Energy Digest

A feedback system can achieve robust stability with uncertainties induced by symmetric graphical regions in the plane. The separation between a specific variant of scaled relative graphs (SRGs) and the region provides necessary and sufficient conditions for matrix robust nonsingularity (MRN). Sufficient conditions for robust stability of multi-input multi-output linear time-invariant systems under frequencywise symmetric uncertainties can be derived using these concepts, including connections to system characteristics such as disc-boundedness.

Operation at steady state is often not optimal when optimizing over an economic cost objective, and periodic operation yields better performance. An economic model predictive control scheme has been developed to achieve asymptotic stability guarantees with terminal conditions for systems undergoing optimal periodic operation. A non-averaged closed-loop performance bound has also been established using Cesàro summation.

DC microgrids use converter-based electrical networks in data centers and industrial distribution systems, requiring the maintenance of a stable DC-bus voltage within predefined limits. A new impedance-based framework is introduced to evaluate and compare converter control algorithms for grid-forming behavior, providing a basis for superior DC-bus voltage regulation performance. Three novel indices are proposed to quantify voltage-forming and current-forming behavior in converters.

A neural network surrogate is embedded as a constraint replacement in optimal power flow (OPF) formulations to address scalability limitations and improve accuracy. The proposed method achieves high voltage accuracy during post-solution AC power flow validation, with maximum deviations of less than 1.0 V, and solves NN-OPF problems to global optimality within a mixed integer linear programming (MILP) solver tolerance. This approach reduces computation time compared to nonlinear OPF models.

A comparative study found that incorporating temporal autocorrelation outperforms static meteorological features for fine-grained residential energy consumption forecasting, with Long Short-Term Memory (LSTM) recurrent networks achieving higher coefficients of determination than Multilayer Perceptron (MLP) models. The LSTM model resulted in R^2 values of 0.883 and 0.865 for two Melbourne households, outperforming the weather-driven MLPs by 93.8% and 45.5 percentage points. The study also highlights an asymmetry introduced by solar generation, where implicit solar forecasting can be achieved from weather-time correlations.

A vectorized Gaussian belief propagation framework is proposed for near real-time fully-distributed phasor measurement unit-based state estimation in electric power systems. The framework achieves fast convergence and high estimation accuracy, often within a single iteration, and supports scalable, distributed, and multivariate state estimation formulations. It reduces factor graph complexity by combining multiple measurements associated with the same set of variables.

Non-minimum-phase zeros in multi-converter power systems impose bandwidth ceilings on feedback control due to strict realness of these values. These zeros can be expressed as singular values of a matrix constructed from grid admittance and steady-state power injections, providing a quantitative measure at the system level. A zero reshaping strategy is proposed to steer the dominant zero away from the origin, suppressing sensitivity peaks and improving stability margin by identifying optimal sites for voltage droop deployment without iterative search.

A new framework called MAC (Mobility-Aware Coordinated EV charging) is introduced to quantify the maximum potential of leveraging electric vehicle demand flexibility to mitigate distribution network overloading risk. The framework expands feasible scheduling by coupling charging decisions over a full mobility horizon, making it computationally scalable via an alternating direction method of multipliers-based decomposition. Using high-resolution mobility data and feeder hosting-capacity data in a 30% EV adoption scenario for the San Francisco Bay Area, MAC dramatically reduces overload-driven upgrade requirements compared to unmanaged charging.

Lithium Iron Phosphate (LFP) Battery Energy Storage Systems (BESSs) are impacted by inaccurate State of Charge (SOC) estimations, which can lead to delivery failures and inability to meet frequency reserves. Researchers have proposed three optimization approaches to account for SOC uncertainty, finding that an uncertainty-aware model outperforms others in maximizing revenue while ensuring reliable frequency reserve provision. The study highlights the importance of treating SOC uncertainty as an endogenous process within operational strategy.