A Simulation Study on Sieving as a Powder Deposition Method in Powder Bed Fusion Processes.

Powder deposition of even and homogeneous layers is a major aspect of every powder bed fusion process. Powder sieving is commonly performed to powder batches outside of the PBF machine, prior to the part manufacturing stage. In this work, sieving is examined as a method of powder deposition rather than a method to solely filter out agglomerates and oversized particles. Initially, a DEM powder model that has been validated experimentally is implemented, and the sieving process is modelled. The sieving process is optimized in order to maximize mass flow, duration of its linear stage and total mass sieved during linearity. For this, a Taguchi design of experiments with subsequent analysis of variance is deployed, proving that the larger the initial powder loaded in the sieve, the larger the sieve stimulation necessary, both in terms of oscillating frequency and amplitude. The sieve's aperture shape is also evaluated, proving that the more sides the canonical polygon has, the less the mass flow per aperture for the same maximum passing particle size. Then, the quality of the layer produced via controlled sieving is examined via certain layer quality criteria, such as the surface roughness, layer thickness deviation, surface coverage ratio and packing density. The findings prove that controlled sieving can outperform powder deposition via a non-vibrated doctor blade recoater, both in terms of layer surface quality and duration of layer deposition, as proven by surface skewness and kurtosis evaluation.

Convolutional Neural Networks for Raw Signal Classification in CNC Turning Process Monitoring.

This study addresses the need for advanced machine learning-based process monitoring in smart manufacturing. A methodology is developed for near-real-time part quality prediction based on process-related data obtained from a CNC turning center. Instead of the manual feature extraction methods typically employed in signal processing, a novel one-dimensional convolutional architecture allows the trained model to autonomously extract pertinent features directly from the raw signals. Several signal channels are utilized, including vibrations, motor speeds, and motor torques. Three quality indicators-average roughness, peak-to-valley roughness, and diameter deviation-are monitored using a single model, resulting in a compact and efficient classifier. Training data are obtained via a small number of experiments designed to induce variability in the quality metrics by varying feed, cutting speed, and depth of cut. A sliding window technique augments the dataset and allows the model to seamlessly operate over the entire process. This is further facilitated by the model's ability to distinguish between cutting and non-cutting phases. The base model is evaluated via k-fold cross validation and achieves average F1 scores above 0.97 for all outputs. Consistent performance is exhibited by additional instances trained under various combinations of design parameters, validating the robustness of the proposed methodology.