摘要
针对粒子群算法在大规模数据场景下效率不高的问题,基于Spark并行技术对粒子群算法进行扩展设计,以提高算法性能;针对传统粒子群算法在收敛性能上的不足,分析了基于惯性权重的多样优化策略,设计并实现了一种开放式可视化评估系统,以验证各策略的有效性。试验结果表明,该扩展设计能有效提高大规模数据场景下粒子群算法性能,并启发大数据技术与数据算法的有机结合。
Aimed at low efficient problem of particle swarm optimization(PSO)algorithm in large-scale data scenarios,extended design of PSO algorithm is carried out based on Spark parallel technology to improve its performance.Aimed at the deficiency of the traditional PSO algorithm on convergence performance,multiple optimization strategies are analyzed based on inertia weight.An open visual evaluation system is designed and implemented to verify the effectiveness of each strategy.Experimental result shows that the extended design can effectively improve the performance of PSO algorithm in large-scale data scenarios,and the organic combination of big data technologies and data algorithms can also be inspired.
作者
崔同科
陈启安
高岩
CUI Tongke;CHEN Qi′an;GAO Yan(School of Information and Communication,National University of Defense Technology,Wuhan 430019,China;School of Computer Science and Engineering,Hunan University of Science and Technology,Xiangtan 411201,Hunan,China;China Academy of Launch Vehicle Technology,Beijing 100076,China)
出处
《指挥信息系统与技术》
2023年第1期62-69,共8页
Command Information System and Technology
关键词
SPARK
粒子群算法
优化策略
可视化评估系统
Spark
particle swarm optimization(PSO)algorithm
optimization strategy
visualiza‑tion evaluation system
