• Developed a life estimation model for metal fatigue that considers multiple influencing factors. • Based on the mean and amplitude distribution characteristics of the extrapolated load signal, a two-dimensional joint probability distribution is established using a mixed distribution model. • An eight-level, two-dimensional stress spectrum and life spectrum have been developed for the engine support. The impact of random loading on the fatigue life of mechanical components contains numerous uncertainties. This study aims to enhance the precision of life prediction by developing mean and amplitude distribution models based on the road load spectrum and extrapolating the operational load beyond the threshold value in the time domain. Additionally, it considers the influence of high cycle fatigue (HCF) and low cycle fatigue (LCF) loads on the initiation life of metal cracks. To address this, a load correction factor is introduced, and critical stresses for high and low cycle loads are proposed. Furthermore, a two-dimensional programmed load spectrum is prepared, taking into account the mean and amplitude. The accuracy and rationality of the fatigue life model are validated through examples using 45# steel and 316L, while experiments on engine support are conducted to confirm the effectiveness and accuracy of the over-threshold extrapolation life prediction method under random loading. [ABSTRACT FROM AUTHOR]