A Novel Method for Product Quality Control Based on Production IoT Data

This study proposed a method of factor analysis to select key impact factors according to the results of importance values in production and to conduct quality prediction in continuous production process. Three different models were employed to compare the effectiveness and accuracy of factor searching and quality prediction. Through data analysis, compared with the empirical model by human knowledge with production IoT (Internet of Things) sensor data, the proposed data-driven based model in this paper can discover impact factors ignored by empirical model. For the quality prediction, the proposed models could also perform well in the aspects of correlation coefficient and standard deviation, respectively. The study in the paper found that the model using effective feature and correcting impact factors can effectively improve the predicting accuracy.