Automation of the interpretation of surface response to excitation (SuRE) method by using neural networks

Abstract

In this study surface response to excitation (SuRE) method with a neural network was used for structural health monitoring of an aluminum beam. SuRE method excited and monitored the elastic guided waves on the structure. The frequency-transfer function was captured over a range of high frequencies (20–200 kHz) using a low cost Digital Signal Processor (DSP) system. For magnitude estimation, the amplitude of the received signal for each frequency value is calculated. With the magnitude estimation, the frequency domain spectrum response is calculated. Using an aluminum plate as the experimental surface, load in different points of the beam were applied and the response signals organized in a database. With the experimental data, a neural network is trained with the Levenberg-Marquardt algorithm and then trained with the scaled conjugated gradient algorithm. The study indicated that the SuRE method may be used as a low cost alternative to detect surface changes.