Gains:
- Can list what the shift report should include
- Can convert raw SCADA data into manageable reports
- Knows the limits of report production with artificial intelligence
SCADA systems record hundreds of data points per second. This raw data is not read by anyone. The shift report is a matter of turning that pile into a one-page decision.
What makes a good shift report
- Production: milled wheat, obtained flour and by-product, calculated yield.
- Quality: ash and moisture values of the samples taken, deviation from the target.
- Stoppages: duration, reason, person intervening.
- Energy: consumption and kWh per ton.
- Alarm summary: recurring alarms are highlighted.
- Shift note: operator's observation (information not included in the number).
Contribution of artificial intelligence
Producing this summary from raw data is a repetitive task, and AI does it well. More valuable is his interpretation of the deviation: a connection such as "efficiency is 0.6 points lower than yesterday's shift; C2 motor current increased by 8% in the same period" may escape one's eyes.
However, artificial intelligence does not tell the reason, it tells the relationship. "Current increased, efficiency decreased" is an observation; The reason may be roller wear or feed imbalance. The decision is made by the miller.
Automate the report but do not remove the operator note field. The uncountable observation ("the wheat was dustier today") is often the most critical clue.