Research And Application Of Genetic Clustering Algorithm On Device Defect Classification

In this paper, clustering algorithms and genetic algorithms are learned, and their advantages and disadvantages are analyzed. K-medoids algorithm is easy to fall into local optimal solution and sensitive to initial value; while the genetic algorithm has global search ability and implicit parallelism. So, this paper study proposed genetic clustering algorithm and improve cross factor, which can greatly improve the effectiveness of clustering. This paper introduces the idea of the semi-supervised clustering to solve incompatibility between clustering results and a small amount of the prior knowledge. The advantage of semi-supervised clustering is that can improve the accuracy of clustering results by using a small amount of supervised sample information to unlabeled sample data set for clustering. Based on the idea of Semi-supervised clustering, this paper proposes a method that dynamic granularity combine with genetic clustering algorithm. Finally, this method is validated the correctness and validity by using the data sets in the UCI. This algorithm is applied to classicize electrical equipment defect, which lay the foundation for defect prediction of the power equipment and improve efficiency and accuracy equipment of defect prediction.