Which is your choice?

Let’s assume that you have to integrate data from 10 PLCs, whose scan time is 12 msec for each, to feed up data to A.I(Artificial Intelligence), and the mission of A.I is to predict the time when machines such as motors should be replaced or repaired. All data from 10 PLCs is scanned in every 10 msec in system configuration a) and 50 msec in system configuration b). Then which is your choice?

As we know, all mechanical components in industrial fields would show anomalistic phenomena before going into the permanent fault status that is very serious in factory operation for continuous processes. Unfortunately these precursors would rarely happen at beginning status and even the their alive interval is too short to detect. As time goes on, these anomalistic phenomena will be accumulated, and at last machines would be dead.

So the best way to predict the time when machines should be replaced or repaired before permanent faults is to detect these anomalistic phenomena early if possible and make patterns with A.I. It means that data acquisition should be fast enough to detect these precursors. If the scan time of data acquisition is faster than the scan time of PLC, then it is best because data which can be acquired physically would not be lost. Even if the devices such as PLC would not support fast communication speed, then the scan time of data acquisition should be fast as much as possible.

SoftDCS has super fast scan time for data acquisition which would not lose any data, and the system configuration a) and its performance were tested and proved on actual running PLCs in POSCO. Which is your choice if you have to feed up data with good quality into A.I? For related video, see this.