Selective Laser Melting (SLM), as a common metal additive manufacturing (AM) technology, achieves high-precision complex part formation by layer-by-layer melting of metal powder using a laser. However, the dynamic behavior of the melt pool during the SLM process is influenced by the heat source model, which is crucial for suppressing porosity defects and optimizing process parameters, directly determining the reliability of numerical simulations. To address the issue of traditional surface heat source models overestimating the melt pool width and volume heat source models underestimating the melt pool depth, this study constructs a three-dimensional transient heat conduction finite element model based on ANSYS Parametric Design Language (APDL) to simulate the evolution of the temperature field and melt pool geometry under different laser parameters. First, the temperature fields and melt pool morphology and dimensions of four heat source models—Gaussian surface heat source, volumetric heat source models (rotating Gaussian volumetric heat source, double ellipsoid heat source), and a combined heat source model—were investigated. Subsequently, a dynamic heat source model was proposed, combining a Gaussian surface heat source with a rotating volumetric heat source. By dynamically allocating the laser energy absorption ratio between the powder surface layer and the substrate depth, the influence of this heat source model on melt pool size was explored and compared with other heat source models. The results show that under the dynamic heat source, the melt pool width and depth are 128.6 μm and 63.13 μm, respectively. The melt pool width is significantly larger compared to other heat source models, and the melt pool depth is about 17% greater than that of the combined heat source model. At the same time, the predicted melt pool width and depth under this heat source model have relative errors of 1.0% and 5.5% compared to the experimental measurements, indicating that this heat source model has high accuracy in predicting the melt pool’s lateral dimensions and can effectively reflect the actual melt pool morphology during processing.

Digital twin-based real-time quality control in gear machining lacks closed-loop feedback, classification without visual recognition, and low-latency I/O synchronization. This paper proposes a digital twin-driven intelligent quality control framework built upon a four-layer cyber–physical architecture comprising three tightly coupled subsystems: a parent-object assignment mechanism that enables precise workpiece type tracking and routing through hierarchical container queries, eliminating the need for computationally expensive visual type classification while maintaining near-perfect tracking accuracy; a rhythm-adaptive multi-robot behavioral control scheme that adjusts production cadence via a global rhythm coefficient without altering spatial trajectories; and a lightweight in-memory key-value store-based I/O synchronization mechanism that achieves architecturally bounded signal update latency (communication over a dedicated localhost TCP/IP path) well below the critical process cycle. Validation on a physical gear hub production line with six operations and three robots demonstrates a single-piece inspection cycle of 11.5 s, collision-free operation, and average error rates of 0.4% for inline quality inspection. The framework provides a low-cost, low-latency, and formally grounded solution for real-time quality control in gear machining workshops.