Neural networks in extrusion process identification and control

摘要: Abstract Although neural networks have become one of the key research objects within artificial intelligence, relatively little information is available on related to food process control. The interest in such areas as dynamic modelling processes has increased, not least due dramatic improvement and availability calculation methods hardware. In present case, flat bread extrusion was used an example process. Dynamic changes torque, specific mechanical energy (SME) pressure were identified (modelled) controlled using two independently taught feed-forward (ANN). SME, torque are system parameters which can be with parameters, feed moisture, mass rate screw speed. Target product expansion index, bulk density, etc. normally difficult measure on-line, but estimated functions parameters. For whole cooking a MIMO (multi input multi output) approach necessary. network topology for model 21-9-3 controller 18-20-2. 629 real data samples 115 synthetic created model. When testing controller, SME set points quite well reached. One clear advantages design ease constructing complex controller.