Artificial Neural Network (ANN)‎ modeling to predict the twin-screw extrusion processing variables of soy protein isolate and corn flour blend formulations on the physical properties of extrudates

Abstract EN

The objectives of this study were to develop soy protein isolate and corn flour blend cereal-like extrudates and to develop an artificial neural network (ANN) model to predict the physical properties of soy protein isolate and corn flour blend extrudates as a function of soy protein isolate content, moisture content and extrusion temperature.

as per the central composite rotatable design (CCRD), 20 processing conditions were selected with soy protein isolate content (33.2-66.8%), feed moisture content (31.6-48.4%) and extrusion temperature (126-194℃) and extruded products were developed using a twin-screw extruder.

the physical properties (expansion ratio, piece density, breaking stress and rehydration ratio) of the extrudates were determined as response variables.

An ANN model was developed to predict the physical properties of extrudates as a function of extrusion processing variables.

the results indicated that the ANN model could predict the expansion ratio, piece density, rehydration ratio and breaking stress of soy protein isolate and corn flour blend extrudates with an 89-98% accuracy depending on the physical properties.

the study also indicated that about 60% soy protein rich extruded cereals and snacks can be produced to replace the carbohydrate cereal and snack products.

the ANN model can be used to setup the extrudates production conditions in a commercial scale.