Data is trending in today’s industrial environment, and many companies are advertising artificial-intelligence (AI) applications to enhance or help industrial manufacturing. There is plenty of data that goes underutilized in the U.S. industrial space. Machine learning (ML) and AI should be well-fed in our industrial space, but the problem is the cost of setting up this new technology and having the people to develop how it’s used. This is part of the constant debate about Industry 4.0 being a continuous process, as opposed to a step in developing smart manufacturing. The Clean Energy Smart Manufacturing Innovation Institute (CESMII) is attempting to give manufacturers access to AI applications with less cost. They are also training people to have a smart-manufacturing mindset.
Chartered by the Department of Energy and with around 200 member organizations, including universities, automation suppliers, factories and plants, CESMII is enabling small and large manufacturers to find ways of improving machines and integrating old and new platforms by using machine learning and artificial intelligence. Projects include load optimization in furnaces to save fuel during thermal processing. Machine diagnostics are being utilized to train on upset conditions. Digital twins are being used to simulate real-time process control and derive solutions without wasting product.
Other application projects include hybrid modeling for energy-efficient CNC grinding, data modeling for aerospace additive manufacturing and using self-powered sensors for data acquisition. For pulp and paper, CESMII is collaborating with Auburn University to use statistical pattern analysis to use machine learning predictive control for brown stock washer lines. The motivation is to bring forward thinking to industries in the United States that have ingrained processes and find it difficult to modernize with new ideas. Collaboration between industry, academia and government allow for funds and resources to be shared for better manufacturing.
Phoenix Contact hosted CESMII representatives at Automate 2024 in Chicago. Its booth also included a demonstration of Sorba’s industrial AI software, utilizing GraphQL and representational state transfer (ReST) application programming interface (API), integrated with PLCnext. The AI automatically can adapt software on the business network side and add new profiles as new hardware is added to the programmable logic controller (PLC). This alleviates time on the operational-technology (OT) engineering side.
This is a smart industrial application of machine learning/artificial intelligence, or advanced data acquisition. The downside is whether we would be able to populate all the profiles properly for the vast amount of OT hardware/software configurations. The idea is amazing because it would allow the controls engineer to populate the PLC at the lowest level and then not have to transfer tag by tag to an upper OT level or to the business side. However, the profiles would be vast, and the cybersecurity would be complex. Cybersecurity would have to be limited to one direction.
Regardless, this is a tangible application of ML and AI integration that could affect machine builders. Perhaps one day, a machine may be added to an industrial process and the network be smart enough to detect the PLC and tie in the connections to the supervisory control and data acquisition (SCADA) system automatically—integration by signal recognition and a software profile. Think of the protocol headers then. They would have to be huge.
This is the idea behind the smart-manufacturing mindset. If we think outside the current box, then perhaps we are not confined to doing things the same way we did them for the past 20 years.
