Engineering researchers from the Argonne National Laboratory of the Department of Energy (DoE) will use social media artificial intelligence technology to better optimize3D printingThe geometry of the part. Argonne’s chief mechanical engineer, Mark Messner (Mark Messner), was one of the people who initially developed this new method as early as 2019. He claims that it is a better way of predicting the material’s possible performance under extreme temperatures and pressures. Way. Although the current simulation-based prediction methods have been successful, they often require supercomputer-level processing power (and a lot of patience) to accurately predict the possible behavior of geometry under certain conditions.
“Messner said: “You usually have to run a lot of physics-based simulations to solve this problem. This is especially true if researchers already have a set of specific properties, such as stiffness, density, and strength, and want to determine the optimal component structure they need to produce these properties. According to reports, as an alternative method, Argonne’s method is more than 2,000 times faster than modern part performance simulations and can run on ordinary laptops with consumer-grade GPUs.
3D printingParts” style=”cursor:pointer;border:1px solid #e5e5e5;” width=”640″ height=”779″ />
Messner’s artificial intelligence is repeatedly optimizing the geometry of the parts. Image source: Mark Messner.
How does social media work?
Messner’s work can be traced back to when he was a postdoctoral researcher at Lawrence Livermore National Laboratory, when he and his team were trying to3D printingComplex micron-scale structure. According to reports, the team’s progress is slow, so they set their sights on artificial intelligence to see if they can speed up their research.
At that time, the emerging social media giants in Silicon Valley had begun to use Convolutional Neural Networks, an artificial intelligence that could find patterns in large data sets for facial and object recognition in images. Messner thinks he can apply this concept to the three-dimensional field. “He explained: “My idea is that the structure of a material is no different from a three-dimensional image. “It makes sense that the 3D version of this neural network will recognize the attributes of the structure well–just like a neural network learned that an image is a cat or something.” To see if his idea is feasible, Messner designed a deterministic three-dimensional geometry and used traditional physics-based simulations to create a set of 2 million data points. Each data point relates its geometry to the “ideal” values of density and stiffness. He then fed these data points into a neural network and trained it to find the required attributes.
Finally, Messner used genetic algorithms-an iterative, optimization-based category of artificial intelligence-together with a trained neural network to determine the structure that would produce the characteristics he was looking for. What is impressive is that his artificial intelligence method found the correct structure, 2760 times faster than traditional physical simulation.
The topology of the convolutional neural network. Image source: Mark Messner.
artificial intelligence,3D printingAnd the nuclear sector
One of the most promising applications of artificial intelligence methods is in3D printingfield. Since this method tends to propose extremely complex geometric shapes, the traditional manufacturing process will be difficult to actually produce the structure suggested by the model.3D printingThe additive nature of it makes it possible to manufacture these optimized structures, no matter how complex the geometry is, enabling scientists to achieve the characteristics they seek.
Messner believes that “the future of mechanical engineering” is likely to be a combination of artificial intelligence and additive manufacturing. “You can give the structure determined by the neural network to the owner3D printingThe people of the machine, they will print it out according to the performance you want. We have not fully reached this goal, but this is our hope. “A more direct application of this technology is in the nuclear industry for material design. In fact, Messner’s team is currently working with nuclear power startup Kairos Power to design a molten salt nuclear reactor core using artificial intelligence. Argonne The model will ultimately help the Kairos team predict how stainless steel 316H will handle the inherent high temperatures and pressures of nuclear reactor cores within decades.
“This is a small part of what we are doing for Kairos Power, but it is crucial,” concludes Rui Hu, a nuclear engineer at Argonne. “Kairos Power hopes to have very accurate models to illustrate the behavior of reactor components in its reactor to support its application for a license to the Nuclear Regulatory Commission. We look forward to providing these models.”
Artificial intelligence and machine learning have undoubtedly entered3D printingIt has applications from material design to defect detection. Earlier this month, researchers from the Computer Graphics Group (CGG) of Charles University developed an ML-based technology that can help unlock higher fidelity c
Artificial intelligence and machine learning have undoubtedly entered3D printingIt has applications from material design to defect detection.Earlier this month, researchers from the Computer Graphics Group (CGG) of Charles University developed an ML-based technology that can help unlock higher-fidelity color3D printing.By simulation3D printingDuring the process, the team was able to train an algorithm to find the best construction parameters to limit color leakage and improve part accuracy.
Elsewhere, at Argonne and Texas A&M University, scientists have previously developed a novel ML method to detect3D printingDefects of components. Using real-time temperature data and ML algorithms, scientists can establish a correlation between thermal history and the formation of subsurface defects.
(Editor in charge: admin)
0 Comments for “Scientists will use social media AI technology to optimize 3D printed parts”