Bump Energy for Durability Prediction of Coil Spring Based on Local Regularity Analysis

This paper aims to study the identification of bumps in vibrational signals and develop bump-energy-based durability predictive models for a suspension coil spring. The bump energy of the loading signal is affected by high frequency noises and can lead to inaccurate results. Therefore, it is necessary to eliminate high frequency noise during bump identification. Local regularity analysis was employed to determine the singular points in road signals. Bump signals were then reconstructed from these singular points. Subsequently, bump-energy-based models were developed by correlating with the fatigue lives estimated using the Coffin–Manson, Morrow and Smith–Watson–Topper strain-life models. The results show that the bump signals extracted from the road excitations had a frequency band within 0–50 Hz, indicating that the high frequency noises had been successfully removed during extraction of the bumps. The bump-energy-based models predicted a fatigue life ranging from 3.98x104 to 4x109 cycles within a 95% confidence interval, where the Coffin–Manson-based model showed the highest fatigue life. This is because the Coffin–Manson model did not consider the mean stress effects. When compared with the experimental results, the Coffin–Manson-based model indicates the highest accuracy, given its highest R2 of 0.948. The bump-energy-based models developed in this study contributed an accurate durability prediction of coil springs.