Analysis of similarity and difference in manufacturing processes based on petri nets

Most traditional similarity and difference analysis methods adopt relatively single analysis criteria; for instance, only considering the dependency relations between activities, which leads to inaccurate analysis results. This problem becomes particularly prominent for manufacturing processes with complex structures that are significantly affected by the relations between operations and components, where accurate evaluation is far more difficult. This paper proposes a new method for similarity and difference analysis of manufacturing processes that comprehensively considers both the relations between operations and components and operation profile. Firstly, the concept of operation profile is defined, and the similarity of the manufacturing processes is analyzed from three perspectives: operation items, operation profile and operation-aware profile. Secondly, the candidate operation profile matrix and operation items are constructed through clustering, separation and deviation calculation to identify the process differences. Finally, comparative experiments are conducted to verify the effectiveness of the proposed method, and a paired t-test (with a significance level of 0.05) is introduced to validate the statistical significance of the performance differences. The experimental results show that the proposed method, which comprehensively considers both the relations between operations and components and operation profile, outperforms traditional methods in precision, recall, F1 -score, and accuracy for similarity measurement. Verified by the paired t-test, the performance improvement is statistically significant, with all p -values less than 0.05. Compared with the advanced OTV method, the proposed method still has a significant advantage in the precision indicator ( p -value less than 0.05), whereas no statistically significant difference is observed in accuracy ( p -value is approximately 0.1039). In terms of difference analysis, the proposed method can identify the deep-seated differences between processes more comprehensively and precisely, providing referable and optional directions for the subsequent optimization and upgrading of manufacturing processes.