Defects are unavoidable during manufacturing processes, and a tremendous amount of research aimed at improving defect prevention has been conducted by scholars. Zero Defect Manufacturing (ZDM) seeks to eliminate defects in production. In addition, technological advancements now allow the repair of defective products. This creates the need to re-schedule productions more frequently in order to take into account the actions necessary to fix defective parts. This study focuses on detection and repair-based ZDM strategies. It implements a newly developed, hybrid Decision Support System (DSS) that uses data-driven and knowledge-based approaches to detect defects and then automate the necessary decision-making processes. The system uses an ontology based on the MASON ontology in order to describe the production domain and enrich the available data with contextual information. Real time production data and past knowledge are utilized to analyse defects, identify their type and severity, and suggest alternative repair plans. Possible repair plans are evaluated using a dynamic multi-criteria approach that determines the plan most suited to production conditions at the time of defect detection. To test the efficacy of the DSS developed for this study, it was integrated with a dynamic scheduling tool and was also used in an industrial application in the semiconductor domain. The simulations and the real-world implementation both show that the proposed DSS system can efficiently detect defects and automate the post-detection decision-making process. The multi-criteria approach adopted by this study proves that the DSS can make well-adapted decisions, handle the dynamic nature of a production system, and help manufacturers move closer to Zero Defect Manufacturing.

在制造过程中缺陷是不可避免的，学者们进行了大量旨在提高缺陷预防的研究。零缺陷制造 (ZDM) 旨在消除生产中的缺陷。此外，技术进步现在允许修复有缺陷的产品。这导致需要更频繁地重新安排生产，以便考虑修复缺陷零件所需的操作。本研究侧重于基于检测和修复的 ZDM 策略。它实施了一个新开发的混合决策支持系统 (DSS)，该系统使用数据驱动和基于知识的方法来检测缺陷，然后自动执行必要的决策过程。该系统使用基于 MASON 本体的本体来描述生产领域并用上下文信息丰富可用数据。利用实时生产数据和过去的知识来分析缺陷，确定其类型和严重程度，并提出替代维修计划。使用动态多标准方法评估可能的维修计划，该方法确定最适合缺陷检测时生产条件的计划。为了测试为这项研究开发的 DSS 的功效，它与动态调度工具集成，并且还用于半导体领域的工业应用。模拟和现实世界的实现都表明，所提出的 DSS 系统可以有效地检测缺陷并使检测后决策过程自动化。