<p>An enhanced nonlinear damage accumulation model based on the Manson-Halford (M-H) model is proposed in this work. To address load interaction effects, the square of the back-to-front stress ratio is incorporated into the action coefficient. By comparing the fatigue forecasting results derived from the improved model with those from other models, its applicability has been confirmed. Higher prediction accuracy is exhibited by the improved model, with over 79% of the relative errors being below 30%, whereas only 60% of the relative errors for the M-H model are below 30%. To assess stability, the standard deviation of the predicted-to-experimental ratio is used to reflect the model’s stability. Under two-stage loading, an average standard deviation of 0.059 is obtained, close to 0, and an average ratio of 0.913, close to 1, is observed, indicating better accuracy. Under two-stage loading scenarios, 98% of prediction errors are effectively confined within a 2 × life factor, whereas under multi-stage loading scenarios, 90% of the predicted errors are smaller than those associated with the M-H model. Collectively, these results validate the model’s enhanced accuracy, stress adaptability, and high robustness, thus establishing its technical superiority in addressing fatigue life prediction challenges involving interacting load sequences.</p>