Laser additive manufacturing (LAM) is an important method for fabricating complex components in aerospace, medical, national defense, and military sectors. However, owing to factors such as molten pool dynamic behavior and steep temperature-gradient variations, forming defects such as balling, pores, lack of fusion, keyholes, and cracks tend to form during the LAM process, severely affecting the mechanical properties, fatigue performance, and service reliability of components. At present, predicting and controlling forming defects in LAM-fabricated parts and improving forming quality and performance are core demands of technological development. Reviewing the research status of defect detection and characterization and process monitoring can promote the high-quality development of intelligent monitoring and closed-loop control systems for LAM.

Significant research progress has been made in the field of detection and monitoring of forming defects in LAM in recent years. Existing studies classify LAM forming defects into two major categories: porosity-type and crack-type defects. The differences among various defects in terms of formation stage, location distribution, physical mechanisms, and morphological and dimensional characteristics are shown in Fig.2 and Table 1. Regarding detection techniques, destructive testing based on metallographic sections, microstructural observation, and high-resolution characterization can effectively reveal the morphological features and formation mechanisms of pores, lack of fusion, and cracks. Nondestructive testing methods such as X-ray, ultrasonic, eddy current, and active thermography enable rapid identification and quantitative assessment of surface, subsurface, and internal defects. A comparison of the applicable scenarios, inspection regions, and output features of different nondestructive testing methods is presented in Table 2. For online monitoring, monitoring methods based on optical, acoustic, thermal, and other sensor signals can acquire process-related physical quantities, including molten pool behavior, spatter and plume evolution, temperature distribution, and acoustic responses, thereby establishing correlation mappings between process physical quantities and forming defects and enabling defect identification and prediction. A systematic comparison of monitoring techniques in terms of typical equipment, ease of use and integration difficulty, monitoring objects, and performance is presented in Table 3. At present, forming defect monitoring in LAM still faces challenges, including the limited characterization capability of single signal sources, difficulties in synchronization and registration of multi-source monitoring signals, insufficient real-time performance, interpretability, and generalization across different processing conditions of multimodal data fusion models, and multiparameter coupled modeling challenges in multi-objective feedback control, all of which require further breakthroughs.

This review summarizes the core research progress in the detection and monitoring of forming defects in LAM and clarifies the formation mechanisms and characteristics of typical porosity-type and crack-type defects. Offline detection techniques can quantitatively characterize defect type, size, and spatial distribution. Online monitoring techniques can acquire process physical quantities such as molten pool behavior, temperature distribution, and acoustic signals, establish correlation mappings between process physical quantities and forming defects, and identify and predict forming defects. These advances promote the development of research from characterization after fabrication to process prediction and online closed-loop control, highlighting the core significance of improving the forming quality of LAM-fabricated components in this field. Future research should focus on high-precision synchronous acquisition and registration of multi-source signals, multimodal data fusion modeling and interpretability analysis, multi-input multi-output multi-objective feedback control, and the construction of integrated monitoring–prediction–regulation systems, with an emphasis on achieving breakthroughs in the characterization of defect formation mechanisms under conditions involving coupled multiphysical processes, multimodal data fusion and defect prediction, and coupled modeling for multi-objective feedback control, thereby promoting the further development of LAM from defect detection and early warning toward quality closed-loop control and intelligent manufacturing applications.