Real-Time Application of Machine Learning in Distributed Generation System
This paper proposes a comprehensive approach to deal with islanding detection problem of distributed generation with concept drift by means of cluster sampling preprocessing and ensemble classifiers First,it proposes using clustering method to do preprocessing and sampling.Then,it analyzes ensemble classifier to promote the accuracy of islanding detection.During the simulation,power imbalance and the disturbance of local load have been taken into account.Eventually,it has been proven that the two processes mentioned above play an important role in islanding detection accuracy and generalization ability.
NETL
NSTL
万方数据
维普
