Manufacturing and inspection in model-based engineering, achieving computer-readable and -interpretable tolerance specification representations, and maintaining unambiguous and consistent semantics throughout the life cycle of product design are issues that have attracted the attention of researchers and require urgent actions. An ontology method based on skin tolerance model is proposed in this study to address the abovementioned problems. Tolerance specification semantics described by the GeoSpelling language can be consistent at different stages of product development. In the ontology as a formal shared concept model, the knowledge described by the web ontology language and the semantic web rule language can be read and interpreted automatically via computer. The ontology is based on a rigorous description logic and describes tolerance information with clear and unambiguous semantics. The inference algorithm of this ontology can be used for ontology consistency checking, knowledge reasoning, and semantic query. The proposed method utilizes the advantages of ontology artificial intelligence technology and GeoSpelling language to describe tolerance semantics. Thus, this method is suitable for consistent semantic data exchange in different heterogeneous systems and automatic interpretation of tolerance semantics in smart manufacturing.