Prediction of Multi-class Industrial Data

Abstract

Industrial plants use many different sensors for processes monitoring and controlling. These sensors generate huge amount of data. These data should be used for improving of the quality of semi and final products in each factory. In this paper, we describe processing of two different datasets acquired from a steel-mill factory using three different methods SVM, Fuzzy Rules and Bayesian classification. Moreover, we describe problems of each method with confrontation with real data. Each of the method used works in different algorithm and is not based on the same theory. Their comparison gives a nice review of the real application of these methods.