Modeling of the Multilayer Perceptrons for Image Recognition of the Steel Microstructures

The relevance to develop new methods and means of automated metallographic analysis with the possibility to diagnose of new metals has been substantiated. The usage of the neural networks to diagnose the metals condition has been proposed. The multilayer perceptrons to recognize the steel microstructures have been modeling. The lowest mean square error of the proposed multilayer perceptrons for image recognition of the metal microstructures has been obtained.