Implementation of Frequency-Based Classification of Damages in Composites Using Real-Time FPGA-Based Hardware Framework

Abstract

Structural Health Monitoring (SHM) systems become an integral part of most technical systems in recent years. An integration of SHM in technical systems is closely related to: i) providing the guaranteed service lifetime of a system, ii) scheduled/planned maintenance actions, and iii) optimized system operation. For these purposes, different system variables can be monitored and utilized for an estimation of aging level of the system. Monitored system variables are therefore correlated to stochastically occurring damage, indirectly also to Remaining Useful Lifetime (RUL). Among challenges related to SHM, high attention is given to the reduction of a large amount of measured data and its real-time signal processing. In this contribution, classification of damages in composite materials using measurements of Acoustic Emission (AE) is proposed. Here, Discrete Wavelet Transform (DWT) is applied to AE signal to identify different damages in composites. As AE-signal is found in high frequency bandwidth, the amount of data captured in a short time period is enormous. Consequently, the calculation of DWT of such signal requires processing time quite far from real time and delays the entire classification procedure. Due to this, real-time implementation of DWT is proposed to cope with huge amount of captured data in this case and to reduce the time required for signal processing. Using FPGA-based system, real-time implementation of DWT is shown. Obtained results are compared with the results of offline DWT calculation to prove the efficiency and accuracy of real-time implementation.