Abstract: Aiming at the problems of long processing time and low accuracy of the traditional machine learning method in the multi-level manufacturing system of the automobile, an intelligent prediction model of body size and assembly quality based on XGBoost is proposed to solve the problem of accurate predictive control of the multi-level manufacturing system body assembly. Firstly, the multi-level assembly process of the car body is analyzed, the data samples processed, and the absolute correlation matrix of different feature elements established based on Spearman coefficients. Secondly, through a real-time collection, cleaning and mining analysis of relevant data in the production process, the data analysis process and data processing framework are proposed, an intelligent prediction model of body size and assembly quality based on XGBoost established, and the effective evaluation of model performance conducted to ensure accurate control of body size assembly. Finally, an example shows that the XGBoost quality intelligent prediction model can solve the problem of car body assembly quality control in multi-level manufacturing systems.