A Concept Drift Detection Algorithm for Power Data Stream
Jiangbin Yu, Han Liu, Qing Guo, Xia Chen, Xue Meng · 2024
Due to the concept mean shift characteristic in the fusion and updating of power multi-source data, a concept drift detection classifier is needed for classification. The concept drift detection classifier is difficult to maintain high classification performance, and there are problems such as error detection and delay detection, a concept drift detection algorithm based on information entropy is proposed in this paper. Firstly, information entropy is used to detect concept drift in dynamic data stream. Then, the detected concept drift information is summarized and counted in the concept pool. Finally, two kinds of public real data are used for experiments, and the experimental results are analyzed to verify the validity and correctness of the model. Experimental results show that the proposed algorithm can effectively detect concept drift and update the classifier, and has good classification performance.