Big Data in Experimental Structural Dynamics by Synchronous Ensembles of Collocated Acceleration Signals (CAS): The Reciprocity Principle Failure and Advanced Diagnostics for Composite Material Structures

Abstract

Acquisition of engineering Big Data (statistically similar large ensembles of structural dynamics signals) accompanied by proper physics-mechanics interpretation could enhance considerably structural health monitoring of complicated structuralmechanical systems, their complex geometric features rendering quite difficult the physics-based interpretation of experimental free wave-vibration signals. Towards exploring the most natural and technologically feasible way of recording (data acquisition and sensors technologies) and interpreting Big Data (processing of arbitrary high resolution space-time samples of motions), we considered composite structures as prototypes of complex structures, the latter being prone to damage both during manufacturing and service cycles. Specifically, we acquired at several regions synchronous Collocated Acceleration Signals (CAS). Introduced then is the concept of intersections of synchronous CAS ensembles by a novel application of the Proper Orthogonal Decomposition (POD) transform. Regarding composite materials-based beams, we have found that three synchronous CAS ensembles intersect strongly partially since they share the same co-dominant POD vibration modes. In addition, partial weak intersection of three synchronous CAS ensembles is detected, indicating that the Reciprocity Principle (RP) fails gradually for all natural time scales. The RP holds strongly partially, macroscopically, when the intersection is sharp over a small number of POD modes. Clearly the failure of the RP forms a needed basis on which one could build analytic procedures on POD-based reduced dynamics, modal like vibration patterns, aimed at developing advanced diagnostics, system identification and damage detection for composite materials-based structures, among others.