Knowledge Discovery for Support Structure Type Selection of Thrust Bearing Using Bayesian Decision Based on Rough Set

A new method of Bayesian decision based on rough set is proposed in order to mine tacit knowledge and latent rules in support structure type selection of thrust bearing. Firstly, rough set theory is applied to reduce all factors considered in type selection for getting determinative factors. By a heuristic algorithm based on improved mutual information, the minimal attributes reduction is obtained and makes up of decision table with decision attributes. Then according to the decision table, Bayesian decision with minimal risk is employed to extract decision rules. In this paper, the concise decision rules are extracted from representative cases and evaluation is made in some successful cases. Experiment results show that it is feasible and effective to use the method to knowledge discovery for support structure type selection of thrust bearing.