Coal Mill Modeling by Machine Learning Based on on-Site Measurements

This paper presents a novel coal mill modeling technique using genetic algorithms (GA) based on routine operation data measured on-site at a National Power (NP) power station, in England, U.K. The work focuses on the modeling of an E-type vertical spindle coal mill. The model performances for two different mills are evaluated, covering a whole range of operating conditions. The simulation results show a satisfactory agreement between the model responses and measured data. The appropriate data can be obtained without recourse to extensive mill tests and the model can be constructed without difficulty in computation. Thus the work is of general applicability.