Development of a cost analysis-based defect-prediction system with a type error-weighted deep neural network algorithm

Abstract With the growing interest in smart factories, defect-prediction algorithms using data analysis techniques are being developed and applied to solve problems caused by defects at manufacturing sites. Cost benefit is an important factor consider, can be obtained applying such algorithms. Existing usually aim reduce error rate of prediction model, rather than focusing on cost for practical application models. Therefore, this study develops a algorithm considering costs systematization field application. To end, type error-weighted deep neural network (TEW-DNN) proposed that applies loss function set different weight each error, conducted search optimal weight. A analysis-based system designed TEW-DNN cyber-physical environment. The efficacy demonstrated through case involving die-casting factory South Korea.