IIC Journal of Innovation 8th Edition | Page 35

The Path from Data to Actionable Information as a Driver for the Industrial Ecosystem Production time can be further broken down into the following categories in Figure 6 21 : Figure 6: Production Time schedule to determine if the machine is operating when it was scheduled to be operating; allowing one to discern Plant Operating Time, Scheduled Operating Time from Potential Production Time and compute Delay Time as well as Lost Production time, which is the Production Time minus the Potential Production Time. To categorize time, one needs to understand the equipment’s current operation states and when essential work is being done necessary to producing parts, even if it is ancillary to the production process. One should start by getting as much data from the controller as possible, use proprietary APIs or binary signals from PLC terminal blocks. The data must be translated into a timestamped stream of tagged values. Q UALITY The quality metric [3] is often ignored in discrete manufacturing since most manufacturing processes have multiple steps and inspection is often performed at the end. There are problems attributing the quality slip to the device and process step since there are many operations that create a single feature; it is often impossible. In most OEE systems, quality is reported at 100%, unless a capability exists at the machine to report scrapped parts or the operator identifies bad parts and enters the data manually. The next phase translates the data from the tag value pairs to the MTConnect standard by taking the tagged data and converting units and determining standardized machine states. After the semantic conversion, rules can be applied to multiple machine vendors and models. Sensor data can be analyzed to identify anomalous conditions and translate the conditions into MTConnect semantics using machine learning. The initial categorization is done during the enrichment stage where the semantic data are matched with patterns that indicate production, repair, setup and non- productive. At the next stage, ecosystem integration with MES gives the plant Processes must be verified at each step for OEE to work properly. When we increase the part mix and variation, it is even more imperative that every step is verified. The - 34 - June 2018