ANN (Artificial Neural Network)

After years of software and signals analysis developing, MaGyc has integrated the neural network concept into the MG-ST-100 accompanying software, aimed at testing parts and components which exhibit a particular vibrational behavior, in that the resulting acoustic signature is of very short time duration, or otherwise described, is heavily damped. In order to increase the good fake detection, MaGyc has introduced neural network concept inside the software algorithm. This strategy allows to apply the biological model, where hundreds neural circuits learn from past experiences. Within the human brain (as well as in many other animals) the resolution of cognitive problems is entrusted to networks neural. These networks, the extent and magnitude varies depending on the required task, may involve different brain regions and their development and training has been fundamental in the evolution of the human species. The unit under test (UUT) classification has been always a difficult activity, in particular for products with complex signals affected by noise. Finally, will be possible to integrate in an automatic testing bench day by day operator experience in identification and evaluation of multiple defects on a component.

Su Giulia Rossi