Knowledge Discovery For Quality Improvement In Finish Milling

Jin Cheng,Jian Wang
DOI: https://doi.org/10.1109/cac.2017.8243592
2017-01-01
Abstract:A novel knowledge discovery method for quality improvement in finish milling is proposed in this paper. Instead of quality experiment, the method is based on manufacture data applied in a more complex production environment to identify the factors leading to quality defects. Akaike's Information Criterion is modeled as an evaluation function to select process features based on condition where logistic regression is as the good-of-fit term. By differential evolution, the best process feature set is obtained in a certain condition. Then 19 conditions accompanied with quality defect in finish milling are extracted from historical records to analyze the cause. The result shows the relationship explored between processes and conditions can assist expert to make decision for production planning or maintenance to improve quality of strip.
What problem does this paper attempt to address?