Data-driven Impulse Excitation Technique via Statistical Signal Analysis for Determining Materials' Elastic Properties

Abstract

Characterising materials’ mechanical properties is vital in engineering applications to ensure the components' performance. This study emphasised the Impulse Excitation Technique (IET), a non-destructive dynamic testing to determine the materials’ elastic properties. It comprised the standard resonant-frequency-measurement IET and alternative statistical signal data analytic method via Integrated Kurtosis-based Algorithm for Z-notch Filter (I-kaz^TM). The equipment included an impact hammer as an exciter, a piece of piezoelectric film patch as a sensor, a data acquisition device for signals recording, and six materials with the same geometry as samples. They comprised 6582 alloy steel, bronze, T250 cast iron, copper, P20 plastic mould steel and SKD-11 cold work tool steel. The experiment setup was designed based on ASTM E1876 standard for flexural test mode. Raw data were processed through MATLAB software, involving two different signals: impact nd vibration. I-kaz^TM was utilised to decrypt the data statistically and generate a relationship by correlating the vibration and impulse signals with tabulated Young’s Modulus of respective materials. The result indicated that the standard IET achieved a mean error value of 7.6%. In contrast, I-kaz is at 13.1% with approximately 5.5% differences. Although the latter was inferior, improvement could be done as it was driven and relied upon the exploited data. Obtain relevant features through multivariate analysis, adding more materials, and undergoing signal filtration in the pre-processed stage could remove irrelevancy and significantly boost the accuracy. Yet, this alternative method opens a new avenue to advance material characterisation and offering flexibility to researchers with an additional instrument for analysing data other than the standard IET.