CREMONA, ITALY — While Ocrim values tradition as an integral part of the company’s success, it also recognizes artificial intelligence (AI) likely will have a role to play in aiding the quality and efficiency of the milling industry, even if the emerging technology may not reinvent everything as some claim.

Over its two-day “Wheat Flour And…” at its headquarters in Cremona, Italy, Ocrim examined what AI is, the rapid expansion of the technology, early applications in the milling industry and what the future might hold.

And while much is uncertain when it comes to AI, one message reverberated throughout the event: AI will not replace the need for humans.

“AI will never replace all this,” said Alberto Antolini, chief executive officer of Ocrim, who compared the miller to the director of an orchestra, ensuring all machines are operating efficiently and in sync. “We have to really pay attention to what we are doing. AI can help and assist but it cannot give a different color to our work.”

Sergio Antolini, president of Ocrim, shared a similar message as he opened the event, noting that humans are endowed with the neocortex, allowing for the most sophisticated movements, perceptions and cognitive abilities.

“If artificial intelligence must be considered a tool, it must remain under human management, avoiding the error of being overwhelmed by automatisms, algorithms and data structuring,” he said.

The Open Day event brought together 270 flour millers, researchers and others from across the world for learning opportunities, factory tours, cultural experiences and networking opportunities.

 Artificial intelligence

Davide Zacchetti, IT manager at Ocrim, introduced AI to the audience, outlining its basic principles and potential applications. He noted how quickly new AI programs are being developed, with changes happening in a matter of months and weeks instead of years and decades.

AI first started stirring excitement in 1950, but it wasn’t until 1980 that machine learning started to flourish. Starting in 2010, deep learning breakthroughs started driving AI, and from 2021 to 2024 there were huge developments in AI’s abilities.

“We were talking about years, not decades, and now it’s so fast,” Zacchetti said. “Week after week there are changes. AI models are becoming bigger and more complex.”

AI is transforming industry by improving performance, quality and competitiveness, said Luca Lambri, smart machines department manager at Ocrim. The smart machines department was created just a few months ago following requests from customers. Automation today is more than a PLC that performs simple logical functions, he said.

“We can have advanced tools that send data to then be processed by AI systems so as to be able to intervene promptly in management of the plant,” Lambri said.

"We have to really pay attention to what we are doing. AI can help and assist but it cannot give a different color to our work."

AI includes four subsets: learning, vision, robotics and the Internet of Things (IoT). Within learning, there is machine learning, which deals with creating systems that learn and improve performance based on acquired data.

Deep Learning, a subset of machine learning, uses mathematical models but is inspired by and emulates the functioning of the human brain, Lambri said.

“This technology allows machines to learn and improve autonomously by processing large volumes of data,” he said. “The goal of Machine Learning and Deep Learning is to make a machine that is able to learn from its own experience to then complete new tasks it has never faced in a very accurate manner.”

Artificial vision represents a fundamental point of AI, Lambri said, because thanks to the use of image recognition algorithms there is a faster response in detecting defects, and this leads to greater product selection, contributing to a high-quality result. An example is the optical sorter, which is trained to reject product through image acquisition and use of deep learning. 

Robots are one of the most concrete aspects of AI, because robots are used to automate physical tasks. 

“The use of robots allows us to have very accurate repeatability in operations, something that humans would not be able to maintain,” Lambri said. “AI algorithms are essential for robots to perform complex tasks. For example, a robot might use a path-finding algorithm to navigate a warehouse or to optimize a trajectory.”

IoT, a network of connected objects and devices equipped with sensors that allow them to transmit and receive data, is increasingly used in the industrial sector. For example, machine data from roller mills could be sent to the cloud and then visible via customized apps or an analytical format. 

“This would allow authorized personnel to interact with the machine even when the system is unattended,” Lambri said. 

Another field in which artificial intelligence is heavily used is in customer service or predictive maintenance. A virtual assistant can help operators manage problems, guide them in maintenance or find the part to order, he said.

Paolo Molinari, electrical and automation department manager, delved into specific applications of AI in milling, while emphasizing AI can improve plant performance, but not alone. 

Data is the fuel that powers AI, so data quality is key, he said. Inaccurate or incomplete data can lead to incorrect predictions. Data also may contain biases, making it important to recognize and mitigate those biases. Managing large volumes of data requires adequate infrastructure for storage and processing.

Ocrim offers its customers Mill@management to collect and store data, which eventually can be shared to the Ocrim Cloud that is being designed, Molinari said. Data types include stocks, recipes, cycles, maintenance, yield reports and power consumption. 

The company is in the process of formalizing some AI applications for the milling process, including the Advanced Production Planner to optimize the planning and scheduling of production activities. Using advanced algorithms, data analytics and software solutions it will generate a task list organized by time and day that minimizes downtime and maximizes machine utilization. It accounts for constraints such as machine maintenance schedules and labor availability and allows for adjustments in response to unforeseen issues. 

“It will not replace the operator, but it will support them,” Molinari said. 

AI also can be used to create a Predictive Maintenance Tool, allowing operators to intervene before an equipment failure occurs. The condition of equipment would be continuously monitored to identify any deviations from normal operating conditions while trends and patterns in the data are identified to predict future failures. 

Algorithms will be trained on historical data to predict when a component might fail and techniques such as regression analysis would forecast equipment lifespan and maintenance needs. 

“By predicting failures before they occur, maintenance can be scheduled during non-peak times, minimizing disruption to operations,” Molinari said. “Preventing unexpected failures can save costs associated with emergency repairs and production losses.”

Finally, Ocrim is developing an Automation Virtual Assistant, an advanced system designed to assist operators in monitoring, managing and controlling industrial plants. It could generate alerts for abnormal or dangerous operating conditions, offer recommendations based on data and analytics to help operators in critical decisions and simulate “what-if” scenarios to evaluate the impact of different decisions.

Two academic speakers shared how AI can be used in agriculture to study and improve wheat cultivation. Ilva Licaj, researcher at the Department of Science and Technology of the University of Sannio in Benevento, explained how she uses AI to study the roots of traditional and modern wheat to highlight the potential of ancient varieties in handling environmental stresses.

Domenico Felice, researcher at the Management Engineering Department of the Polytechnic University of Milan, discussed the ability to use AI to detect drought stress in wheat roots.


Ocrim_Wheat Flour And_Alberto Antolini_left_Sergio Antolini_right_©SOSLAND PUBLISHING CO._e.jpg

Alberto Antolini, left, chief executive officer of Ocrim, shares a moment on stage with brother Sergio Antolini, right, president of Ocrim.

|
Credit: ©SOSLAND PUBLISHING CO.

 Wheat cleaning systems

The second day focused on “The World Granary and it Changes,” with speakers examining global challenges for wheat production, the impacts of international unrest on grain trade and alternative cereal options. Marco Galli, technological department director of Ocrim, concluded the presentations with an overview of Ocrim’s wheat cleaning systems.

The milling industry is faced with a series of challenges, including market globalization, market fluctuation, crop quality, availability and food safety, Galli said. Consumers are more sensitive to food safety concerns, regulations are tighter, contaminant limits are reduced and there are advanced tests for monitoring, controlling and preventing contaminants in food.

This leads to higher attention to chemical and biological contaminants and higher attention to the quality of investments, Galli said. Mycotoxin contamination presents a particular challenge because not all contaminated wheat has the same characteristics or requires the same separation process, he said.

“There is no single machine on the market today that can identify directly the mycotoxin-contaminated wheat,” Galli said. “The separation must be done indirectly. This means that the separation must take place through a process, not through a single machine.”

A high-efficiency cleaning system maximizes detection and removal of mycotoxin contaminated wheat and provides flexibility in processed wheat. The system creates a division between the heavy and light fractions to concentrate potential contaminated kernels.

Machines in the system include color sorters, infestation destroyers, high efficiency scourers and tempering water purification. The color sorter allows for the separation of contaminated kernels by size, shape and color shade, which can be set for the whole kernel or just for the defect. The machine can be easily integrated with new AI concepts, Galli said.

Deep learning technology allows for processing of large amounts of data images to make predictions with high precision. The software uses the database of images as a reference to detect impurities and defects, looking at color and shadows, size and shape of the grain and the defect, surface and density.

Infestation destroyers are an important step because of the negative impact of insect-damaged kernels on the rheological flour qualities, Galli said. It could lead to cross contamination and infestation, and the flour produced from infested wheat has a shorter shelf life.

The destroyer is positioned on the light fraction and breaks the infected kernels against the static corrugated surface. It is combined with an open channel aspirator to ensure removal of the various fractions and avoid cross contamination.

The high efficiency scourer is designed to reduce the kernel’s surface contamination and features a screen with a high abrasion surface. It reduces ash, heavy metal and pesticides, Galli said.

Tempering is essential for grinding, but it is also one of the most critical phases regarding microbiological contamination. Water and temperature are ideal conditions for the development of contaminants, he said. Using purified water is one solution, but this is not always sufficient and may not apply to all countries.

“This is why further purification phases are applied and the most common method is the use of chlorine,” Galli said. “On the other hand, the use of chlorine is increasingly restricted.”

Ocrim has decided to apply UV technology, using a defined wavelength to destroy the molecular bonds of the DNA of microorganisms. It is easy to install and maintain, it does not subject the plant to corrosion risks, has no flow rate limits and does not use chemicals, Galli said.