IIC Journal of Innovation 12th Edition | Page 23

Digital Twin + Industrial Internet for Smart Manufacturing: A Case Study in the Steel Industry accessibility, availability and elasticity at low cost through economies of scale. Furthermore, these technologies have matured, making it feasible deploy in small datacenters and small clusters of servers to enable small-scale distributed computing on the edge in the manufacturing environment—with the benefits of scalability, reliability and ease of management. On the other hand, due to the large amount of data expected to be stored and managed in the manufacturing environment, scale-out capabilities in big data are needed. Finally, machine learning modeling has increasingly become an analytic capability mutually supplementing the traditional first-principle-oriented modeling. Introducing machine learning capabilities in the manufacturing environment has become fruitful. Analytical model frameworks offer a unified execution framework that draws data from the data framework below it, running multiple analytic models as plug-ins simultaneously and efficiently. To complete closed feedback loops, insights drawn from the data analytics are combined with operational and business logics to transform into actions. Often, there are many applications involved in manufacturing processes. To avoid building new chimney- like closed applications, these applications are run and managed in a unified application development and operation (DevOps) environment. Such an environment would enhance the reliability of applications, decrease the effort in application development and reduce the complexity of system operations and maintenance management. Built on such a broad set of technologies as outlined above, an industrial internet platform for the manufacturing environment should seek to abstract a set of common functions that are required and shared by data-driven smart software applications and offer them as horizontal platform services to reduce the otherwise repetitive implementation of these functions in conventional architectures. These key common platform functions coincide with core elements of the industrial internet, namely data, analytical models and applications (implementing business logics). Furthermore, a Digital Twin framework offers a unified, systematic approach to represent, configure and manage the real- world objects in the digital space. It also provides a unified interface to the real-world objects for application development, akin to The data framework offers unified data collecting, processing and storing capabilities to achieve full lifecycle management of production data, avoiding the data silos commonly found in existing manufacturing environments. - 18 - November 2019