Hierarchical PSD damage detection methods for smart sensor networks

摘要

Structural health monitoring(SHM)will transform the management and maintenance of civil infrastructure as available technology and methods continue to improve.Realizing the full potential of SHM requires the development of dense arrays of multi-scale sensors running robust and efficient programs.However,the scale of a densely instrumented civil infrastructure implies will generate vast amounts of data.Organizing the sensors in a hierarchical computing environment will store and process more data locally on each smart-sensor.This organization will greatly reduce the amount of data broadcast to a base station or multihopped along the network,thereby reducing its power consumption.However,because hierarchical computing only shares data locally,damage detection algorithms need to effectively perform in the presence of this constraint.This paper examines a damage detection algorithm that analyzes the changes in the structure’s Power Spectral Density(PSD)in a hierarchical,distributed computing environment.The method is modelindependent,requiring only output measurement.The effect of group size,sensor overlap,and frequency range are considered using numeric simulation.The results show that limitations exist for these variables to maintain the functionality of the damage detection algorithm.Nevertheless,when the hierarchical distribution properly addresses these limitations,the proposed algorithm is effective in accurately detecting damage.