Methods for editing random load spectra based on multi-parameter index preservation for accelerated durability testing

Abstract

Herein, a multi-parameter index preservation-based (MPIP) load spectrum editing method is proposed for improving the efficiency of the accelerated durability testing of mechanical components. The editing process was conducted in the frequency domain by adopting a wavelet transform to decompose the load spectrum into different wavelet levels. The high-frequency wavelet levels were squared and summed at each time interval to obtain the wavelet energy distribution. The damage segments with large amplitude cycles were identified and extracted from the load spectrum based on the wavelet energy distribution. A shorter spectrum was obtained without significant changes in the indexes (the damage index, statistical index, and power spectrum density index). The effectiveness of the MPIP-based load spectrum editing method was demonstrated through the editing process of a coil spring load spectrum, along with a fatigue and damage assessment. The results show that the time duration reduction ratio of the load spectrum after editing was considerably lower than that of the load spectrum before editing. Further, the statistical indices (i.e., the mean, root mean square, and kurtosis) and power spectrum density Pearson correlation coefficient of the MPIP-based edited load spectrum were comparable to those of the original load spectrum. The results also demonstrate that the MPIP-based edited load spectrum indices and time-domain-based edited load spectrum have certain differences. However, the proposed method is more effective for accelerated durability tests. The results regarding the coil spring fatigue damage and lifetime assessment further validated this conclusion.