State feedback control supported by learning function based on neural network

The authors propose a state feedback control system supported by a self-learning system, considering its application to a heat exchanger. The system fractionalizes the heat exchanger along the direction of flow, calculates the state variables for each fraction segment using mathematical models, and feeds back those variables. This system simplifies the control law and makes possible the use of nonlinear mathematical models. In order to deal with the problem of performance drift of the heat exchanger, two automatic learning functions supported by a neural network are built into the system: the automatic mathematical model learning function and the automatic state feedback gain learning function. The effects of these ideas were demonstrated in sample calculations and in a simulator of an actual ultra-super critical pressure boiler