Breaking News

Machine Learning Technique Improves Material Testing in Manufacturing – Machine Design

Harnessing the power of machine learning, scientists have developed a material testing technique that improves the accuracy of a previous technique used in manufacturing. Nano-indentation is a process used for poking a sample of a material with a sharp needle-like tip to see how the material responds by deforming. The technique has had poor accuracy in obtaining certain key mechanical properties of a material, which has prevented it from being used widely in the manufacturing industry.The new technique is a hybrid approach that combines machine learning with nano-indentation techniques. It involves using the standard nano-indentation process and feeding its experimentally measured data to a neural network machine learning system. The scientists then developed and “trained” the system to predict samples’ yield strength 20 times more accurately than existing methods.According to the researchers—a team from Nanyang Technological University, Singapore (NTU Singapore), Massachusetts Institute of Technology (MIT) and Brown University—the new analytical technique is able to reduce the need for time-consuming and costly computer simulations in the manufacture of parts. The technique tests whether parts are safe to use in real-life conditions and can be employed in structural applications such as airplanes and automobiles, as well as those made from digital manufacturing techniques such as 3D printing. “By incorporating the latest advances in machine learning with nano-indentation, we have shown that it is possible to improve the precision of the estimates of material properties by as much as 20 times,” reported NTU Distinguished University Professor Subra Suresh, the senior corresponding author of the paper. In addition to improving the precision of the estimates of material properties by as much as 20 times, Suresh noted, the research also validates the system’s predictive capability and accuracy enhancement on conventionally manufactured aluminum alloys and 3D-printed titanium alloys. “This points to our method’s potential for digital manufacturing applications in Industry 4.0, especially in areas such as 3D printing,” he said.The findings will be published in the Proceedings of the National Academy of Sciences.Scientists at Nanyang Technological University, Singapore (NTU Singapore), Massachusetts Institute of Technology (MIT) and Brown University have developed new approaches that significantly improve the accuracy of an important material testing technique by harnessing the power of machine learning.MITA Hybrid ApproachThe process starts with pressing a hard tip—typically made of a material-like diamond—into the sample material at a controlled rate with precisely calibrated force, while constantly measuring the penetration depth of the tip into the material being deformed.To improve accuracy, the NTU, MIT and Brown team developed an advanced neural network—a computing system modelled loosely on the human brain—and “trained” it with a combination of real experimental data and computer-generated data. The researchers’ “multi-fidelity” approach incorporated real experimental data as well as physics-based and computationally simulated “synthetic” data (from both two-dimensional and three-dimensional computer simulations) with deep learning algorithms.Previous attempts at using machine learning to analyze material properties mostly involved the use of synthetic data generated by the computer under unrealistically perfect conditions—for instance, where the shape of the indenter tip is perfectly sharp, and the motion of the indenter is perfectly smooth. The measurements predicted by machine learning were inaccurate as a result.However, training the neural network initially with synthetic data, then incorporating a relatively small number of real experimental data points, was shown to improve the accuracy of the results. The researchers also report that the training with synthetic data can be done ahead of time, with a small number of real experimental results to be added for calibration when it comes to evaluating the properties of actual materials.“The use of real experimental data points helps to compensate for the ideal world that is assumed in the synthetic data,” said Suresh. “By using a good mix of data points from the idealized and real world, the end result is drastically reduced error.” 2020 is set to be the year in which enterprise resource planning (ERP) drives fresh advancements in technology and changes the way whole organizations interact with data—from engineering to HR and finance departments. Machine learning (ML), artificial intelligence (AI), cloud technology and open APIs (application programming interfaces) are all to be taken advantage of in the coming decade. But beyond bringing these emerging technologies to a single platform, ERP developments will not be limited to technology.Following are five trends expected to shape the future of ERP:1. Simplicity Before Anything Else Simplicity—it trumps everything, even in our personal lives. The lessons learned from the business-to-consumer world can be applied to industrial production, too, in that companies that can simplify their user experience will triumph over competitors. Similarly, manufacturers shouldn’t have to fight with disparate modules in their ERP applications that need integration, modification and extensions. They need seamless processes and reporting, and as ERP evolves, it must also keep these functions simple.2. Complete Cloud Adoption Cloud has been a trend for so long that it is now a basic fact of life. Yet cloud is still not universally adopted, its vast benefits notwithstanding. When organizations take their first steps toward the technology, they often start with a hybrid model to “test the waters,” then later go for a full-cloud approach so they can reap the benefits. This gives them the opportunity to scale at their own pace and still migrate to the cloud without major disruption. Expect to see more industrial organizations realize this as the 2020s progress, and for the full-cloud trend to continue. 3. AI and Machine Learning Targeting Productivity Improvement AI and ML is a pairing of technologies whose full potential has yet to be realized. In a November 2019 IFS study 40% of manufacturers said they were planning to implement AI for inventory planning and logistics, and 36% for production scheduling and customer relationship management. Overall, 60% said they were targeting productivity improvements.This reveals the scale of appetite for AI and ML because they can uncover fresh, innovative insights that boost the bottom line. Take the example of just-in-time manufacturing. This is designed to deliver components at the last possible moment to alleviate storage and associated costs. If an AI solution is asked to look at just-in-time scheduling it can come up with a whole new approach to the production process, creating new options that can make a real difference and generate significant productivity gain. 4. Open APIs Offering Flexibility Open APIs are another big trend that is transformational for ERP. Based on open standards that are readily available to developers, a wide range of client-side functions will be quickly integrated into systems. They allow development to happen at pace because there are no hidden complexities, and they permit inbound and outbound connections to the digital core with the greatest possible flexibility and speed. In addition, open APIs make it easy to connect to applications, platforms, services and databases that are external to the core. APIs are the foundation of the way IFS’s front-end works, and is expected to gain even more traction across the industry.5. A Service-Based ApproachThe traditional manufacturing practice of selling goods by intermediaries is changing as manufacturers are moving closer to their customers. There are two ways this is happening—one is via subscription-based sales, the other is through integrating services into products. As this trend continues, expect servitization to become a more widely recognized and adopted business approach.In fact, while in 2018, 62% of manufacturers reported profitable aftermarket service operations through planned maintenance or service contracts, only 4% offered products entirely as a service. There are many unexplored opportunities for the evolution of the business-to-business-to-customer model. For example, a filtration systems customer has transformed itself from selling air cleaners to selling clean air by helping customers set and meet their individual air quality goals. Automation is key, as sensors monitor air quality, call out technicians, order parts and monitor the equipment to action a condition-based maintenance program. A More Customer-Centric and User-Friendly Decade for ERPThe decade ahead for ERP looks positive. There are early signs of this: AI and ML are already introducing new efficiencies to operations, open APIs are allowing for more seamless integration of applications and data, and manufacturers are moving closer to customers by directly delivering the products and services they really want—all supported by the cloud.The primary role of ERP will be to simplify using the means at our disposal. To the end, simplicity will cut through all of these trends and will be the watchword for ERP implementations in the coming decade. Darren Roos is CEO of IFS.
Source: https://www.machinedesign.com/automation-iiot/article/21126695/machine-learning-technique-improves-material-testing-in-manufacturing