Smart Manufacturing Connectivity for Brown-field Sensors Testbed
least from today’ s perspective, the incorporation of communication capabilities in every sensor is too expensive. Furthermore, in terms of size, many sensors need to be very small to be incorporated into the appropriate locations at the machines and thus cannot be equipped with a receiver.
This testbed suggests using option 2 to extract data from the first aggregation level and to transfer that data through an additional communication path to the enterprise IT system without impacting the real-time operations.
The architecture proposed by this testbed can also be applied if there is a modern PLC and if it is beneficial to extract data close to the sensors, e. g., to reduce the traffic on the Industrial Ethernet system. However, modern PLCs are definitely suitable to process the required amount of data and many PLC suppliers offer modules with integrated OPC UA servers. So, with a modern PLC available, users would likely select option 3 above.
Another objective of the Testbed was to provide a retrofit solution for the factory floor. Many older machines are delivered without sensors integrated for diagnosis purposes. Thus, there is a distinct potential of increasing the Overall Equipment Effectiveness( OEE) by adding sensors and running machine learning algorithms.
Machine learning and advanced analytics are currently not used in the Testbed. Simple monitoring and analytics applications were used within the validation usage scenario, but it was not the focus of the Testbed. Nonetheless, this could be a point of connection with additional IIC members to perform advanced analytics on the Testbed.
Elements of the Technical Solution
This testbed introduced a special IO module which connects up to 8 IO-Link sensors and which provides interfaces to the real-time control system( Industrial Ethernet) and the enterprise IT system( TCP / IP) at the same time and that is why the name“ Y-Gateway” is used for it. The Y-Gateway is a retrofit-able hardware solution intended to substitute classical IO modules where data with relevance for higher-level processes are passing through. This approach creates a new proponent suitable for brown-field installations; namely, extracting data from where it is actually generated while the original system architecture is preserved. Existing cabling can be re-used.
The Y-Gateway implements computational and storage capabilities and is thus able to perform some pre-processing of data from the eight IO-Link sensors and to buffer data for later bulk retrieval if required by the application.
There are some estimations that only about 5 % of the sensor data are actually used by the governing PLC for running the automation task. As explained above, elder PLC are typically not suitable to process more than that, so the Testbed’ s sensor-tothe-cloud approach enhances the data availability for brown-field installations by up to a factor of 20. This higher quantity of data is especially interesting when considering to implement machine learning algorithms at enterprise IT level that require a higher volume of training data prior to actually running real-time analysis tasks.
September 2018- 80-