent state.
Continuous graph ODE
Between observations, the system continues to change. We represent that evolution with a graph ordinary differential equation, which describes how a compact latent state changes over time. Shared interaction functions exchange information between neighboring buses, while sensor-specific readouts determine their latent derivatives. A typed decoder converts the resulting trajectory into a fused reconstruction of electrical quantities for the state estimator.
Residual state correction
The fused reconstruction provides an initial estimate, which the state estimator then refines. A graph transformer learns a correction for each bus using that estimate, the observation mask, and connections between buses. The final state estimate is the fused estimate plus this learned correction. This residual formulation lets the estimator refine information already recovered by data fusion. In the three test feeders, the refinement improves accuracy over fusion alone.
Citation / BibTeX
@inproceedings{algikar2026sensoraware,
title={Sensor-Aware Joint Data Fusion and State Estimation under Sparse Multi-rate Distribution Grid Sensing},
author={Algikar, Pooja and Fenu, Stefano and Jhala, Kumarsinh and Plathottam, Siby},
booktitle={NeurIPS Workshop: AI Foundations for Power Grids: From Models to Deployment at Scale},
year={2026},
eprint={2610.15156},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/2610.15156}
}
Elapsed0:00Remaining1:47Total1:47
01Probe evidence and attacker belief02Planning guides the learned policy03Physical probing and measured feedback
Virtual power plants often lack the feeder model needed by network-based cyber defenses. This method tests small, secret DER perturbations through measured voltage responses, updates a Bayesian belief over attacker profiles, and uses Stackelberg guidance within reinforcement learning. The learned policy balances informative probing with perturbation cost, adapting defense using DER-level observations rather than explicit feeder topology.
A virtual power plant needs to detect interference even when it lacks a detailed feeder model. We test the integrity of its observations by applying small, secret power changes through selected distributed energy resources. These changes should create a recognizable voltage response at observing resources. A statistical detector compares the measured response with the expected signature, accounting for noise, and reports a decision with a measure of confidence.
Bayesian attacker belief
Detection evidence also helps us learn what kind of attacker may be present. A Gaussian process relates candidate attacker profiles to detection outcomes. Combining this likelihood with the previous belief produces an updated probability distribution over profiles. Hamiltonian Monte Carlo draws samples from that distribution. Their mean and covariance summarize the estimated attacker behavior and the uncertainty that the defender should consider when planning its next action.
Stackelberg-guided RL
The defender must choose a useful probe while anticipating how an attacker might respond. We model this as a Bayesian leader-follower game: the defender proposes an action, accounting for possible attacker responses under its current belief. The proposed action guides reinforcement learning through the policy input and a reward term penalizing deviation. Voltage and risk feedback help the policy adapt the final probing action.
DER probing feedback
The chosen action becomes a small adjustment to active and reactive power, subject to the resources' operating limits. Observing resources measure the voltage response, providing fresh evidence for detection and the attacker-belief update. Repeating this cycle links learning to physical feedback. The objective is useful detection information with limited probing effort and voltage disturbance. The strategy uses local voltage measurements and does not require explicit network connectivity.
Citation / BibTeX
@article{algikar2026gametheoretic,
title={Game-Theoretic Reinforcement Learning based Topology-Agnostic Defense for Virtual Power Plants},
author={Algikar, Pooja and Fenu, Stefano and Jhala, Kumarsinh and Plathottam, Siby},
journal={IEEE Transactions on Smart Grid},
year={2026},
note={Early Access},
doi={10.1109/TSG.2026.3741535},
eprint={2610.15183},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2610.15183}
}
Elapsed0:00Remaining1:57Total1:57
01Sensing constraints and method families02The approved research-dependency map03Uncertainty-aware decision support
1 / 3
01Sensing constraints
02Method families
03Coupled research needs
04Decision support
IEEE Access
Real-Time Grid Visibility in Active Distribution Networks and Microgrids: A Review of Data Fusion, State Estimation, and Topology Identification
Grid visibility depends on more than an accurate state estimator: sensing, communication, data fusion, and topology inference interact. This review organizes classical, probabilistic, learning-based, and physics-informed methods around their operational requirements. It examines how uncertainty and residual evidence should pass between these functions, and identifies gaps in field validation, adaptation, and decision support for active networks and microgrids.
