Defect Prediction in Hot Strip Rolling Using ANN and SVM

One of the largest factors affecting the loss for steel manufacturing are defects in the steel strips produced. Therefore the prediction of these defects forehand would be very important. In this study we used classifiers - feedforward neural networks and a support vector machine - to solve this problem. We also used different kinds of feature selection methods such as a preprocessing step for the classifiers. As a result, these two classifiers confirmed the same grade of classification error in this study. Keywords. Hot steel rolling, feature selection, classification, neural networks, support vector machine

• Type: Conference
• Year: 2008
• Where: SCAI
• Authors: Manu Hietaniemi, Ulla Elsilä, Perttu Laurinen, Juha Röning
• Keywords: Artificial Intelligence, Feature Selection, Neural Networks, SCAI 2008, Support Vector Machine