Material Properties Prediction Model Using Sensitivity Based Linear Learning Method
Yahaya Mohammed Onimisi, Sunday Olusanya Olatunji · 2010
This paper presented a new prediction model for material property (strength of materials for gray cast iron) based on composition and microstructure using a recent learning algorithm called Sensitivity Based Linear learning Method (SBLLM). This method was proposed in order to address the problems of back propagation learning algorithm for feed forward neural network. Thus we have made use of this unique method to develop a material property model. Comparative studies have also been carried out between the proposed model and some earlier used methods, on same dataset, that include multiple regression, fuzzy regression, the generalized regression neural network and fuzzy neural network. The result obtained from the proposed model is reasonable and promising, and it is very stable and consistent in it prediction compared to others.