Typical process route discovery method based on multi-dimensional fusion similarity calculation and Markov clustering

In the era of knowledge economy and intelligent manufacturing, a large number of process cases and data accumulated by manufacturing enterprises have become valuable assets of enterprises. Effective mining and high-quality reuse of these cases is an important way to improve the efficiency of product development and carry out design and manufacturing process innovation. In reality, the first task is to determine the typical process with high representativeness and reusability in the process case database so as to realize efficient mining and reuse of the manufacturing process cases accumulated by enterprises. In order to scientifically determine the typical process, this paper proposes a typical process route discovery method based on multi-dimensional fusion similarity calculation and Markov clustering. From the perspective of the longest subsequence with the same sequence in the process route and the calling frequency of the process method, this method proposes the order measurement method and the disorder measurement method of the similarity between the process routes. Furthermore, by means of the coefficient of variation, the results of the order measurement and the disorder measurement of the similarity between process routes are fused, and the calculation of the fusion similarity value between process routes is realized. Finally, with the help of the idea of spectral clustering and the extracted cluster centers, the typical process routes are mined for reuse. Finally, based on three validation examples, it is shown that the proposed method can realize effective mining of high-value process reuse objects, which, in turn, can support manufacturing instance reuse more.