Predictive Modeling for Early Detection of Quality Defects in the Output of an Aerospace Batch Manufacturing Process Open Access
Downloadable ContentDownload PDF
Quality defects from the output of batch manufacturing processes are a major cause of disruption within the aerospace industry supply chain. Nonconforming parts result in late deliveries and the accrual of additional costs, including repair and scrap costs. The quality of the parts must be within a specified range and are determined by conformance to the design specification during post-manufacturing inspection. Current industry research focuses primarily on predictive maintenance as a measure for early detection of manufacturing equipment failures. This research provides an approach for forecasting defects in aerospace batch manufacturing process using a time series regression predictive model by focusing on the historical critical quality attributes data as the primary model input. Time series regression is used to provide the probability of defect for a part within a specified future time frame. Predicting impending product quality defects will provide a batch manufacturing organization the ability to take the necessary preventative action. The proposed method shows 97% accuracy in prediction and four times improvement as compared to the industry standard method of statistical process control.