基于聚类思想的弹道外推算法研究

为了提高炮位侦察校射雷达中炮位侦察定位精度同时提升外推结果一致性,本文引入聚类思想,建立了基于K-均值聚类的弹道外推模型。该模型采用七态无迹卡尔曼滤波算法对量测数据进行多次滤波,然后利用4 阶龙格-库塔积分方法对火炮位置进行外推,最后对多次外推结果进行K-均值聚类处理,采用综合多因子方法计算簇品质,选取最优簇对应的聚类中心作为最终的火炮位置进行输出。实验结果表明,该弹道外推算法显著提升了外推结果的一致性及定位精度。

Research on Extrapolation Algorithm Based on Clustering Theory

To improve the positioning accuracy and conformance of emplacement reconnaissance radar, this paper introduces clustering theory and establishes a ballistic extrapolation model based on K-means clustering. The model uses seven-state UKF to filter the measurement data for many times, and then uses fourth-order Runge-Kutta method for artillery position extrapolation. Finally, the results of multiple extrapolation K-means clusterings are performed K-means clustering,and the cluster quality is obtained by using the comprehensive multi-factor method. The clustering center corresponding to the optimal cluster is selected as the artillery position output. Experimental results show that the trajectory extrapolation method improves the consistency of extrapolation results and the location accuracy significantly.