A product quality monitoring framework using SVM-based production data analysis in online shop floor controls

Abstract

As the manufacturing supply chain is getting more global and complex, the real-time prediction of product quality is becoming a critical issue in global manufacturing business, especially in the automotive industry, where most subcontract enterprises still lack a systematic and fast quality assessment for their products in the supply chain. On the manufacturing shop floor, product quality is still assessed by total visual inspection in the post-manufacturing stage, which requires a lot of time and human resources to manage quality issues. In this paper, a real-time, inprocess, and remote quality monitoring system for small and medium sized manufacturing enterprises is proposed to provide an online quality monitoring framework using real-time production data. The proposed framework imposes a real-time quality assessment tool based on a support vector machine (SVM) algorithm that enables users to classify the product quality patterns from the in-process production data. At the end of this paper, the door trim production data from an automotive company is used to verify the proposed quality monitoring/prediction model.