To enable condition monitoring and fault prediction of aero-engines under operational conditions，this study investigates the prediction of metal debris content in aero-engine lubricating oil.This study proposes a comprehensive prediction model based on support vector regression （SVR），wavelet neural network，and BP neural network，which analyzes multi-source test flight data，including aero-engine lubricating oil spectral data，aircraft attitude data，and engine operating state data.Using the typical Fe element in the lubricating oil spectral data as a case study，the results indicate that the prediction accuracy of the established comprehensive prediction model is 9.6%，which is a significant improvement compared to tradition-al machine learning methods （e.g.，support vector machine，wavelet neural network，and RBF neural network）.An information fusion method utilizing multi-source test flight data under aero-engines operational conditions is proposed to evaluate the feasibility and effectiveness of the integrated prediction model for predicting the metal content of lubricating oil，providing essential technical support for fault prediction and health management of aero-engines