As industrial technologies develop, the manufacturing industry is globally changing in more automated and complex manners, and the prediction of real-time product quality has become an essential issue. Although many of the physical manufacturing activities are getting more automated than ever, there still exist many uncovered parameters that, either directly or indirectly, affect the product quality. In many manufacturing sites, the quality tests in their processes still rely on few skilled operators and quality experts, which requires a lot of time and human efforts to manage the product quality issues. In this thesis, thus, a real-time/in-process quality monitoring system for small and medium size manufacturing environments is proposed to provide the data-driven product quality monitoring system framework. The proposed framework consists of a product quality ontology model for complex manufacturing supply chain environments, and a real-time quality prediction tool using the support vector machine (SVM) algorithm that enables the quality monitoring system to classify the product quality patterns from the in-process production data. Additionally, we propose a framework for analysis of the quality inspection results from the monitoring system with respect to quality costs, including inspection and warranty costs. In addition, this thesis establishes a relationship between the warranty cost and the severity of customer-perceived quality. Finally, we suggest a future work that a prescriptive product quality assessment concept using the Hidden Markov Models (HMM) that analyze and forecast possible future product quality problems using production data from manufacturing processes based on data flow analysis. Also, a door trim production data in an automotive company is illustrated to verify the proposed quality monitoring/prediction model.
Publisher
Ulsan National Institute of Science and Technology