基于TMS320C6678的粒子群算法并行设计  

The parallel processing design of particle swarm optimization based on TMS320C6678

作　　者：张庆祥[1] 郭绪涛 王晶 

机构地区：[1]哈尔滨工业大学电子工程技术研究所,黑龙江哈尔滨150001

出　　处：《电子技术应用》2018年第2期23-26,共4页Application of Electronic Technique

基　　金：国家自然科学基金(61401120)

摘　　要：针对粒子群算法在实际应用中的实时性需求,对算法的并行性进行分析,并根据TMS320C6678多核处理器的架构特点,设计出局部并行全局串行的并行模型,高效地将应用程序映射到多核处理器中。实验数据表明,该设计充分发挥了TMS320C6678的性能优势,有效提高了系统的实时处理能力。该设计有效地推进了PSO算法在实际中的应用,对其他各种群智能算法有重要的借鉴意义。In view of the real-time requirement of PSO in practical applications, the parallelism of the algorithm is analyzed. According to the architecture characteristics of TMS320C6678 multi-core processor, this paper designs a local parallel and global serial model, and maps the application program to multi-core processor efficiently. The experimental data shows that the design gives full play to the performance advantage of C6678 and improves the real-time processing capability of the system effectively. The design promotes the application of PSO algorithm in practice effectively and is of important significance to other kinds of swarm intelligent algorithm.

关 键 词：粒子群算法 TMS320C6678 并行处理 加速比 

分 类 号：TN915.04[电子电信—通信与信息系统]