Material property prediction using neural-fuzzy network
Minyou Chen · 2002
A neural-fuzzy network based adaptive fuzzy modelling approach that includes the initial fuzzy model self-generation, significant input selection, partition validation and parameter optimisation was developed for alloy material property prediction. In this approach, the whole procedure of structure identification and parameter optimisation is carried out automatically and efficiently. The proposed adaptive fuzzy modelling approach has been used to construct composition microstructure property fuzzy models for hot rolled alloy steels. Simulation studies demonstrate that the predicted mechanical properties have good agreement with the measured data by using the obtained fuzzy model with only a few rules.