This harvested data also has inherent value: statistical tools look for patterns and contexts in the information from which interesting anomalies emerge. These anomalies can help predict when a machine is not functioning normally and prevent it from failing well in advance by highlighting which parts need to be serviced. “¬Predictive ¬maintenance” is one of the most important trends in engineering 4.0: companies can now avoid failures due to avoidable defects and save on the cost of expensive routine system maintenance, which may not even be necessary.
“Consolidating the necessary data and developing explanatory models is currently one of the most demanding tasks in the Industrial Internet of Things,” says IBM IIoT expert Holtmann. Ideally the system would not only identify the maintenance needed but also take on more of a project management role by suggesting actions to take based on other data sets: for example, when is the next service, when is a specific technician available and what are the latencies for each process affected by the machine being shut down. “In the most comprehensive scenarios, the alternative actions created by the computer flow directly into production control,” says Holtmann.