Implementation of Radar Signal Sorting Based on K-means Clustering and Cuckoo Search Methods

Abstract

Abstract In this paper, a fusion clustering radar signal sorting algorithm is proposed to overcome the shortcoming of the traditional K-means clustering sorting algorithm, which is sensitive to the initial clustering center and easy to fall into a local optimal solution. The improved algorithm combines K-means clustering algorithm and cuckoo search, to reduce the dependence on the initial clustering center and avoid falling into local optimum. Further, a new definition method of fitness function and a new adjustment mechanism of discovery probability are proposed, and sine and cosine guidance are introduced in the process of location updating. The number of clusters is proposed to be adjusted adaptively according to the distance between classes, and the initial cluster center is determined by the data field. Implementation results show that the sorting accuracy of the new algorithm is 96.6%, which is 54% higher than that of the traditional K-means clustering. Compared with PSO-Kmeans, the accuracy is improved by 22%, and the convergence time is reduced by 10 seconds on average.