A new inference control system for the flame position in the combustion chamber of a power plant system boiler is presented. The system is based on enhanced multi-softsensing at three operational levels – basic level, providing estimates of all necessary technological variables; a separate Mill Fan (MF) level, and a total Dust Preparation System (DPS) level. The control system involves a subsystem for the stabilization of the position of the common MF ventilation rate momentum in a given threshold area in a burner horizon, which is supervised by an inference correction based on softsensed 2D flame position in the output section of the combustion chamber. A hybrid approach is accepted in softsensing, using fusion of the first principle models, statistical models, neural networks and fuzzy logic based models. Real experimental results are presented from TPP.

Publication timeframe:
4 times per year
Journal Subjects:
Computer Sciences, Information Technology