Generating Random Vibration Targeting Acceleration Kurtosis and Probability Distribution by Combining Gaussian Vibrations

ABSTRACT Random vibration testing is essential for ensuring the safety of packaged products during distribution; however, conventional test methods that rely on only the power spectral density of acceleration differ from real vibrations measured on truck beds. These discrepancies lead to overpackaging or underpackaging, and they can be quantitatively evaluated using acceleration kurtosis. Thus far, numerous studies have proposed methods to control kurtosis during vibration testing to match real vibrations. Among these approaches, leptokurtotic vibration tests can be conducted using nonstationary methods, wherein Gaussian‐distributed random vibrations are applied sequentially with varying acceleration standard deviations and durations. However, previous studies showed that vibrations synthesized through nonstationary methods often exhibit kurtosis values deviating from those of the vibrations. To address this issue, this study proposes a novel approach for generating vibrations with targeted kurtosis using a nonstationary method. Standard deviations and durations of each Gaussian random vibration are determined based on the definitions of standard deviation and kurtosis. From many possible combinations of standard deviation and duration, the ones that minimize the difference between the probability distribution of synthesized vibration and that of the real vibration are selected. The proposed method enables a more accurate implementation of vibration tests for packaging by achieving targeted kurtosis values.