The neural network control application in a power plant boiler

Abstract

Two neural networks are used in the control of power plant boiler throttle pressure and megawatt load, where one network acts as an emulator, and the other as a controller. The learning scheme is a two-phase procedure in which the first involves training the emulator in mapping the plant dynamics and the second to train a controller network to learn the desired performance using a backpropagation algorithm and minimize plant output error cost function. This example illustrates the potential application of neural network technique in the power plant control area.