Identification and modeling of a rotary kiln in cement plant based on ANN (MLP)

Abstract

The objective of this study is to identify and model a rotary cement kiln based on production history data by using an artificial neural network MLP algorithm. The usefulness of this algorithm is that it provides a reliable empirical relation between the inputs parameters (Flow, Temperature, and pressure) and the outputs, which indicate the cement quality. Where, the most critical process in a cement production facility is cooking the mixed raw material in a rotary kiln; its task is to gradually burn and bakes a suitable mixture of input material to produce clinker. Therefore, the rotary kiln is the most important part in a cement factory. From another side, the control of a cement kiln is a complex process due to many factors namely: The Non linearity of the system caused by the chemical reactions, its dynamic and high dimensionality. Therefore, identification, modelling, prediction and simulation of Kiln system is very crucial step in managing and optimizing the cement production. Since the ANN has demonstrated its effectiveness in identifying a large class of complex nonlinear systems, it has been proposed in this case study to model cement Kiln of plant based on Multi-Layer Perceptron (MLP) approach. The MLP algorithm has been trained by using history data of twenty four months, and it has been tested and validated through comparison with production data of the next six months after the training. The obtained results have demonstrated the superiority of the proposed ANN approach over the conventional modelling approaches