Case-based reasoning system based on Bayesian rough set and hierarchical mixture of experts model

Abstract

An efficient case retrieval method and an adjustment strategy are proposed in this paper to build a case-based reasoning (CBR) system for oxygen calculation in Basic Oxygen Furnace (BOF) steelmaking. In the process of case retrieval, the Bayesian rough set technology is adopted to establish the weights of the case attributes. Then, the k nearest neighbors algorithm is implement to retrieval the most similar cases as a reference. The adjustment step executed by mixture of experts model is introduced to make up the gaps between current case's problem attributes and the retrieved case's. Finally, the parameters in mixture of experts model are optimized by Particle Swarm Optimization (PSO) method. Practical production data are used to test the CBR system. Using actual production data converter simulation Results show that proposed system is effective.