Process faults pose a significant challenge for metal additive manufacturing (AM), resulting in structurally compromised parts or a print failure. To increase reliability and confidence in the metal AM process, we present a dynamic data-driven application system (DDDAS) detection framework that leverages a data-driven time series model to predict the nominal behavior. In-situ images of the melt pool are used to monitor the quality of the weld, and the sequence of melt pool area is used as the time series to be modeled. The deviation between the online, monitored melt pool size and the predicted melt pool size is used for defect detection; a large prediction error indicates the corresponding image deviates from the expected. This distinction can be performed automatically using a statistical threshold metric, which forms the basis of a statistical detection algorithm. Furthermore, by basing the detection criteria on the statistics of the nominal signal, the detection algorithm performs unsupervised, with no labelled anomalies necessary to distinguish process faults.
Figure 1: Metal AM production and monitoring testbed. Laser powder bed fusion (LPBF) relies on thin, deposited layers of metal powder melted by a laser. A coaxial camera that follows the laser path allows for real-time melt pool video.
Figure 2: Presented is a sample anomaly detection with localization. Through modeling the melt pool time series (top), significant errors in the prediction (bottom) lead to deviation in the model parameter statistics (center). This allows for unsupervised detection in the time series domain, which can be localized on the time series-geometry mapping (right).
Aspects of this work have been presented in the following papers: [1], [2], [3].
References
2024
An unsupervised online anomaly detection method for metal additive manufacturing processes via a statistical time-frequency domain algorithm
Alvin Chen, Fotis Kopsaftopoulos, and Sandipan Mishra
Anomalies often occur in metal additive manufacturing from processing inconsistencies and uncertainty. A robust fault detection system that uses sensor measurements such as melt pool imaging has the potential to improve part quality and save production time by anticipating print failure. Toward this goal, we develop and validate a fault detection technique using melt pool geometry-related measurements from an in situ near-infrared optical camera. This method is unsupervised and is trained on a small dataset, mitigating human error in classifying fault types, and reducing lead times for preparing training datasets. Furthermore, this method uses learned geometry-informed nominal behavior of the melt pool signal to make informed decisions on the process health. There are spatial-temporal characteristics embedded in the melt pool images, caused by the periodicity in the geometry-dependent raster pattern. These characteristics can be captured in the frequency domain using the signal spectrogram, a representation of the frequency content over time. Defects will appear in the spectrogram, disrupting the healthy spectral response. To quantify healthy spectrograms, we use principal component (PC) decomposition to extract the features of these spectrograms as a set of nominal basis vectors. Anomaly detection is then performed by calculating the error between the original and reconstructed spectrogram vector by projection of the spectrogram PCs onto the nominal basis. The reconstruction error for anomalous signals is larger than that from healthy signals, which is then used for fault detection. A one-tailed statistical test is used to determine the fault detection threshold for the reconstruction error signal. This method is tested on three raster patterns and performs better than a comparative time-series thresholding method. We demonstrate that this time-frequency algorithm can detect both temporal faults (which occur at a single time instant) and spatial faults (such as those introduced by an improper sintering), differentiating them from nominal operation.
@article{chen_unsupervised_2024,title={An unsupervised online anomaly detection method for metal additive manufacturing processes via a statistical time-frequency domain algorithm},volume={23},issn={1475-9217, 1741-3168},url={https://journals.sagepub.com/doi/10.1177/14759217231193702},doi={10.1177/14759217231193702},language={en},number={3},urldate={2024-10-20},journal={Structural Health Monitoring},author={Chen, Alvin and Kopsaftopoulos, Fotis and Mishra, Sandipan},month=may,year={2024},pages={1926--1948}}
2023
A DATA-SPARSE APPROACH TO IN-SITU FAULT DETECTION AND IDENTIFICATION FOR METAL ADDITIVE MANUFACTURING
Alvin Chen, Fotis Kopsaftopoulos, and Sandipan Mishra
In Proceedings of the 14th International Workshop on Structural Health Monitoring, Sep 2023
With increasing adoption of metal additive manufacturing (AM) in manufacturing, detecting faults in the printing process has the potential to reduce waste from failed prints and streamline the production process. To increase robustness of anomaly detection, a statistical method of detecting faults from melt pool images is presented. This method uses parametric identification of 1D compression of melt pool images to build a nominal predictive model. Nominal melt pools result in residuals that are Gaussian white noise processes, whereas anomalous melt pools will not follow this distribution. Detection is performed through statistical comparison of incoming data with a nominal reference generated on sparse data. This approach successfully applies statistical time-series methods to detect anomalous melt pools in a metal AM process.
@inproceedings{chen_data-sparse_2023,title={A {DATA}-{SPARSE} {APPROACH} {TO} {IN}-{SITU} {FAULT} {DETECTION} {AND} {IDENTIFICATION} {FOR} {METAL} {ADDITIVE} {MANUFACTURING}},isbn={9781605956930},url={https://www.dpi-proceedings.com/index.php/shm2023/article/view/36939},doi={10.12783/shm2023/36939},urldate={2024-10-20},booktitle={Proceedings of the 14th {International} {Workshop} on {Structural} {Health} {Monitoring}},publisher={Destech Publications, Inc.},author={Chen, Alvin and Kopsaftopoulos, Fotis and Mishra, Sandipan},month=sep,year={2023},}
Fault detection techniques in metal additive manufacturing (AM) have explored a variety of monitoring methods to flag anomalies as they occur during the sintering process. Although many in-situ techniques are able to adeptly detect these abnormalities, several utilize machine learning black box methods that do not easily transfer to varying print geometries. An approach that is adaptable to a multitude of geometries holds an advantage in determining anomalies for more complex cross-sections and raster patterns. To address this lack of a geometry agnosticism, we propose a method that detects faults using the frequency content of the melt pool image response through an unsupervised approach. Scan line length and scan speed extracted from known geometry can be translated to associated frequencies via a spectrogram. We examine three specific geometries to determine detection performance on each by comparing the frequency content to the nominal response. A deviation from the expected performance will signify that an anomaly has occurred. We verify this approach is feasible for fault detection and is accurate in detecting anomalies that are hard to observe in the image time series. A feasible geometry agnostic method and the current interpretability will be discussed in this paper. The results reached in this paper strongly indicate that the approach is promising, has potential for improvement, and that a geometrically independent method is sensible. With further work, a generic algorithm applicable on any geometry will be achievable.
@inproceedings{chen_unsupervised_2022,address={Columbus, Ohio, USA},title={Unsupervised {Online} {Anomaly} {Detection} of {Metal} {Additive} {Manufacturing} {Processes} via a {Statistical} {Time}-{Frequency} {Domain} {Approach}},isbn={9780791886625},url={https://asmedigitalcollection.asme.org/IMECE/proceedings/IMECE2022/86625/V001T01A007/1156737},doi={10.1115/IMECE2022-94486},urldate={2024-10-20},booktitle={Volume 1: {Acoustics}, {Vibration}, and {Phononics}},publisher={American Society of Mechanical Engineers},author={Chen, Alvin and Kopsaftopoulos, Fotis and Mishra, Sandipan},month=oct,year={2022},pages={V001T01A007}}
