Energy Digest

Network topology reconfiguration can reduce power system operating costs by optimizing generation dispatch and substation switching, but existing methods leave a gap in finding efficient transition routes between target points. A new approach, Optimal Transition Planning (OTP), co-optimizes the switching sequence and dispatch trajectory subject to AC feasibility at every intermediate point, reducing operating cost by up to 18.4%. The method uses a receding-horizon framework to exclude infeasible topologies and achieve efficient transitions on commodity hardware.

A dynamic model is developed for an integrated electrolyzer-electric-driven compressor system to coordinate its operation under transient disturbances. The proposed model uses PID controllers to regulate the system's response to disturbances from either component, improving stability and reliability. The coordinated model demonstrates effectiveness in mitigating transient fluctuations and ensuring operational reliability.

A grid-forming control method regulates positive-sequence active power delivery and suppresses negative-sequence converter voltage to maintain constant voltage magnitude during unbalanced faults. The method uses a disturbance observer for synchronization, integral, and resonant action, estimating positive- and negative-sequence grid voltages. Experimental results show the proposed method can operate effectively during severe balanced and unbalanced faults.

XGBoost outperforms Long-Short Term Memory (LSTM) in forecasting transmitted heat energy in District Heating Systems, achieving better performance in scenarios where conventional machine learning algorithms excel. The study highlights the computational cost and environmental benefits of using XGBoost, which consistently reduces costs and minimizes carbon footprint associated with data analysis tasks. XGBoost's superiority is attributed to its ability to minimize errors in intervals of limited data availability.

A new benchmarking system for cyberattack detection in electric vehicle charging infrastructure has been developed, using a leakage-controlled session-level approach that models legitimate revisions as normal behavior. The proposed Dual-Branch Masked-Autoencoder (Masked-AE) Transition Boost model evaluates whether current requests are normal and whether transitions resemble benign updates, achieving strong robust validation performance. This system detects malicious request manipulations without rejecting legitimate user choices.

The authors tested various methods for detecting multivariate outliers in District Heating System data, including Z-score, Mahalanobis distances, PCA, Isolation Forest, and Hotelling's T-squared test. The proposed approach identified relevant results from PCA, Isolation Forest, and Hotelling's method, which were then combined using an ensemble method to detect outliers. This approach aims to uncover irregular plant operation and reduce gas consumption and CO2 emission.