Abstract: To address the problem that condition monitoring data of wind turbines is prone to sparse sampling and irregular missing due to long-term operation under complex and harsh environmental conditions, a remaining useful life prediction method for wind turbines under sparse data conditions is proposed. A life prediction framework integrating degradation modeling and adaptive optimization is developed for non-uniform time sampling scenarios. First, an exponential degradation model incorporating a time-scale shifting mechanism is established to characterize the continuous evolution of equipment degradation under sparse sampling conditions. Then, an adaptive weighted loss function combining the Huber loss and degradation monotonicity constraints is designed to suppress the influence of abnormal disturbances while ensuring the physical rationality and stability of the degradation modeling results. On this basis, an adaptive backtracking step-size strategy driven by the Armijo condition is introduced to achieve efficient and robust parameter optimization under sparse observation conditions. Finally, the proposed method is validated using actual operational data from a wind farm, and the results demonstrate that the proposed method achieves high prediction accuracy in sparse data scenarios.