Optimal Recloser Placement in Radial Distribution Networks Using LSTM-Based Fault Risk Pattern and Artificial Bee Colony Algorithm

Abstract

Determining the best locations for reclosers in distribution networks requires a method capable of dynamically detecting fault patterns. This study presents a technique that integrates LSTM-based fault pattern predictions with the ABC optimization algorithm to determine optimal recloser placements. The main goal is to model fault risk dynamically and apply it directly to recloser placement decisions. The approach involves two stages: first, training the LSTM model on historical fault time-series data to assess risk for each feeder, indicating the vulnerability of network segments; second, using the ABC algorithm to find optimal recloser locations by minimizing a weighted objective function that includes SAIDI, SAIFI, and the LSTM-predicted fault risk. The method was tested on the IEEE 69-bus radial distribution system with three fault-pattern scenarios. Results demonstrate that combining LSTM predictions with ABC optimization provides an adaptive and effective recloser placement strategy that reduces SAIDI and SAIFI in the test cases.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint