In spite of being a relatively new solution in technology, artificial intelligence has found its place in various businesses. From eCommerce, retail, all the way to customer service and advertising, AI is helping in developing valuable solutions for many companies.
Driven by knowledge-based systems which are feeding themselves on input data, it also helps manufacturing businesses develop more sustainable working models, while at the same time increasing productivity and quality of the delivered products.
Today, we're going to take a look at the ways of implementing AI in the steel industry. Complex by nature, each of its processes generates an astonishing amount of data which provides useful insights if properly managed. The data is primarily gathered from many field sensors of the plant automation through the supply chain. Altogether, it contains information that feeds all the activities of production.
The Importance of AI in the Steel Industry
As an early adopter of AI technologies, the steel industry leads all heavy industries in the improvement of sustainability and competitiveness on the market. It resembles a perfect ground for an approach based on data exploration and exploitation.
Steelmaking industries deal with complex, multi-physics processes, where a lot of variables and correlations are not completely understood. Besides that, environmental conditions play an important role in the process and are keen to change over time.
There are many obstacles steel industries face in the manufacturing process. First of all, decision-making processes are often made by human operators. An operator needs to go through a thorough educational process before entering the manufacturing process itself. And not only that, they usually need to commute far away from the nodal points, which aggravates the process of team selection.
While using AI in steel industries, many processes can be done remotely through personalized processes powered by AI and machine learning solutions.
In the context of processes, data plays an important part, and it often changes in regular intervals. With cases such as a global pandemic, manufacturing capacities can drop down, which can lead to a decrease in the data generated over time.
With sensor systems that can identify deviations in the data trained on historical events, AI can help in creating arrangements and modifying the process in a way where errors won’t be repeated.
Reasons for the Implementation of AI in the Steel Industry
Unlike humans, machines cannot only learn the execution of common activities, but they can also think in patterns. Using high-dimensional data human brain cannot conceive, the usefulness of AI in the steel industry is more than apparent.
Not only, AI is capable of understanding more information than a human, but it also can perform extremely dangerous actions commonly carried by operators. By removing operators from risky situations, they are given the opportunity to work on tasks with more added value.
Artificial intelligence and machine learning have been disseminated in various fields of engineering. Some literature examples even show data-driven ML techniques for predicting secondary deformation mechanisms in steel manufacturing.
Some other examples include data-driven models for strip temperature prediction in line heating processes, diagnosing cooling temperature deviation defects, but also steel surface defect classification.
DETECT YOUR ANOMALIES IN SECONDS
Check out Wizata's Anomaly Detection Wizard
Polishing the Imperfections – A Case Study
In the business world, case studies represent a beneficial source of information based on customer experience. While we at Wizata provide the tools for improvement, it is on our clients to make their groundbreaking strategies possible.
Some of our primary goals include maintaining high-quality standards when working with our clients, and thanks to AI and ML, such standards became easy to achieve.
One of our clients, a South American steel producer, had continuous quality issues in their steel coil production process. All of the traditional processes they’ve implemented have not resulted in the desired outcome.
The defect consisted of permanent deformations of the coil surface as multiple small scratches in the rolling direction of the coil. In order to prevent these deformations, it was important to identify the root cause that led to these defects.
While increasingly happening at an uncontrolled pace, the root of these defects was unknown. Unfortunately, this problem resulted in a significant decrease in sales and additional costs of reprocessing and scrap handling.
To solve this problem, the quality and production data were collected and integrated into the Wizata platform from different sources. Wizata scientists, with the help of local operators, identified 3 main data sources inside the production data. These sources were telemetry data, chemical composition, and quality inspection.
Focusing on those 3 data sets, hypotheses were made based on the pattern recognition and anomaly detection algorithms.
The main issue was found in the first 7 seconds of the process. The combination of the hot roller pinch position, bending force, and coiler speed resulted in creating these tiny scratches.
As the pinch roller was not fully retreated when it was supposed to, production flaws were made. But with the help of machine learning models, the reprocessing rate has been reduced, and the defect rate has been decreased by 25%.
The Future of Manufacturing is Happening Now
The key concept of AI and ML lies in its ability to extract knowledge from data. While older computers weren’t capable of performing these actions, machines now can learn and perform much-desired tasks in various industries.
Patterns in data are too complex to be detected by a human easily. With the use of AI in the steel industry, operational costs are reduced, product quality is increased, and revenue is constantly growing.
Digital technologies can also improve existing models that estimate and predict events by extracting the already captured information and patterns.
As frameworks of AI and Ml are being deeply integrated into the automation systems, industries are now able to implement them in their manufacturing processes, and so AI in the steel industry takes an important place.
However, there are very specific requirements the steelmaking industry has in terms of automation and information technology. Thus, it poses challenges of implementing AI and limits its spreading.
Industrial automation systems have a vast number of complex components in production. Therefore, knowledge extraction from data signifies a core part in transforming industrial plants into smart factories.
Conclusion
An idea of future industry solutions relies on machinery's ability to impact processes with self-optimization and independent decision-making strategies. This will result in the improvement of safety, transparency, both effectiveness, and efficiency, but also in the creation of self-organized and autonomous management.
Having that in mind, the implementation of AI in the steel industry will create a space for much greater production processes that can only evolve into the creation of the best products, while maintaining safety not only for the workers but for the product itself. Additionally, it is cost-effective and has a great impact on creating a competitive market.
