Scientists in the US have examined three smart drying methods that use AI-enabled optical sensing technologies, by developing a convective heat oven to test them on apple slices. These techniques can allow F&B manufacturers to monitor drying processes precisely and gain valuable insights for industrial applications.

The team at the University of Illinois Urbana-Champaign, studied drying techniques such as RGB imaging with computer vision, near-infrared (NIR) spectroscopy and near-infrared hyperspectral imaging (NIR-HSI). Manufacturers can use these methods independently or in combination, depending on their needs and cost-effectiveness.

Food Ingredients First speaks with Marcus Vinicius da Silva Ferreira, the lead author of the study and a postdoctoral researcher at the Department of Agricultural and Biological Engineering at the University of Illinois Urbana-Champaign. He explores how these precision food drying methods can enhance drying accuracy and the role of AI in ensuring product consistency and quality.

“We are working on research using a smart dryer device we built in our lab that uses NIR to determine the end point of drying, accessing the moisture content. We have built a convective drying oven that we use at some drying conditions to precisely determine the moisture content in the apple slice so we can predict the drying time (endpoint).”

“These [RGB imaging, NIR and NIR-HSI] are methods to monitor drying processes in real-time rather than drying methods themselves, as they do not enhance drying based on their principles. Instead, they are used as tools in drying systems,” he explains.

Some possible food applications go beyond fruits, for example, meat and fish, powder and grain drying.

The findings are published in Food Engineering Reviews and also provide an overview of standard industrial drying methods, such as freeze drying, spray, microwave, or hot-air oven drying, which can be integrated with the precision monitoring techniques.

Handling data with AI

The three methods analyzed must be combined with AI and machine learning to process the information and the models must be trained for each specific application, states the study. 

Meanwhile, the idea of using AI to power the team’s proposed smart drying system is that AI helps manufacturers treat huge amounts of data, similar to that generated by smart sensors, explains Ferreira.

“The idea of implementing them in the smart drying system is validated by this role. For example, in a large-scale machine, we are dealing with mass transfer and thermodynamic outcomes for thousands of products in a production line that involves managing and processing an enormous volume of data.”

Even when these phenomena are simplified into a few parameters, “traditional data treatment techniques often fall short due to the scale and complexity of the task,” he notes.

Moreover, in traditional drying processes, most industries deal with determining end points by taking each product over time by using a smart moisture meter to establish quality control. “Our intention is to eliminate this step and offer a real-time measuring system.” 

Retaining product quality

The F&B industry has long used food drying to preserve fruits and meat. However, the scientists note that drying can alter the food’s quality and nutritional value, which precision techniques, such as optical sensors and AI, can help reduce.

“In food production, AI-driven sensors can address inconsistencies by ensuring optimal product throughput without compromising quality, thereby maximizing efficiency and minimizing loss,” states Ferreira.

“These new techniques can not only help manufacturers produce consistent products but also enable the development of energy-efficient equipment, which plays a crucial role in supporting the United Nations’ agenda of reducing carbon emissions to achieve net-zero by 2050.”

Additionally, AI-based methods can help manufacturers boost food production “to the next level.” This is possible as they determine the endpoint in real-time (minimal delay) and ultimately minimize production loads since efficiency is a function of time in this scenario.

“By ensuring high-quality products are delivered in the shortest possible time, food manufacturers can save resources.” He believes the approach will reduce waste, maintain product quality and promote both efficiency and sustainability-core principles of the Cen­ter for Advanced Research in Drying’s mission, a US-based research centre that financially supported the study.

Modernizing drying operations

In the future, Ferreira expects AI-based NIR, NIR-HSI and RGB sensors to offer the new drying systems the ability to “work smarter” in every aspect of food processing, not exclusively in food but also pharmaceuticals and materials science.

“The existing drying methods will be positively impacted by these new AI-driven optical sensing harnesses that leverage the power of real-time datasets.” 

This will facilitate predictive modeling to anticipate drying outcomes and enable autonomous process control systems to optimize parameters for peak efficiency and quality “dynamically and consistently,” he tells us.

“This, we believe, will be the key to modernizing drying operations in the industry in the future.”

Analyzing implementation possibilities

Notwithstanding the technology’s “undeniable benefits,” Ferreira acknowledges that its adoption may be limited by cost, particularly for NIR and HSI systems, which can cost “several thousand dollars.”

“Validation will be a crucial factor in driving purchasing decisions and ensuring widespread implementation of these new technologies.”

The team is now evaluating NIR-HSI systems to better understand the moisture distribution within the tested products.

“Our future goal is to keep evolving in optimizing the system as well as the integration and application of cutting-edge AI algorithms. Understanding the intricate aspects of the diffusion process will make possible the perfect product quality for manufacturers and consumers,” he concludes.