Reliable grid operation requires knowing the system's electrical state and how its equipment is connected. Yet measurements are often sparse, noisy, delayed, and collected at different rates. This review examines how those sensing and communication constraints affect grid visibility. It brings data fusion, state estimation, and topology identification into one discussion, so methods can be assessed against the information actually available to an operator.
Method families
Different methods address different parts of this challenge. Classical estimators use electrical models to relate measurements to system states. Robust methods reduce the influence of unreliable observations, while probabilistic methods represent uncertainty explicitly. Learning and graph methods capture relationships in data, and physics-informed methods incorporate electrical constraints. The review compares these choices through their assumptions, data requirements, computational demands, and suitability for real-time operation.
Coupled research needs
Grid visibility depends on several interacting tasks. Data fusion combines heterogeneous observations for state and topology inference. The inferred network configuration determines how a state estimator interprets those observations. Disagreement between predicted and measured values can motivate checking data quality or connectivity. A central research need is to share uncertainty and residual evidence across these tasks, so each inference can benefit from the others.
Decision support
For operational decisions, an estimate must come with evidence of when it can be trusted. The review identifies field validation, adaptation to changing topology, uncertainty-aware fusion, and scalable computation as important research priorities. It also highlights the need for fallback when sensing or inference becomes unreliable. Its contribution is a coordinated view of what these methods must deliver before they can support dependable real-time decisions.
Citation / BibTeX
@article{algikar-grid-visibility-review,
title = {Real-Time Grid Visibility in Active Distribution Networks and Microgrids: A Review of Data Fusion, State Estimation, and Topology Identification},
author = {Pooja Algikar and Kumarsinh Jhala and Stefano Fenu and Siby Plathottam},
journal = {IEEE Access},
note = {Accepted; publication details pending}
}
Elapsed0:00Remaining1:20Total1:20
01Weight unusual input locations02Limit large residual influence03Infer the latent-function posterior
Outliers in both inputs and responses can distort Gaussian process regression. This work combines projection pursuit weights for unusual inputs with a Huber likelihood that limits the influence of large residuals. Laplace approximation or Gibbs sampling then supports posterior inference and prediction. Experiments examine synthetic data, benchmark datasets, and transmission spectroscopy.
Gaussian process regression represents an unknown function probabilistically, using observed input–response pairs. Outliers can distort this learning through unusual responses or unusual input locations. We address input leverage with projection pursuit: examining the data along multiple directions and measuring robust distances from their central pattern. Larger projection distances receive smaller weights, limiting the influence of unusual input configurations before posterior inference.
Huber residual likelihood
A residual is the difference between an observed response and the latent function value. We standardize it using the input weight, noise scale, and a robust estimate of residual spread. The Huber likelihood applies a quadratic penalty to small standardized residuals and a linear penalty to large ones. This preserves squared-error behavior near the center while limiting the influence of extreme responses.
GP posterior inference
Combining this likelihood with a Gaussian process prior gives a posterior over the latent function, without a closed-form solution. We develop two inference routes. The Laplace route uses a smooth pseudo-Huber loss and a Gaussian approximation near the posterior mode. Gibbs sampling represents the tails through Gaussian distributions with random variances, then samples conditional distributions. Both support predictions and uncertainty at new inputs.
