基于板块元法的Benchmark亮点聚类优化算法  

Optimized Benchmark Highlight Clustering Algorithm Based on Planar Element Method

作　　者：陈原 郝保安[1] 万亚民[1] 杨伏洲[1] 吕维[1] 范若楠[1] 

机构地区：[1]中国船舶重工集团公司第705研究所,陕西西安710077 [2]水下信息与控制国家重点实验室,陕西西安710077

出　　处：《水下无人系统学报》2017年第5期432-436,共5页Journal of Unmanned Undersea Systems

摘　　要：针对现有鱼雷自导仿真中潜艇目标亮点模型精细化程度不高的问题,应用板块元法的基本原理,引用k-means聚类算法的主要思想,提出Benchmark潜艇亮点聚类优化算法,为鱼雷自导仿真构建精细化的潜艇亮点模型。文中对Benchmark潜艇3D模型面元进行划分,计算面元回波声势函数,然后利用初步的聚类算法进行面元运算,建立Benchmark亮点模型;最后研究板块元法中的面元划分质量对仿真结果的影响及二次划分方法,得出亮点模型的聚类优化算法。仿真结果表明,文中所提聚类优化算法构建出的Benchmark亮点模型与现阶段常用的亮点模型相比精细化程度更高,在纵轴方向上的起伏较平稳。文中研究可为鱼雷目标尺度识别研究提供参考。Aiming at the problem that the available highlight model of submarine target in torpedo homing simulation is not exquisite enough, the fundamental principle of planar element model and the main idea of k-means clustering algo- rithm are employed to propose optimized Benchmark highlight clustering algorithm. First, a more exquisite highlight model of submarine target was built for torpedo homing simulation. A three-dimensional Benchmark submarine model was divided into planar elements, and the acoustic potential functions of each element were computed. Then, the ele- ments were disposed by using the primary clustering algorithm, and a Benchmark highlight model was built. At last, the influence of division quality of the elements on the result of simulation was analyzed and the secondary division method was discussed to optimize the algorithm. Simulation indicated that the highlight model of Benchmark based on the pro- posed clustering algorithm is more exquisite with longitudinal stability compared with the available method. This re- search may provide the reference for target recognition of a torpedo.

关 键 词：鱼雷自导 板块元法 聚类算法 BENCHMARK 亮点模型 

分 类 号：TJ630.34[兵器科学与技术—武器系统与运用工程]