Comparison of linear and nonlinear stati

This paper presents the comparison of Multiple Linear Regression Analysis (MLRA) and Artificial Neural Networks (ANN) as the statistical analysis tools. The main objective of this study was to investigate the applicability and constraints of these linear and nonlinear methods of statistical analysis. Both approaches were used for modeling the same data set obtained during an experiment conducted under industrial conditions, specifically focusing on aluminate solution decomposition as part of the Bayer alumina production process. Important statistical parameters influencing the choice of the modeling tool were evaluated through an analysis of real statistical data collected from the industrial environment. The investigation highlighted the merits and limitations of MLRA and ANN in effectively modeling the process parameters which included various input variables such as concentration of Na2O, caustic ratio, crystallization ratio, and temperature conditions. The degree of aluminate solution decomposition was considered as the output variable of the model. The findings of this research support the use of both statistical methods for different types of industrial processes, demonstrating their relevance and utility in process modeling.