Energy Digest

Thermostatically controlled loads, electric vehicles, and other flexible devices can reduce power peaks in distribution networks by coordinating their use under some level of central control and limited information about the devices. A novel optimization-based control scheme has been proposed to flatten total load curves by restricting operating times while preserving customer comfort. The improved scheme achieves greater peak reductions in summer compared to the original formulation, nearly matching half of the potential daily peak reduction achieved by an ideal controller with perfect knowledge.

The paper addresses the robust $\mathcal{H}_2$ synthesis problem for LFT systems subject to structured uncertainty and white-noise disturbances, providing convex synthesis conditions in terms of linear matrix inequalities (LMIs) for both robust and gain-scheduled controller design. The proposed framework preserves the classical interpretation of the $\mathcal{H}_2$ criterion while providing certified robustness guarantees. Numerical examples demonstrate improved disturbance rejection compared to conventional robust $\mathcal{H}_{\infty}$-based designs.

A trustworthiness layer for foundation models in power systems is introduced using stratified conformal prediction, providing adaptive and statistically valid confidence bounds for each output. This method enhances the accuracy of power system modeling by offering richer and more accurate information than traditional methods, with 2x-3x higher precision at faster speeds. The developed trustworthiness layer can also generalize to unseen high-order contingencies.

Quantifying resilience for distribution system customers involves tracking impact of routine smaller outages using conventional reliability indices, but large blackout events' customer minutes interrupted are difficult to quantify. A new resilience metric called System Average Large Event Duration Index SALEDI is proposed to logarithmically transform customer minutes interrupted, providing a more accurate measure. SALEDI has been compared with alternatives and illustrated in practice with standard outage data from five utilities.

A primal-dual-based active fault-tolerant control scheme has been proposed for cyber-physical systems, enabling optimality guarantees and stability in post-fault steady-state operation. This approach casts the problem as a constrained optimization problem and provides a framework that enforces network-level constraints and ensures exponential stability. The scheme's effectiveness is demonstrated through numerical experiments on a DC microgrid.

Thermostatically controlled loads and electric vehicles can be used to reduce power peaks in low-voltage distribution networks by coordinating their operation centrally, using limited information about the devices. A novel optimization-based control scheme with prediction capabilities was proposed and tested, achieving greater peak reductions than an original formulation and nearly half of the potential average daily peak reduction achieved by an ideal controller. The improved scheme showed technical feasibility and design flexibility for real-world applications.

The paper introduces a framework for robust and gain-scheduled ${\cal H}_2$ control techniques for linear fractional transformation systems subject to structured uncertainty and white-noise disturbances. It derives convex synthesis conditions in terms of linear matrix inequalities (LMIs) that enable both robust and gain-scheduled controller design for parameter-dependent systems. The proposed method provides certified robustness guarantees while preserving the classical interpretation of the ${\cal H}_2$ criterion.

A trustworthiness layer for foundation models in power systems introduces adaptive, statistically valid confidence bounds for each output of a foundation model, allowing users to obtain uncertainty estimates and support conservative decisions. This method enhances GridFM with statistically valid prediction intervals, providing richer and more accurate information than traditional methods, while running faster. The new approach demonstrates improved precision and speed compared to DC Power Flow and AC Power Flow for N-k contingency assessment.