A high-quality assembly process information model forms the key foundation for the digital assembly of mechanical products. This paper addresses the challenges inherent in traditional diesel engine assembly process information modeling, including the large number of components, complex structures, vast amounts of process data, and the high precision required for local assembly. It proposes a multi-scale, assembly feature-driven information modeling method tailored to the diesel engine assembly process. This method is designed to meet the actual needs of the entire diesel engine assembly process and involves reconstructing the diesel engine assembly process BOM through a comprehensive analysis of all relevant elements. The process is divided into three scales: assembly units, assembly elements, and assembly components. At each scale, different methods are employed to define assembly process information: skeleton models for units, ontology models for elements, and interactive feature pairs for components. This approach not only enhances the structuring and integration of assembly process information but also reinforces hierarchical relationships within the assembly, laying a solid foundation for the subsequent design of three-dimensional process procedures and visual guidance in the diesel engine assembly workshop.