The Application of Neural Networks for Studying Phase Transformation by the Method of Acoustic Emission in Bearing Steel/ Zastosowanie Sieci Neuronowych Do Badania Przemian Fazowych W Stali Łożyskowej Metodą Emisji Akustycznej

Abstract

Abstract The research was carried out on steel 100CrMnSi6-4 under isothermal austempering resulting in forming the duplex structure: martensitic and bainitic. The kinetics of transformation was controlled by the acoustic emission method. Complex phase transformations caused by segregation and carbide banding occur at the low-temperature heat treatment of bearing steel. At the temperature close to M S , a certain temperature range occurs where an effect of the first product of prior athermal martensite on the bainitic transformation can be observed. In the registered signal about 15 million various events were registered. There were considered three types of acoustic emission events (of high, medium and low energy), with relatively wide sections and with different spectral characteristics. It was found that the method of acoustic emission complemented by the application of neural networks is a sensitive tool to identify the kinetics of bainitic transformation and to show the interaction between martensitic and bainitic transformations.