Neurosymbolic inference and fuzzy control enable risk-aware reconfiguration of active distribution networks

This research looks at the challenges of actively reconfiguring distribution networks when there is uncertainty from distributed generation, electric vehicles, and changing load demand. The paper proposes a hybrid control architecture that combines graph‑convolutional inference, symbolic topology verification, and an interval type‑II fuzzy controller. The symbolic layer works as a safety filter that blocks any network topology that is not radial or that becomes disconnected. The fuzzy controller, whose parameters are tuned offline using quasi‑opposition arithmetic optimization, changes the reconfiguration priorities in real time. Tests on a modified IEEE 33‑bus system show that: – Reconfiguration alone cuts active power losses by 31.15 %. – Adding distributed generation to the reconfiguration

cuts losses by 53.72 %. – In fault‑recovery cases, the method lowers the amount of energy not supplied by 77.2 % compared with particle‑swarm optimization. – Over seasonal load profiles, the online policy saves 25.87 % of lost energy while using only 16 switch operations and keeping the network radial. The same framework was also tried on a 69‑bus feeder, showing that it can be transferred to larger systems. The authors point out two main limitations: reconstruction errors can still appear, and the approach still needs extra checks of electrical limits before it can be used in real life. This work gives M.Tech students a clear example of how to

model and test neurosymbolic and fuzzy control methods in smart grids.

Message: Hi FE, I need help with “Neurosymbolic inference and fuzzy control enable risk-aware reconfiguration of active distribution networks” in “Electrical Engineering”