Facilities collaboration in cloud manufacturing based on generalized collaboration network

Wenxiang Li, Chunsheng Zhu, Edith C. H. Ngai, Laurence T. Yang, Lei Shu, Yuxia Sheng

Introduction

Cloud manufacturing aims to transform individual factories into a shared pool of resources, but coordinating these resources across a regional industrial cluster is a massive logistical challenge. Traditional scheduling often overlooks the "collaborative" nature of production—where multiple machines or facilities must work in tandem to complete a task.

This paper introduces the Facility Collaboration Network (FCN), a complex network model designed to optimize how dispersed manufacturing resources interact. By analyzing real-world data from a Chinese industrial cluster, the authors demonstrate how different facility selection strategies—ranging from pure random assignment to load-balanced preference—drastically alter the efficiency, connectivity, and "social" structure of the manufacturing ecosystem.

Methodology

The researchers model the manufacturing environment using Complex Network Theory. Their approach involves a Bipartite Mapping, where they first create a two-part graph connecting "Tasks" to "Facilities."

Bipartite mapping between tasks and facilities.

This bipartite graph is projected into a one-mode Facility Collaboration Network (FCN). In this network, nodes represent facilities, and an edge is drawn between two facilities if they have collaborated on at least one shared task. To see how the network evolves, they test four distinct scheduling algorithms: Random Selection (RAN), Random with Preference (RAN-P), Balanced Selection (BAL), and Balanced with Preference (BAL-P).

The study utilizes a real-world dataset from a packaging and printing industrial cluster in Hubei Province, China, covering 2011 to 2013. The scale includes 25 enterprises and 2,007 facilities of 51 different types, handling approximately 5,500 task orders.

Results and Discussion

The analysis reveals that different selection strategies lead to vastly different network topologies. Random selection strategies (RAN and RAN-P) result in a "scale-free" network, where a few "hub" facilities handle a disproportionately large number of tasks, mirroring the behavior of social networks. These strategies also consistently maintain a "small-world" property, with an average shortest distance of ~2 to 3.

Growth of active vertices (facilities) under different selection strategies compared to real-world data.

As shown in the vertex growth analysis, Balanced strategies (BAL/BAL-P) activate almost all available facilities (~2,000) very quickly to distribute load, whereas random strategies see a much slower, continuous increase in the number of active nodes. Interestingly, by comparing simulations to actual industrial data, the authors found that real-world behavior is a hybrid of random selection and preference-based selection, rather than a perfectly optimized load-balanced system.

Furthermore, strategies that favor specific enterprises (the "-P" variants) utilize significantly fewer facilities overall, leading to higher network density. In late-stage scenarios, the BAL-P strategy shows a dramatic spike in high-order cliques (groups of 4 or 5 facilities), suggesting that as a balanced network matures, it naturally forms highly specialized clusters.

Conclusion

The FCN model provides a formal framework to describe how facilities connect, helping to predict system bottlenecks and evaluate load-balancing strategies. While the current model treats collaborations as unweighted, it establishes a foundation for future research into "collaboration intensity" and the impact of irregular task arrival patterns on system stability. This work is crucial for ensuring cloud manufacturing platforms remain resilient and efficient as they scale.

Summary updated

Abstract

In cloud manufacturing for regional industrial cluster, there is increasing necessity of collaboration among enterprises or facilities. It is valuable to explore the characteristics of these collaboration behaviors for effectively scheduling dispersed manufacturing facilities and organizing their collaboration. The collaborative relation of manufacturing in regional industrial cluster can be described as a generalized social collaboration network. In this paper, we introduce the relevant entities and relations of facilities collaboration, and propose the method for building Facility Collaboration Network (FCN). We further design the dynamically growing process of FCN for different facility selection strategies, including random selection, balanced selection, random selection with preference and balanced selection with preference. Based on the metrics such as network scale, node degree distribution, act degree distribution, average shortest distance and number of cliques, we present the statistical characteristics of FCN, and analyze relevant characteristics and laws for efficient facilities selection in cloud manufacturing.
BibTeX
@inproceedings{li2015facilities,
  author = {Wenxiang Li and Chunsheng Zhu and Edith C. H. Ngai and Laurence T. Yang and Lei Shu and Yuxia Sheng},
  title = {{Facilities collaboration in cloud manufacturing based on generalized collaboration network}},
  booktitle = {QSHINE},
  year = {2015},
}
Figure