The progress in machine learning technologies allowed us to develop another method for the optimization of the cement and clinker production: data analysis.
Process engineers spend most of their times looking at trends from the control system and trying to find out how certain process parameters influence others and how the performance of the plant can be increased. Given the advances in artificial intelligence, this is now a much better task for an automated data-analysis tool:
- Our tools can handle many hundreds of data signals while the human brain is limited to only a few dozen.
- Our tool does not need to compress and average the data in order not to get lost, but it can find interrelations even in high frequency data of 1 Hz and higher.
- Our data analysis tool can provide not only analysis but also basis for a data-driven prediction. We can teach an artificial neural net to behave like the cement plant and then use this model for prediction, operator assistance and closed loop control.