INTELLIGENT DEFECT PREDICTION IN METAL ADDITIVE MANUFACTURING USING PROCESS DATA ANALYTICS

Abstract

Metal additive manufacturing has emerged as a transformative production technology for fabricating complex, lightweight, customized, and high-performance metallic components across aerospace, automotive, biomedical, energy, tooling, and advanced engineering applications. Despite its advantages, the industrial adoption of metal additive manufacturing remains constrained by defects such as porosity, lack of fusion, keyhole formation, balling, cracking, residual stress, delamination, dimensional distortion, surface irregularity, and inconsistent microstructural characteristics. Conventional quality assurance approaches frequently depend on post-process inspection, destructive testing, fixed process windows, and operator experience, which can identify defects only after substantial material, machine time, and energy have already been consumed. This paper proposes an intelligent defect prediction framework for metal additive manufacturing using process data analytics. The proposed methodology integrates machine configuration data, process parameters, layerwise sensor streams, thermal observations, meltpool characteristics, image-derived indicators, environmental context, equipment condition, material information, and historical quality outcomes. A multi-stage analytics pipeline performs data acquisition, synchronization, quality validation, feature engineering, contextual fusion, intelligent model training, defect risk classification, explainability, digital twin synchronization, and adaptive quality decision support. The framework predicts major defect categories and identifies abnormal process states before final component completion. Machine learning models evaluate interactions among laser power, scanning speed, layer thickness, hatch spacing, thermal history, meltpool behavior, build position, powder characteristics, and equipment state. A digital process twin maintains the evolving condition of each build, while secure API services support controlled integration with manufacturing execution, quality management, and analytical platforms. Representative experimental analysis indicates that the proposed framework can improve defect prediction accuracy, increase recall for high-risk build regions, reduce false alarms, decrease scrap rate, shorten inspection effort, and improve process stability compared with fixed-threshold and isolated machine learning approaches. The proposed framework provides a scalable foundation for predictive quality assurance, data-driven process optimization, and intelligent production control in next-generation metal additive manufacturing environments.