A digital twin-driven perception method of manufacturing service correlation based on frequent itemsets

Manufacturing service composition is a key technology in service-oriented manufacturing systems. Service correlation is a mix-order correlation, which is supposed to be defined as adjacent-order correlation (AO-C) and non-adjacent-order correlation (NAO-C). The existing works mainly focus on AO-C without considering NAO-C, and constantly lead to the failure of composite service execution path (CSEP). In this paper, with the support of digital twin, firstly the non-uniform transitivity of correlation from AO-C to NAO-C is analyzed. Then, the basic model of AO-C, multi-order model of NAO-C, and its relevancy degree formula are proposed based on workflow and modular design. Meanwhile, a perception method based on improved Apriori algorithm is designed and the relevant supporting data is collected by digital twin technology, so as to percept AO-C relevancy degree and calculate the relevancy degree of mix-order correlation in CSEP in the proposed AO-C and NAO-C models. Finally, a case study of magnetic bearing manufacturing service composition is conducted to verify the effectiveness of proposed method.