Citation / BibTeX
@inproceedings{algikar-gp-huber-2025,
title = {Robust Gaussian Process Regression with Huber Likelihood and Projection Pursuit},
author = {Pooja Algikar and Lamine Mili},
year = {2026},
booktitle = {ECML PKDD 2025 · LNCS 16013, pp. 3–19},
doi = {10.1007/978-3-032-05962-8_1},
url = {https://link.springer.com/chapter/10.1007/978-3-032-05962-8_1}
}
Elapsed0:00Remaining1:23Total1:23
01Robust process models02Bayesian Gaussian-process inference03Quantify operator uncertainty
Reliable data-driven decisions require attention to both corrupted observations and propagated uncertainty. This dissertation connects robust Gaussian process modeling with uncertainty analysis of Koopman-operator approximations. Its chapters develop robust hyperparameter and likelihood approaches, then study measurement-noise propagation into DMD and extended-DMD operators. The organizing theme is quantifying reliability alongside learned predictions and dynamical representations.
Data-driven models must remain useful when observations are contaminated and learned quantities are uncertain. This dissertation addresses those problems through three complementary research threads. One thread develops robust Gaussian process models: probability models for an unknown response function. Projection statistics and residual weighting reduce harmful observation influence while the mean and covariance parameters are fitted iteratively for prediction.
Bayesian GP inference
A second thread takes a Bayesian approach to robust regression. A Gaussian process prior describes plausible latent functions, and a Huber likelihood models observations while limiting extreme residual influence. Projection-based weights also account for unusual inputs. The resulting posterior, the distribution updated by the data, is inferred using either Laplace approximation or Gibbs sampling. It represents the underlying response function and its uncertainty probabilistically.
Operator moment analysis
A third thread studies uncertainty in learned dynamical operators. Dynamic mode decomposition fits a linear evolution map from state snapshots; its extended form uses functions of the states as features. Under Gaussian modeling assumptions, random matrix theory propagates measurement uncertainty into first and second moments of individual operator entries. This complementary thread quantifies reliability in learned dynamics, alongside the dissertation's robust regression methods.
Manuscript dated December 2024; deposited at Virginia Tech in February 2025.
Citation / BibTeX
@phdthesis{algikar-dissertation-2025,
title = {Robust and Data-Driven Uncertainty Quantification Methods as Real-Time Decision Support in Data-Driven Models},
author = {Pooja Algikar},
year = {2025},
school = {Virginia Tech},
url = {https://vtechworks.lib.vt.edu/items/2c689a22-9256-4e8f-a160-5f9036491bcb}
}
Measurement noise affects the linear operator estimated by dynamic mode decomposition. This paper uses random matrix theory to propagate normally distributed measurement uncertainty through the pseudoinverse and into individual operator entries. Their first and second moments provide expectations and confidence bounds, evaluated against Monte Carlo simulations on spring–mass and power-system examples.
Dynamic mode decomposition estimates a linear map between consecutive measurements of a system's state. Measurement noise makes that fitted map uncertain. We organize the records into current and one-step-shifted snapshot matrices and model their entries as normal random variables with specified means and variances. This establishes the measurement uncertainty that must be carried through the operator calculation.
Pseudoinverse moments
The difficult step is the pseudoinverse: the generalized matrix inverse used in the least-squares fit. Its entries depend nonlinearly on the noisy snapshots. We express an entry as a ratio of linear and quadratic expressions, then use moment-generating functions and integration under normality to derive its first and second moments. These supply the information needed for its mean and variance.
Operator-entry uncertainty
The moments of the generalized inverse are combined with the moments of the shifted measurements to estimate a mean and variance for each fitted operator entry. The mean summarizes its central value; the variance measures its spread under the assumed noise model. Comparisons with repeated noisy simulations examine these calculations on spring–mass and power-system examples, connecting the statistical analysis to learned dynamical relationships.
Citation / BibTeX
@inproceedings{algikar-dmd-2024,
title = {Statistical analysis and method to quantify the impact of measurement uncertainty on dynamic mode decomposition},
author = {Pooja Algikar and Pranav Sharma and Marcos Netto and Lamine Mili},
year = {2024},
booktitle = {63rd IEEE Conference on Decision and Control (CDC), pp. 4373–4379},
doi = {10.1109/CDC56724.2024.10886110},
url = {https://doi.org/10.1109/CDC56724.2024.10886110}
}
Elapsed0:00Remaining1:16Total1:16
01Measure candidate dependence02Estimate the copula parameter03Update the source estimates
1 / 3
01Candidate-source CoS
02Copula parameter regression
03KL-guided W update
2024IEEE PES Innovative Smart Grid Technologies Conference (ISGT), pp. 1–5
Power-system signals may mix sources with nonlinear statistical dependencies. This method uses copula statistics to estimate a dependence parameter, then updates a demixing matrix by minimizing a copula-based Kullback–Leibler divergence. The resulting source estimates retain a modeled dependence structure rather than assuming independence. Tests include synthetic mixtures and coupled generator-speed signals.
