Abstract: The concept of "Service-oriented Holonic Manufacturing Network"(SHMN) is put forward as a new business model, which integrates the advantages of Holonic Manufacturing Systems (HMS) and Service-Oriented Architectures(SOA). HMS is a manufacturing system where key elements, such as machines, cells, parts, products, operators and working teams, etc., are modeled as manufacturing holons, Which have both "autonomous" and "cooperative" properties. The cooperative relationship among holons to achieve system reconfiguration and optimization, make HMS a new paradigm with great potential to meet today's agile manufacturing challenges.
Complicated with cooperation, the SHMN begins to show complex network's characteristics. In this paper, the holon collaborative demand problem of an SHMN based on the complex network theory is analyzed. And an evaluation method is presented considering intuitional normal fuzzy number. Specifically, the traditional SHMN mainly has integrated star networks, chain networks that formed by the manufacturing industry chain extending to both ends, combined networks of production and manufacturing industry chains, etc., and all of above networks are regular networks with full links. As the integration of manufacturing and service deepens, the SHMN evolves with product services bundled with services, multiple services, multi-dimensional services, and full lifecycle services. At the same time, holon collaborative demand of SHMN changes with a certain probability, so it's possible to form a complex SHMN. Further, the holon collaborative can be both synchronous and synergistic. The collaborative demand evaluation problem considering the holon links in SHMN in the synergistic state. According to the stability of similarity measurements of SHMN in the synergistic state is discussed, the evaluation method is designed. Firstly, an evaluation index on the basis of composite and recursive collaboration for holon is constructed. Secondly, the holon collaborative demand using the similarity stability measurements of nodes is studied and the holon's network structure entropy is measured. Thirdly, a two-way feedback mechanism for network is studied. And finally using INFCWA(INFCWA_R) operators or INFCWG(INFCWG_R) operators for the information aggregation, the potential application value of the method is illustrated by an example. And the illustration takes 247 holons as SHMN. Revised by nodes' stability of similarity measures, it is found holon collaborative demand corrected by COS,SAL,JAC and SOR (the second type) is different from CN and AA (the first type), LHN is similar to the second type, and PC is similar to CN, and PA is different from all above. Assuming the SHMN is stable, methods performances are compared, where the recursive coefficient is 0.6, the evaluation group could draw a conclusion in 10 rounds considering holon collaborative demand, and without considering the negotiation need 40 rounds or more.The method proposed in the paper could not only overcome the limitations of the traditional manufacturing network but also compensates for the SHMN with complex networks' characteristics, to a certain extent.

Key words: SHMN, collaborative demand, similarity stability, two-way feedback