Recurrent auto-associative artificial neural network model of Biomass Steam Boiler System

In the paper a recurrent auto-associative artificial neural network structure is used to obtain a dynamic model of Biomass steam boiler system. Upon offline real process data a model was derived and results were compared to model derived by classic identification method. A good dynamic model was needed for design, testing and tuning of the Fuzzy controller, which resulted in optimization of steam production. By using recurrent auto-associative neural network instead of locally adequate state space model, more general model with a better fit to the wide range of process data was achieved. Implementation of such complex neural network was tested on typical industrial PLC and compared to Matlab simulation results.