Optimizing Reliability and Quality in AI-Based Manufacturing Diagnostics: Machine Learning and Risk Assessment Perspectives  
Author Kuo-Yi Lin

 
Co-Author(s) Ying-Hong Yang; Meng-Hua Li; Ming-Lang Tseng

 
Abstract In response to the increasing demand for intelligent transformation in manufacturing, this study proposes a machine learning-driven diagnostic framework that integrates risk assessment models to enhance reliability, quality, and operational efficiency. Leveraging real enterprise diagnostic data, key AI-driven application scenarios—including process failure prediction, quality optimization, and cost reduction—are systematically constructed. To achieve a data-driven reliability assessment, this framework incorporates Failure Mode and Effects Analysis (FMEA) and predictive reliability modeling, enabling the identification of critical failure risks and guiding robust design decisions. Additionally, it utilizes structured and unstructured production process data to train AI models for anomaly detection, defect classification, and predictive optimization. The integration of machine learning algorithms with probabilistic risk modeling allows for continuous system adaptation and uncertainty quantification, improving overall diagnostic accuracy. Empirical validation demonstrates that the proposed approach enhances equipment reliability, reduces unplanned downtime, shortens production cycles, and improves decision quality. By providing a scalable and AI-enabled intelligent diagnostic consultation system, this research contributes to the advancement of reliability engineering and quality-driven decision-making in modern manufacturing.

 
Keywords UNISON, Reliability Modeling, Risk Assessment, Quality Optimization, Intelligent Diagnostic System, Industry 4.0, Industry, Innovation and Infrastructure
   
    Article #:  RQD2025-236
 

Proceedings of 30th ISSAT International Conference on Reliability & Quality in Design
August 6-8, 2025