Research On Structure Feature Calculation Methods Based On Compressed Sensing In Networks

Structural health monitoring(SHM)systems sense the dynamic response characteristics of the engineering structure,including natural frequency and vibration mode shapes,and can detect structural damage in time,which become the guarantee for the safe operation of large civil engineering.However,in order to extract structural features,the data volume of each sensor may reach thousands or even tens of thousands bytes.For the wireless sensor network with limited energy consumption,it will undoubtedly pay an unbearable transmission cost.Therefore,there is a practical requirement: how to reduce the data transmission cost in SHM system network greatly under the condition of ensuring the structural dynamic response characteristic accuracy? In order to solve this problem,the compressed sensing theory and its extended matrix completion are applied to the structural feature calculation of the system.Compressed sensing is a new sampling theory that can be compressed at the same time as the signal is sampled.This paper first optimizes the key part of compressed sensing—the measurement matrix,and proposes an optimization algorithm based on matrix decomposition to improve the performance of compressed sensing sampling.Furthermore,the method of applying compressed sensing to the SHM system network is studied,that is,the data is no longer a simple relay transmission method,but a series of data weighted sums are transmitted.At the same time,the method of combining with the routing strategy of the minimum connected dominating set(MCDS)is proposed,so that the data transmission overhead in the SHM system network is very small during the calculation of the structural characteristics.In addition,as an extension of compressed sensing,matrix completion utilizes the low-rank characteristics of the matrix to reduce the amount of sampled data.Based on this idea,this paper also studies the method of applying adaptive sampling rate matrix completion to the system network.At the same time,the structural feature calculation method combined with the shortest path tree(SPT)routing strategy is proposed,which is convenient to adapt to different engineering structures and further improves this paper jobs.Experimental simulations results show that the algorithm based on the theory ofcompressed sensing and matrix completion can further reduce the data transmission overhead of SHM system and effectively extend the system life under the condition that the accuracy of the structure characteristics is equivalent to the existing distributed computing methods in the network.