Abstract
An unsupervised learning technique and an associative memory have been used for encoding weights by a special type of Petri network named Petri neural net for modelling the influence of alloying elements on the final property of the high strength low alloy steel. The combined effects of alloying elements for different strengthening mechanisms is predicted when weights and threshold values are chosen on the basis of metallurgical understanding. The technique is found to be effective to create an associative memory of input-output relations in unknown data sets so that the same can be subsequently be used as a predictive tool.
