Parallel Damage Identification Of Frame Structure Based On Cloud Computing

In recent years,with the development of Structural Health Monitoring(SHM)and technology of structural damage detection,a number of scholars paid more attention to damage detection methods based on particle swarm optimization(PSO)or relevant algorithm,owing to their superior computation performances.With the accumulation of massive structural monitoring data,it is difficult for traditional structural monitoring systems to real time assess on structural states and furthermore efficiently provide alarming.Due to the fact that the iterative process of standard PSO is complex and cost a long time,the ability of traditional PSO is strictly limited in real time processing the massive structural monitoring data.In order to solve the above-mentioned issues,on the basis of cloud computing technology and parallel distributed computing concepts,the parallel structural damage identification approach based on cloud computing is studied.The primary studies and achievements were as follows.(1)The parallel processing was investigated on the Multi-Particle Swarm Coevolution Optimization(MPSCO),and thus a Parallel Improved Multi-Particle Swarm Coevolution Optimization(PIMPSCO)algorithm was proposed.The proposed algorithm has the abilities of having access to distributed parallel computing on the cloud computing platform,and the expandability.Three kinds of trial functions were used to validate the stability of traditional MPSCO and PIMPSCO,and the results show that the proposed PIMPSCO outperforms the traditional MPSCO in the stability.(2)Based on PIMPSCO,a parallel physical parameter identification algorithm of frame structure based on Matlab cloud computing platform was studied,and a novel parallel physical parameter identification algorithm was proposed that can overcome the MPSCO drawback of not easy dealing with massive data effectively.The results of a 15-story frame numerical experiment test and a 7-story steel frame test in lab show that:1)the parallel algorithm can accurately identify the layers rigidity and damage extent of the frame structure;2)the computational time significantly reduces;3)the computational efficiency increases exponentially.(3)A parallel identification algorithm of frame structure parameters based on Spark cloud computing platform was studied,and a numerical experiment of a 30-story frame and a lab experiment of 7-story steel frame were conducted and validated the algorithm.The results show that the proposed algorithm can keep good recognition accuracy and stability and expansibility,and the computation efficiency increases as the amount of data increases.The comparison found that the identification algorithm based on Spark cloud platform has more advantages than Matlab algorithm in the analysis of massive data.(4)On the basis of unitarian hypothesis test and multiple hypothesis test,a parallel intelligent damage identification method based on cloud platform was studied,and a two-stage intelligent damage identification method of frame structure based on cloud computing was proposed,and a numerical experiment was validated the efficiency and noise tolerance of the presented algorithm.The results indicate that the speed and accuracy have improved significantly compared with the traditional structural damage identification method.In summary,the achievements provide a new solution and strategy to the problems,like data expansion and poor computing ability of large-scale structural in SHM system.Furthermore,they have theory and practical significance to some extent.