Power-system observations can be linear mixtures of sources with nonlinear statistical dependence. Blind source separation seeks the underlying signals without knowing the mixing matrix. We allow dependent sources and apply a demixing matrix to form source candidates. Their ranks define an empirical copula, which describes dependence separately from each signal's individual distribution. A copula statistic then quantifies pairwise association among these candidates.
Copula parameter regression
Within a supplied copula family, the unknown dependence parameter controls the form of the source relationship. We calibrate a quadratic regression using simulated samples with known parameter values and their copula statistics. Applying that family-specific regression to the candidate-source statistic estimates the dependence parameter. This supplies a parameterized target copula for the separation procedure.
KL-guided W update
K L divergence measures the discrepancy between the candidate-source copula density and the target density. We minimize this discrepancy with gradient descent on the demixing matrix. Each update changes the candidate sources, so their copula statistics and estimated dependence parameters are recalculated. Iteration stops when the matrix stabilizes, producing source estimates guided by the supplied dependence model.
Citation / BibTeX
@inproceedings{algikar-ccca-2024,
title = {Quantifying Non-linear Dependencies in Blind Source Separation of Power System Signals using Copula Statistics},
author = {Pooja Algikar and Lamine Mili and Kiran Karra and Akash Algikar and Mohsen Ben Hassine},
year = {2024},
booktitle = {IEEE PES Innovative Smart Grid Technologies Conference (ISGT), pp. 1–5},
doi = {10.1109/ISGT59692.2024.10454117},
url = {https://doi.org/10.1109/ISGT59692.2024.10454117}
}
Elapsed0:00Remaining1:15Total1:15
01Build injection–voltage pairs02Limit contamination during fitting03Predict bus voltages
1 / 3
01Injection–voltage training pairs
02Robust alternating GP fit
03Voltage mean + covariance
2024IEEE Transactions on Power Systems 39(1), pp. 693–705
Corrupted power-injection and voltage data can undermine a learned power-flow model. This work estimates process-model hyperparameters with a Schweppe-type robust estimator, using projection statistics and residual weights to bound outlier influence. The trained surrogate predicts voltage phasors and quantifies associated variance. Studies include the IEEE 33-bus system and a real-world unbalanced 240-bus network.
Stochastic power flow asks how uncertain injections from loads and renewable generation affect grid voltages over time. We train a Gaussian process surrogate using power injections paired with voltage magnitudes or phase angles at matching times. The voltage targets come from measurements when available, or from a power-system simulator. These aligned examples let the model learn the injection-to-voltage relationship.
Robust alternating GP fit
Training data can be contaminated in both injections and voltage responses. We combine projection-based leverage weights with weights derived from the Huber loss for large residuals, the differences between observed and fitted voltages. Iteratively reweighted least squares updates the mean coefficients. With those coefficients fixed, likelihood fitting updates the kernel and noise parameters. Alternating these fits gives the robust process model.
Voltage mean + covariance
For new power injections, conditioning the fitted process on the training data gives a voltage prediction with a mean and covariance. The mean is the surrogate's voltage estimate. The covariance summarizes model uncertainty and correlations among predictions at the requested inputs. Applying this process across an injection time series yields a statistical surrogate that describes changing voltage behavior together with model-based uncertainty.
Issue year 2024; first published online in 2023.
