Free-model based neural networks for a boiler-turbine plant

The concept of the free model is implemented in the design of neuro-identifier and neuro-controller in the model reference adaptive inverse control scheme for a boiler-turbine plant. The free model approximates the future outputs of a system on the basis of the historical changes in the past and possible inputs currently applied. This feature makes it possible to identify and control a nonlinear system for which a mathematical model is not known or uncertain, such as boiler dynamics in a power plant. In order to be more economical on the fuel burning at the throttle valve, the functional mapping is applied in the reference model for the drum pressure set point from the power demand. The integrator in the reference model reduces the large initial errors for the neuro-controller training. After the training is completed, the trained neural networks are applied in the control scheme and tested under various operating conditions. It is shown that the proposed scheme can be applied to a nonlinear boiler-turbine plant with satisfactory performance globally