During the battery's production, various surface defects can arise, leading to unknown problems and potentially causing serious consequences. Product defect inference systems, which employ semantic retrieval, not only reduce labor costs in comparison to traditional manual quality inspection but also offer versatility in the field of industrial defect inspection. In this paper, we construct a battery cell surface defect inference system based on fuzzy production rules. We generate an expert knowledge base for battery cell inspection, conduct qualitative analysis on the battery cell surface defect rules using fuzzy logic, and formulate judgment rules that establish correlations between causes and types of defects through fuzzy matching. According to the fuzzy rules, the system efficiently performs fuzzy inference to identify the most probable defect types on the cell surface, thereby reducing product quality inspection time. We evaluate the performance of the expert system, the results demonstrate its reliable performance. It effectively infers the surface defect types of cells and continuously updates the detection standards with excellent accuracy, sensitivity, and specificity.