Citation / BibTeX
@article{algikar-powerflow-2024,
title = {A Robust Data-Driven Process Modeling Applied to Time-Series Stochastic Power Flow},
author = {Pooja Algikar and Yijun Xu and Somayeh Yarahmadi and Lamine Mili},
year = {2024},
journal = {IEEE Transactions on Power Systems 39(1), pp. 693–705},
doi = {10.1109/TPWRS.2023.3238385},
url = {https://ieeexplore.ieee.org/document/10023983/}
}
Elapsed0:00Remaining1:08Total1:08
01Build time-delay coordinates02Estimate frequency and damping03Select the modal order
1 / 3
01Time-delay coordinates
02Copula HOBI + Hilbert
03Modal-pair order selection
2023IEEE Power & Energy Society General Meeting (PESGM), pp. 1–5
A measured power-system oscillation can contain several coupled modes, even when few channels are available. The proposed method combines copula-based high-order blind identification with the Hilbert transform to estimate modal frequency and damping. Iterating over delay-embedding dimensions helps identify the modal order. Numerical and 11-bus simulations examine its performance under nonlinear coupling.
A recorded oscillation can contain several modes with different frequencies and damping, even when only one measurement channel is available. The method constructs delayed copies of the signal to expose its temporal structure. For each candidate number of modes, it uses twice as many delay coordinates, creating the observation matrix needed for modal estimation.
Copula HOBI + Hilbert
A copula statistic captures both linear and nonlinear dependence among the observations. This dependence information supports whitening and the construction of dependency matrices at several time delays. Joint diagonalization then estimates the underlying source signals. Applying the Hilbert transform to those estimates yields modal frequencies and damping coefficients, connecting statistical separation to the oscillation properties of interest.
Modal-pair order selection
Each oscillatory mode is associated with a pair of conjugate components whose frequency and damping estimates should agree. The method measures their divergence and repeats the analysis across candidate embedding dimensions. Comparing these pair divergences supports the selection of modal estimates and the number of modes, providing frequency and damping estimates together with an estimated modal order.
Citation / BibTeX
@inproceedings{algikar-hobmi-2023,
title = {Identification of Power System Oscillation Modes using Blind Source Separation based on Copula Statistic},
author = {Pooja Algikar and Lamine Mili and Mohsen Ben Hassine and Somayeh Yarahmadi and Almuatazbellah Muataz Boker},
year = {2023},
booktitle = {IEEE Power & Energy Society General Meeting (PESGM), pp. 1–5},
doi = {10.1109/PESGM52003.2023.10252977},
url = {https://doi.org/10.1109/PESGM52003.2023.10252977}
}
Elapsed0:00Remaining1:10Total1:10
01Diagnose input leverage02Fit the mean by reweighting03Alternate kernel and mean fits
Outliers can bias Gaussian process hyperparameters and the uncertainty attached to predictions. This preprint develops a robust observational data-driven process using a Schweppe-type Huber estimator and projection-statistics weights. Iterative estimation limits the influence of unusual responses and harmful leverage points. Examples include thick-tailed synthetic data, coastal wind and water measurements, and stochastic power flow.
Unusual input locations can exert disproportionate influence when fitting a Gaussian process. This method assesses the rows of the mean model's basis matrix using robust projection distances. Those distances inform leverage weights, which are combined with a robust estimate of residual spread. The resulting scaling prepares the fit to consider both input outlyingness and disagreement with the mean model.
IRLS robust mean fit
The mean coefficients are fitted through iteratively reweighted least squares. Residuals from the current mean are standardized using the leverage weights and robust scale, then Huber weighting limits the influence of large residuals. A weighted fit incorporating the current covariance updates the mean coefficients. Because that update changes the residuals, their weights are recalculated until this inner fit settles.
Kernel–mean alternation
The mean fit depends on the covariance, and covariance estimation depends on the fitted mean. An outer iteration therefore alternates between them: hold the mean coefficients fixed while fitting kernel and noise parameters through profile likelihood, then refit the robust mean with the updated covariance. At convergence, the fitted process provides prediction means and covariances describing uncertainty within the model.
