When metal 3D printing is widely considered to be a reliable industrial manufacturing method, this day is slowly coming, but we still need to solve some problems before it really comes. A great deal of research has explored the root cause of metal 3D printing defects, which can lead to defects such as splashes and microcracks in the final 3D printed parts - this is unacceptable when dealing with high-risk applications such as aerospace components.
But two researchers at the Carnegie Mellon University School of Engineering (CMU) have come up with how to combine 3D printing with machine learning for real-time process monitoring, which detects abnormalities in parts during 3D printing. .
CMU Mechanical Engineering (MechE) alumnus Luke Scime and NextManufacturing Center director Jack Beuth have created a machine learning algorithm that monitors the process of laser powder bed fusion technology, which is prone to errors due to uneven powder layer dispersion.
Other researchers are using methods such as acoustics, spectroscopy, and temperature monitoring to understand what is going on inside the constructed structure. However, although there are some limited types of monitoring on the market, they usually do not have the ability to automate analysis and only provide data that the machine operator must interpret. But the work of Scime and Beuth has a different direction: computer vision algorithms.
Scime said: "One of the biggest obstacles to making a good looking component and putting it on the plane is to make sure that the parts you make are flawless. Computer vision is a term that uses data analysis techniques to understand what's going on in the image."
Scime's innovative algorithm takes images of the powder bed and extracts features, then groups the features and compares them in different levels of analysis until the fingerprint of the image is created. The machine has learned how to identify different defects because the researchers provided hundreds of pre-marked training images. Now it can compare the fingerprint of the new image it receives and its known fingerprint to isolate various anomalies.
It is reported that Scime and Beuth published a paper entitled "Anomaly detection and classification in a laser powder bed additive manufacturing process using a trained computer vision algorithm" in the magazine "Additive Manufacturing", which demonstrates how the algorithm can detect powders. The millimeter level ç‘•ç–µ. The algorithm can determine what the defect is and where it occurs, which can help improve process stability (printing power).
The paper abstract reads: "This work proposes a method for in situ monitoring and analysis of powder bed images that may become an integral part of real-time control systems in LPBF machines. Specifically, computer vision algorithms are used to automatically detect and classify Anomalies that occur during the powder diffusion phase of the process. Anomaly detection and classification is performed using an unsupervised machine learning algorithm, operating on a training database of appropriately sized image blocks. Demonstration through several case studies, performance of the final algorithm Evaluate and use it as a stand-alone package."
This work is to make metal 3D printing a reliable and safe method of industrial production.
Scime said: "The Holy Grail is deploying this environment in a real-time environment, you will automatically analyze the data, do something, and then move on. The real question is, can we detect it, understand that this is a problem, and then design our Weigh the processing parameters to do something different than what we did to reduce the amount of warpage?"
Scime explained, “Automatic correction may end up working in several different ways, the most basic of which is that once an abnormal condition is detected, the 3D printer sends an alert to the operator to resolve the problem as soon as possible. Then you will continue to teach the 3D printer to Identify critical defects and automate simple fixes."
However, the highest achievement of automated self-correction is to fight high. This anomaly that causes most damage occurs when part of the build begins to distort or curl out of the powder. Although there may be some time before this level of automation is reached, the CMU machine learning algorithm has been able to accurately identify some anomalies and is ready to be applied in the real world. However, Scime hopes to study how to add additional sensor data to its analysis and improve its accuracy.
The size of the fabric Sleeper is small which takes less space and easy to move. You can have a fabric sleeper at your office which is more comfortable for lunch break. And you can use it as a sofa at office time. The fabric sleeper sofa is also easy to intall. There are different types of fabric sleeper, you can choose the fabric sleeper suitable for you.
Fabric Sleeper,Chaise Lounge Sofa Chair,Fabric Chaise Lounge Sofa,Fabric Chaise Lounge Chair
Kaifeng Lanwei Smart Home Co., Ltd , https://www.leather-recliner.com