舰船电子工程
艦船電子工程
함선전자공정
SHIP ELECTRONIC ENGINEERING
2011年
8期
56-59
,共4页
数据融合%并行计算%动态聚类%独立划分
數據融閤%併行計算%動態聚類%獨立劃分
수거융합%병행계산%동태취류%독립화분
data fusion%parallel computing%dynamic clustering%independent compartmentalizing
首先介绍了多传感器数据融合并行技术发展现状,分析了影响多传感器数据融合处理并行化的主要难点,在此基础上提出了一种基于数据空间独立划分的并行数据融合理论框架,并基于该框架给出了一种数据空间的近邻聚类非规则区域＂软＂划分方法,将融合计算自适应地分解成多个独立子任务,均衡地分发到各计算节点自治计算。该方法能较好克服并行数据融合中数据耦合、共享同步等技术问题,提升数据和计算空间并行度,具有良好的并行扩展性。试验结果表明了所提出方法的有效性。
首先介紹瞭多傳感器數據融閤併行技術髮展現狀,分析瞭影響多傳感器數據融閤處理併行化的主要難點,在此基礎上提齣瞭一種基于數據空間獨立劃分的併行數據融閤理論框架,併基于該框架給齣瞭一種數據空間的近鄰聚類非規則區域＂軟＂劃分方法,將融閤計算自適應地分解成多箇獨立子任務,均衡地分髮到各計算節點自治計算。該方法能較好剋服併行數據融閤中數據耦閤、共享同步等技術問題,提升數據和計算空間併行度,具有良好的併行擴展性。試驗結果錶明瞭所提齣方法的有效性。
수선개소료다전감기수거융합병행기술발전현상,분석료영향다전감기수거융합처리병행화적주요난점,재차기출상제출료일충기우수거공간독립화분적병행수거융합이론광가,병기우해광가급출료일충수거공간적근린취류비규칙구역＂연＂화분방법,장융합계산자괄응지분해성다개독립자임무,균형지분발도각계산절점자치계산。해방법능교호극복병행수거융합중수거우합、공향동보등기술문제,제승수거화계산공간병행도,구유량호적병행확전성。시험결과표명료소제출방법적유효성。
The paper introduces the developments of multi-sensor data fusion parallel computing technology,and analyzes key factors of affecting the computing parallelizing of multi-sensor data fusion on multi-processor system.Based on that,this paper presents a theoretical framework of parallel data fusion based on data space independent compartmentalizing thinking,and also proposes a data compartmentalizing approach based on near-neighbor clustering algorithm.This algorithm flexibly breaks up the data space into multiple irregular partitions,decompose fusion computing task into multiple independent child-tasks which are distributed for autonomy computation on some computing nodes according to load-balancing algorithm.This approach can overcome the problems of data coupling,memory-share and synchronization during data fusion parallel computation,improve the parallel intensity of data and computation space,and have a good parallel expansibility.The experimental result shows that the presented approach is effective.