Citation / BibTeX
@misc{algikar-hyperparameters-2023,
title = {Robust Hyperparameter Estimation in Gaussian Process Regression Model},
author = {Pooja Algikar and Lamine Mili},
year = {2023},
howpublished = {Research Square},
note = {Preprint · journal publication not verified},
doi = {10.21203/rs.3.rs-2584837/v1},
url = {https://www.researchsquare.com/article/rs-2584837/v1}
}
Elapsed0:00Remaining1:09Total1:09
01Specify disturbance and grid02Include generator reactance03Compute downstream arrival time
1 / 3
01Grid + disturbance inputs
02Frequency-dependent wave model
03Downstream arrival time
20232023 IEEE Power & Energy Society General Meeting (PESGM)
When an electromechanical disturbance travels through a power system, its arrival depends on more than distance. This study includes generator reactances in a wave-propagation model, making propagation speed depend on disturbance frequency. Combined with admittance-based inertia distribution, the model examines how these effects alter estimated arrival times and rotor-angle behavior.
Assessing disturbance arrival requires a description of both the grid and the disturbance. The network contributes generator reactances, line admittances, and the distribution of inertia. The disturbance contributes its sinusoidal frequency and initial or boundary conditions. Together, these quantities define a physical propagation problem, allowing arrival time to reflect the system's dynamics as well as the distance traveled.
Frequency-dependent wave model
The key refinement is to account for the electrical reactance of each generator. In this model, the rotor angle and terminal voltage angle can differ, while their physical coupling is retained in the wave equation. This coupling makes the modeled wave speed depend on disturbance frequency. A fixed network can therefore have different propagation speeds for different sinusoidal disturbances.
Downstream arrival time
For each disturbance frequency, the model supplies a corresponding wave speed. The travel contribution of a line is its length divided by that speed, and these contributions determine arrival along the chosen path. Different frequencies can therefore produce different modeled arrival times on the same route. The arrival estimate therefore depends jointly on disturbance frequency and the physical parameters along the path.
Citation / BibTeX
@inproceedings{algikar-wave-arrival-2023,
title = {Electromechanical Wave Propagation for Disturbance Arrival Time Assessment in Power Systems},
author = {Somayeh Yarahmadi and Pooja Algikar and Lamine Mili},
year = {2023},
booktitle = {2023 IEEE Power & Energy Society General Meeting (PESGM)},
doi = {10.1109/pesgm52003.2023.10252186},
url = {https://doi.org/10.1109/pesgm52003.2023.10252186}
}
Elapsed0:00Remaining1:15Total1:15
01Distinguish three outlier types02Combine leverage and residual weights03Use the trained voltage emulator
1 / 3
01Outlier taxonomy
02SHGM influence weighting
03Voltage emulator output
2022IEEE Power & Energy Society General Meeting (PESGM), pp. 01–05
Stochastic power-flow emulators trained on measurements are vulnerable to outliers. This paper uses projection-statistics weights and a Schweppe-type generalized maximum-likelihood estimator to limit the influence of vertical outliers and bad leverage points. The learned emulator produces voltage magnitude and angle predictions. A renewable-integrated IEEE 33-bus study tests the method using deliberately contaminated training data.
Robust training must distinguish unusual inputs from harmful observations. A vertical outlier has a large residual at a typical input location. Good and bad leverage points both have unusual inputs, but good leverage agrees with the fitted relationship while bad leverage has a large residual. This distinction matters because unusual inputs can still provide valuable information about the response.
SHGM influence weighting
The emulator combines projection statistics with a Schweppe-type Huber estimator. Projection statistics assess input outlyingness, while weighting based on scaled residuals limits harmful influence without discarding useful leverage. Iteratively reweighted least squares updates the mean coefficients; changing residuals require updated weights. The covariance parameters are also estimated, producing the trained process used to approximate the power-flow response.
Voltage emulator output
The trained emulator maps active and reactive power injections to predictions of voltage magnitude or phase angle at a target bus. Its mean gives the predicted response, while covariance describes uncertainty within the fitted model. Experiments use measured renewable injections in a thirty-three-bus system, with voltage magnitudes and phase angles from simulation, to evaluate the effect of contaminated training data.
