To enhance the accuracy of spare parts quantity prediction, a corrected prediction model was proposed along with a Bayesian updating method, which effectively integrated support experience information and actual consumption data. Based on the national military standard exponential prediction model, a correction coefficient was introduced to compensate for modeling errors, with its prior estimated derived from historical support experience. Subsequently, the correction coefficient was dynamically updated using the Bayesian method according to actual spare parts consumption data, thereby achieving iterative refinement of prediction result. Results show that the proposed method can successfully achieve information fusion for spare parts quantity prediction, significantly improving prediction accuracy and reliability, and providing a solid foundation for the scientific allocation of spare parts.