Internal Model Control for a Continuous, Snack Food Frying Process Using Neural Networks

Based on a developed neural network one-step-ahead prediction model with two inputs and two outputs, aninternal model control (IMC) loop is established for process control of a continuous, snack food frying process. In thismultivariable control loop, each control action is computed iteratively by a modified Newtons and gradient descentmethods to inverse the process prediction model equation at each time instant. Using this control loop, the time-lagsbetween the process inputs and outputs are compensated. Results of controller simulation of the continuous, snack foodfrying process are presented.