Modeling and control of pH neutralization using neural network predictive controller

Abstract

The difficulty of controlling pH neutralization processes resides in the non-linearity of such processes. This behavior is due to the logarithmic relationship between the hydrogen ions concentrations [H <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> ] and the level of pH. The control strategy to be developed very much depends on the feasibility of the mathematical model that represents the process. This paper illustrates feasible modeling of the pH neutralization plant using empirical techniques and investigates the performance of an artificial neural network predictive controller against the more traditional PID controllers. As a conclusion, a feasible empirical model was found closest to a second-order with dead time. The artificial neural network predictive controller has outperformed the conventional PI /PID controllers.