Intel's AI helps speed paper-making process

byteLAKE and Intel are collaborating on a new artificial intelligence (AI) solution to automate wet and dry-line management during the paper-making process. The solution is designed to help paper mills avoid costly disruptions during production.

byteLAKE’s Wet Line Detector for Paper Mills is commercially available and successfully implemented in mills across Europe. It is included in a collection of AI solutions for Industry 4.0 called byteLAKE’s Cognitive Services, which offers AI-assisted visual inspection and big data/Internet of Things sensors data analytics.

The technology
In Europe, byteLAKE optimized Cognitive Services on Intel Xeon Scalable processors with integrated Intel Deep Learning Boost and the Intel Distribution of OpenVINO toolkit, enabling the solution to be used even more efficiently. This resulted in 10 times faster performance without any hardware upgrades or drops in model accuracy.

“At byteLAKE, we have used many Intel technologies for several years and are impressed by their performance and scalability,” said Marcin Rojek, co-founder of byteLAKE. “The use of our technology in the paper industry helps with avoiding many critical situations during the process. The combination of byteLAKE’s Cognitive Services with Intel Xeon Scalable processors with built-in AI acceleration allows us to outperform GPUs and greatly simplify the future implementation of this solution in other Industry 4.0 scenarios.”

Why it matters
Paper production is a multi-phase process during which a natural phenomenon called “wet line” (sometimes “dry line”) is observed. The wet line must be carefully monitored to avoid losses and expensive breaks in production. It’s a highly tedious process where an operator must visually inspect the machine and monitor the wet line and adjust settings to avoid any issues.

“AI already plays a very important role in our daily lives. Nevertheless, many companies still perceive the application of AI in their business as an addition, not its foundation,” said Krzysztof Jonak, EMEA Territory Sales director at Intel. “The application of the Intel Distribution of OpenVINO toolkit in byteLAKE’s Cognitive Services shows that AI works efficiently as an actual tool for optimizing company operations. Moreover, such a combination reduces the barrier of necessary upgrades to IT infrastructure in the company to an additional computer forming the basis for the whole system.”

How it works
Wet Line Detector for Paper Mills is a dedicated AI model designed and trained specifically for the role of wet-line detection. It can work in real time and perform ongoing analysis of the frames from closed-circuit television. AI algorithms examine the surface where the weave is formed and detect the wet line.

Additionally, the algorithms measure and estimate the location of the wet line and its width. This information is then presented to the paper-machine operator, who can react appropriately and adjust the settings as needed.

“This is a breakthrough in looking at AI and its implementation in companies that are able to see the potential in joining the ‘Industry 4.0’ family of businesses,” added Jonak.