To address the issues of complex wiring and low protection accuracy in conventional backup protection devices during the operation and maintenance of distribution automation terminals

it proposes a dynamic relay protection setting method based on historical data prediction and adaptive optimization.This method retains the original main protection while adding a set of temporary protection units

forming an optional dual configuration.The core lies in the utilization of the collected panoramic historical data of the system(voltage

current

power

etc.).The regression model of the advanced extreme gradient boosting(XGBoost) decision tree algorithm optimized by the adaptive moth-flame optimization(AMFO) algorithm accurately predict the future optimal protection setting value and dynamically set the action threshold of the backup protection accordingly.The main and backup protections utilize a singal“or”or“and”logic

effectively preventing false activations from a single protection system

while ensuring dual protection capabilities in the event of a real fault.Simulation results confirm that this algorithm enhances the adaptability and reliability of the protection system

effectively reducing the risks of both false activations and missed activations.