Lime Kiln Process Identification and Control: A Neural Network Approach
Complex systems exhibiting strong non linearities and time delays such as chemical processes are very demanding in control requirements. In this paper we present a neural network approach for multivariable non-linear kiln process identification and control. Neural networks, in control theory, are attractive because of their powerful capabilities to successfully approximate nonlinear functions within a specified approximation error as recent research has proven. They can be used to synthetize non-linear controllers for non-linear processes and it is expected that better results can be obtained as compared to more conventional methods.
The main objective of this work is to train the neural network kiln controller to provide suitable control inputs that produce a desired kiln response. If the neural network plant model is capable of approximating well and with sufficient accuracy the highly non-linear calcination process in the lime kiln, then it may be used within a model based control strategy.
Firstly, the lime kiln process identification is achieved using a feedforward Artificial Neural Network (ANN), namely the plant model. It learns the kiln dynamics through a training process that drives the error between the plant output and network output to a minimum. The nonlinear mapping from control inputs to plant outputs is achieved through the use of the backpropagation learning paradigm. The specifications of the neural network to provide the desired system representation are given.
Secondly, a neuralcontroller was designed to adaptively control the non-linear plant. The neural network topology was selected according to common used performance criteria.
Simulation results of non-linear kiln process identification as well as non- linear adaptive control are presented to illustrate the neural network approach. Analysis of the neural networks performance is underlined.
KeywordsNeural Network Neural Network Approach Manipulate Variable Internal Model Control Fuel Flow Rate
Unable to display preview. Download preview PDF.
- D. Rumelhart, J. McClelland, and the PDP Research Group, Parallel distributed processing: explorations in the micro structure of cognition. Vol. 1 2, MIT Press, 1986.Google Scholar
- R. Hecht-Nielsen, “Theory of back propagation neural network,” in Intl. Conf. Neural Networks, pp. 593-605, IEEE, 1989.Google Scholar