Support Vector Regression Based Model Identification of a Batch Reactor Process: A Simulation Study

Abstract

The present study emphasizes the use of Support Vector Regression-based model identification for acrylamide polymerization reaction in a batch reactor process with three kinds of SVR kernels. The real-time plant is identified by solving the differential equation models of the plant. The first model identification approach is done using Linear Kernel, the second model identification approach is done using Polynomial Kernel, and the third one is done using Radial Basis Function. In all three approaches, the cost function is defined for equal penalization of all errors. The study emphasizes machine learning-based model identification of the non-linear system.