Multi-Dimensional Feature Selection and Combination Method of Aerospace Target Based on K-means Clustering and Information Entropy

Aiming at the shortcomings of K-means clustering analysis algorithm, such as unable to effectively suppress noise features in high-dimensional data and difficult to solve irregular shape clustering, a classification selection method based on K-means clustering and information entropy for multi-dimensional target feature information is proposed. In this method, the information content of characteristic samples contained in information entropy is used as the evaluation distance instead of the traditional Euclidean distance. And combined with the time domain, the aggregation degree and diversity degree are defined to objectively reflect and measure the initial value of feature sample sets and the similarity measure among feature sets, so as to get rid of the serious dependence on the initial value problem and the computational redundancy problem. The simulation results show that the calculation process is more intuitive and clearer, and because the calculation process is relatively simple, the calculation amount is small, and the timeliness is good.