| GE is well known for its 3D printing, and applies this advanced manufacturing technology to the aerospace field where the manufacturing of parts is very demanding, and the fact that strict dimensional tolerances have made 3D printing a key area of focus for GE last year brought together both ConceptLaser and Arcam This is evidenced by the takeover of the metal.While promising, metal 3D printing faces a number of challenges, if quality defects occur in the 3D printing process, it could lead to the entire part being scrapped, which would be serious waste.  For more consistent quality stability, GE applies artificial intelligence technology to 3D printing. According to Joe Vinciquerra, GE's senior chief engineer and head of additive manufacturing technology platforms at GE Global, GE is recently working to improve the performance of machines and materials through ways to increase AI, and machine learning is being played at GE's Additive Manufacturing Materials Laboratory Through Artificial Intelligence, research teams are dedicated to improving the process and quality of additive manufacturing to produce better parts and reduce quality issues. Artificial Intelligence allows any factor that can affect quality to be detected during the process, allowing the operator to ensure that appropriate adjustments are made to reduce quality defects and avoid material waste The ultimate goal is to achieve a perfect 100% quality control result Without wasted material and without failing 3D printing, this is often a distant dream, however, through machine learning, a smarter system is approaching its goal and using digital twins to create a simulated simulation model so that The process is more predictable. GE's 100% quality control results are designed to make 3D metal printing devices their own quality inspectors. GE hopes to achieve 100% visibility at every level of the part manufacturing process, Going through the machine training to identify any problems with the 3D printing process is what GE's digital twins play. GE's digital twins bridge the physical and digital worlds and learn about each unique asset over time, combining data from sensors and devices with analysis, modeling, and materials science Improve the digital model of industrial components and assets, and even the entire process and plant. It's like a human learning model that learns from experience and can become smarter if they observe that some of the constructs appear to have errors similar to what they have seen before and the device can mark it as an operator response The operator can correct and continue the machining process by stopping the construction or by dynamically adjusting it.Of course, the further case is that the 3D printing apparatus can make these modifications without operator intervention. The combination of machine learning and physical modeling not only helps companies understand the past performance of a product but also predicts the future, and GE engineers are now able to study and test complex physics through digital twins in great detail, with levels of detail passed Physical test methods are hard to come by Part build data from intelligent inspection is fed back to digital twins and the machining bias can be found by continually comparing the data with GE's proprietary 'gold standard'. As a result, metal 3D printing equipment can act as its own inspector, industrial 3D printing can also incorporate artificial intelligence and some degree of automation, further driving 3D printing equipment into the industrialized, more in-depth part manufacturing process for final production use In the case of higher-speed build rates required for production, artificial intelligence enables 3D printing devices to self-monitor and ultimately automate self-correction / compensation to improve quality control. Like many other manufacturing processes, each part built by 3D printing has its own stream of data.Figure a 3D printing device is responsible for producing tens to hundreds of parts each year, then each part represents a learning Process, even if the parts are the same.This is where AI comes in. By capturing those critical learning points and leveraging their own learning processes, you can continually improve the overall manufacturing process. So how many builds does AI need to have a 'full understanding of the additive manufacturing process and thus its effectiveness in process checks? It depends on how it is processed and on the level of what it is trying to do.' 'When you just need to predict material The goodness of control includes bulk porosity control, which may require a small amount of machine learning to be achieved. For ConceptLaser's metal additive manufacturing facilities, GE also customized a testing platform to simulate how to build parts in a ConceptLaser machine. Once the virtual test was completed, the team turned to testing on the actual machine. |
 |
|