Method of model´s parameters classification using neural network for meat characterization

The main objective of this work is to develop the classification application for a new promising method for meat characterization getting information about the state of the vacuum packed meat in supermarket. A supervised training using neural networks and according to the back error propagation method is used. The training ensure a classification with high precision and with ability to answer correctly the inputs then with ability to classify the erroneous inputs which do not exist in the data base, without creating new classes. Method classification consist to classify the model parameters of the physical model of the meat according to a data base including the model´s parameters for different beef muscle in different days.