Citation / BibTeX
@inproceedings{algikar-emulator-2022,
title = {A Measurement-Based Robust Non-Gaussian Process Emulator Applied to Data-Driven Stochastic Power Flow},
author = {Pooja Algikar and Yijun Xu and Lamine Mili},
year = {2022},
booktitle = {IEEE Power & Energy Society General Meeting (PESGM), pp. 01–05},
doi = {10.1109/PESGM48719.2022.9916996},
url = {https://ieeexplore.ieee.org/document/9916996}
}
Generator inertia is apportioned using connected-line admittances before computing propagation paths and arrival times; detailed sequence pending the full paper.
Full text needed to verify the method animation.
20222022 IEEE PES Generation, Transmission and Distribution Conference and Exposition – Latin America (IEEE PES GTD Latin America)
Electromechanical disturbances reach different parts of a power network at different times. This paper accounts for transmission-line admittances when distributing generator inertia in a continuum model. Simulations on a 39-bus system examine how that distribution changes propagation paths, wave velocities, and disturbance arrival estimates, providing information for the design of oscillation-damping controls.
Electromechanical disturbances travel through a power network, so different locations respond at different times. This study examines how representing generator inertia across transmission lines affects that propagation. It uses line admittances to distribute inertia in a continuum wave model. Simulations on a thirty-nine-bus system investigate propagation paths, wave velocities, and arrival times, with the aim of informing oscillation-damping control design. This overview draws on the published abstract; full text remains unavailable.
Citation / BibTeX
@inproceedings{algikar-wave-2022,
title = {Electromechanical Wave Propagation Analysis in Power Systems},
author = {Somayeh Yarahmadi and Pooja Algikar and Lamine Mili},
year = {2022},
booktitle = {2022 IEEE PES Generation, Transmission and Distribution Conference and Exposition – Latin America (IEEE PES GTD Latin America)},
doi = {10.1109/ieeepesgtdlatinameri53482.2022.10037778},
url = {https://doi.org/10.1109/ieeepesgtdlatinameri53482.2022.10037778}
}
Elapsed0:00Remaining1:18Total1:18
01Organize the reviewed literature02Compare two solution families03Map coordination requirements
Flexible distributed resources can relieve grid congestion, but utilities and aggregators hold different information and control capabilities. This survey organizes market-based mechanisms and technical controls across planning, dispatch planning, and real-time operation. It compares distribution-operator architectures and coordination needs, clarifying how flexibility procurement and local network constraints must be considered together.
Congestion management involves decisions at different time scales. Long-term planning considers future network and resource needs; operational planning prepares for anticipated conditions; real-time operation responds to current constraints. The survey organizes the literature by these horizons to clarify when decisions are made and what information and flexibility are available. This framework supports a consistent comparison of the reviewed approaches.
Compare congestion-management tools
The survey compares economic coordination with direct technical action. Market mechanisms use price signals, contracts, and auctions for flexibility to coordinate resource adjustments. Technical options include network reconfiguration and direct resource control. These approaches differ in the information they need and the participants authorized to act. Comparing those requirements helps identify suitable tools for planning, scheduling, or real-time operation.
Coordinate constraints and flexibility
Coordination matters because network constraints and resource flexibility are distributed across participants. The distribution operator can share local limits and available flexibility with the transmission side, which returns prices and dispatch schedules. Aggregators and microgrids exchange requests and feasible offers with the distribution operator. The survey compares coordination designs in terms of who exchanges information and who controls the resources.
Citation / BibTeX
@article{algikar-congestion-survey-2026,
title = {Distribution System Congestion Management: A Survey of Reliable Integration for Aggregated Resources and Microgrids},
author = {Kumarsinh Jhala and Kiran Kumar Challa and Pooja Algikar and Prudhviraj Dhanapala},
year = {2026},
journal = {IEEE Access},
doi = {10.1109/access.2026.3709210},
url = {https://doi.org/10.1109/access.2026.3709210}
}