May 2022 Volume 4

AUTOMATION

forging industry invests in the technologies that make it smarter, the more competitive it will remain. Integral to the Smart Factory is the Internet of Things (IoT), which integrates devices that can communicate with each other in real time and that also generates heaps of data that can be mined in ways not yet even understood. Data generated through advanced sensors can be shared across the entire manufacturing platform to provide the tools to minimize downtime through, for example, predictive maintenance programs. A niche to the IoT, is the Industrial Internet of Things, or IIoT, into which the forging community would fit. One of the smarter things to hit forge shops during the last several decades has been the advent of software platforms that can model process technologies and predict part behavior in the manufacturing process through the application of Finite Element Analysis (FEA) and other predictive analytics. To gain some perspective on this topic we spoke with Jim Miller, Director of Sales and Marketing for Scientific Forming

Machine learning models train on existing data sets to find relationships between process inputs and system response. Courtesy of Scientific Forming Technologies Corp.

Technologies Corp., Columbus, Ohio. His company’s DEFORM software platform falls squarely into the “revolution” that is happening in smart manufacturing among the forging community. “Industry 4.0 takes the computers and automation introduced during Industry 3.0 and, through data analytics and machine learning, are key to the manufacturing revolution. Data analytics represent the process and tools for analyzing large data sets and modeling their data relationships. These can incorporate machine learning

algorithms to make production machinery intelligent enough to build predictive relationships, and autonomous enough to self-correct themselves as needed,” says Miller. Existing data sets are used to “train” machine learning models to identify otherwise hidden patterns and trends in system response. Once a trained model is established, new input values can be fed into it and the associated outputs become predictive. The new data analytics tools are broadly applicable to data sets from many sources such as CMMs, load cells, strain gauges, thermocouples and other instruments. Machine learning models have accounted for simulation and production results in a single model, with the model improving as the data set grows. This offers attractive opportunities to link machine learning into real-time process automation equipment, where the equipment control routines become more intelligent as production data is collected. Miller indicates that there is great interest among forgers in these technologies. Early adopters are already applying them, and that this is where the industry was on forging simulations a few decades ago. The revolution, in other words, is undergoing an evolutionary process. “A main objective is to establish a new framework of tools with basic simulation, allowing for more analysis of greater scope relating to process as well as part,” he concludes.

Researchers and manufacturers are applying data analytics response models in “ intelligent” process control equipment. Courtesy of Scientific Forming Technologies Corp.

FIA MAGAZINE | MAY 2022 27