摘要
结构物理参数辨识是结构损伤识别的一个关键问题。针对结构物理参数辨识精度不高、海量数据处理计算效率低下和在单机环境下运算资源不足的问题,提出一种云平台下改进并行化多粒子群协同优化算法(IPMPSCO)的结构物理参数辨识方法。在云计算平台下,引入Apache Spark云计算平台的弹性分布式数据集RDD,对传统多粒子群协同优化算法(MPSCO)的结构物理参数辨识进行分布式并行化改进。为了验证所提方法的准确性和处理海量数据的能力,在8节点的云计算集群上对一个30层框架数值试验和一个7层钢框架试验进行结构物理参数辨识。结果表明,所提方法具有良好的精度和稳定性,在执行效率上优于单机,且具有较好的并行能力。
The physical parameters identification of structures is a key topic in structural damage detection.Considering the problem of low accuracy and computation efficiency and insufficient computation resources in the physical parameters identification of structures,an improved parallel multi-particle swarm cooperative optimization（ IPMPSCO）algorithm was proposed. Based on the apache spark cloud computing platform,the resilient distributed datasets（ RDD）was introduced to parallelly and distributedly improve the traditional multi-particle swarm cooperative optimization（MPSCO） algorithm for the identification of physical parameters. In order to verify the accuracy of the proposed method and the ability to deal with the huge number of data,a 30-story frame numerical simulation and a 7-story steel frame test were conducted to identify the physical parameters on the cloud computing cluster of 8 nodes. The results show that the approach proposed has excellent precision,stability,and fairly parallel ability in the computation efficiency.
作者
骆剑彬
姜绍飞
任晖
赵剑
LUO Jianbin;JIANG Shaofei;REN Hui;ZHAO Jian(College of Civil Engineering,Fuzhou University,Fuzhou 350116,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2018年第14期67-73,共7页
Journal of Vibration and Shock
基金
国家"十二五"科技支撑计划:"城镇重要功能节点和脆弱区灾害承载力评估"(2015BAK14B02-06)
福建省中青年教师教育科研项目(JAT170066)
