Characterization and statistical approach of hypervelocity impact response of aluminum plates using PVDF sensors

In order to investigate the hypervelocity impact response of homogeneous 6061-T6 aluminum plates of 3 mm thickness, a 2-stage light gas gun was used to carry out hypervelocity impacts up to 4 km/s and polyvinylidene fluoride(PVDF) piezofilm sensors were used. The acquired impact responses were categorized into non penetration, hard penetration, fragmented penetration, and debris cloud impact. The frequency domain characteristics of the impact responses were observed and analysis of variance (ANOVA) of the 4 impact categories and correlation analysis with impact velocity were performed using 82 feature parameters with respect to the time, frequency, and time-frequency domains. The ANOVA results revealed that the spectral entropy and the conditional spectral and temporal moments of the time-frequency domain were statistically highly significant in differentiating the impact categories. Using the feature parameters of the third order temporal moment, correlation dimension, and zero crossing rate, accuracy of 91.3% was obtained for the categorization of impact responses into the impact categories. These feature parameters also showed moderate to strong and statistically significant correlation with the impact velocity. Inverse prediction of the impact velocity based on the linear fit for feature parameters with high correlation coefficients resulted in a mean error of 0.0374 km/s using the third order temporal moment minimum. These results indicate the potential of the conditional spectral and temporal moments and other signal energy related feature parameters investigated in this study for post processing and even real time assessment of transient events such as hypervelocity impacts.

Automated summary

This study investigated the hypervelocity impact response of 6061-T6 aluminum plates using a 2-stage light gas gun and PVDF piezofilm sensors. The results showed that spectral entropy and conditional spectral and temporal moments were effective in differentiating impact categories, with a categorization accuracy of 91.3%. Furthermore, feature parameters such as the third order temporal moment, correlation dimension, and zero crossing rate exhibited statistically significant correlation with impact velocity.