We use cookies to operate this website and to improve its usability. Full details of what cookies are, why we use them and how you can manage them can be found by reading our Privacy & Cookies page. Please note that by using this site you are consenting to the use of cookies.
Food
ARTIFICIAL INTELLIGENCE FOR MORE PROCESS QUALITY
2025-01-13
Every year in Germany, about 11 million tonnes of food are thrown away. Per capita, this is around 78 kilograms. Around 15 percent of this waste are caused by processing industries, 17 percent by the out-of-home catering sector and seven percent by commerce, according to the German Federal Ministry of Food and Agriculture.
There are many different causes for food waste, ranging from overproduction and incorrect storage to transport damage and technical problems in the industry to over-ordering in commerce and from individual customers.
Artificial intelligence (AI) offers a broad range of opportunities to avoid overproduction in the future through optimised processes and to lengthen the shelf life of food. Collecting data and analysing them for process optimisation are of special importance here.
Production plans that are geared towards actual need, quality-led manufacturing processes, production systems that minimise waste or loss-optimised strategies for procuring and sales are some of the complex-sounding ideas for increasing sustainability along the food industry’s value chains.
REIF research project
In April 2020, an AI research project slated to run for three-and-a-half years was launched under the guidance of what was then the German Federal Ministry for Economic Affairs and Energy (today the German Federal Ministry for Economic Affairs and Climate Action, BMWK). The project is called REIF for Resource-efficient, Economic and Intelligent Foodchain (https://www.tha.de/Binaries/Binary_72303/Resource-efficient-economic-and-intelligent-foodchain.pdf).
“Conventional technologies have their limits when it comes to goals like minimising overproduction of food and avoiding waste,” says Prof. Dr.-Ing. Stefan Braunreuther from the Technical University of Applied Sciences Augsburg. “There are enormous amounts of data to be determined and analysed during the manufacturing and delivery processes. This is only possible using AI. The REIF project aims to determine potential ways to minimise losses, especially in the dairy, meat and baked goods industries.”
So how does that actually work, using AI, big data analysis and the manufacture of products in order to get products tailored exactly to customers’ wishes, with consistent quality and optimised shelf life? Products that are guaranteed to be in demand as opposed to gathering dust on the shelves before just being thrown away in the end?
AI for fluffy layer cakes
One concrete example reveals the potential. Let us imagine a lot of sponge cakes that are quickly sold out because the sponge tastes good, each lot is the right type of golden brown that the customers want and the consistency is optimal.
When producing these sponge layers, different ingredients like fat, flour, sugar, eggs and liquids all come together – and their behaviour during the baking process influences the final result.
Each ingredient plays a specific part, for example for the viscosity of the dough or the taste of the finished product. For example, flour gives the dough an even structure and by forming strands of gluten and the resulting gas bubbles, the sponge acquires an even distribution of pores. Sugar caramelises during baking and contributes to the appealing golden colour of the crust, while lastly butter determines the delicate, soft taste and supports the formation of the desired aerated crumb. Egg yolk determines the golden colour which entices consumers to buy, winking at them from the packaging, while egg white contributes to the firmness and stability of the sponge.
All ingredients interact during the manufacturing, baking and cooling process so that a multitude of parameters like humidity, the consistency of the dough, stirring time, air supply as well as the guided temperature curve influence the finished product. Even the shape of the baking mould and its material have an effect on the end product. Each of these parameters can be controlled with the help of artificial intelligence in order to achieve the desired result through optimised interaction of all factors.
Sensor monitors dough in feed pipes
The cooperation between industry, associations and research institutes during the REIF project led, among other things, to the development of a specialised sensor for an industrial manufacturer of sponge cake layers. The sensor analyses the foam structure of the dough before it is filled into the baking moulds. “For this product, the structure of the foam is critical for the final quality of the sponge,” says REIF project head Braunreuther.
The newly created sensor – a soft sensor, i.e. a model that predicts process variables – is now installed at the manufacturer's feed pipes. It uses ultrasound waves which enable to draw conclusions regarding the structure of the foam from its dampening behaviour. At the beginning of the baking process, the dough is wet and has a certain consistency. During this phase, dampening is relatively high, as the dough is still liquid. During the baking process, the dough dampens fewer and fewer ultrasonic waves as it becomes more firm and can transmit them better.
The foam structure is directly linked to product quality over the further process of baking. Precise control of all the different parameters, such as the recipe for the dough, consistency, temperature, baking time as well as a precise analysis of the dampening behaviour now offer exact control of the baking process and over the final product.
Analysing huge amounts of data
In meat processing, artificial intelligence can, for example, determine the optimum rotational speed of the conveying screw and thus the speed at which the meat is transported through the machine for cutting or shredding. The more heat a process generates, the higher the microorganism load and the lower the expected shelf life of the finished meat product.
Among the applications used are fluorescence spectroscopic methods which help to detect microbiological and chemical changes in meat products. Experimenting with vastly different production parameters produces huge amounts of data, which must be analysed in favour of optimising shelf life.
“AI algorithms, especially those related to machine learning, can process huge amounts of data and recognise patterns that correlate with the physical age of germs and the shelf life of food products,” explains process specialist Braunreuther.
Great challenges for new standards
Precise planning of raw materials or early detection of faults in production are other equally important areas of application for AI, as are optimising mixtures or identifying alternative ingredients or the ratios which will increase the freshness and stability of products and thus affect their shelf life.
AI models are able to analyse environmental conditions and the physical and chemical properties of products in order to decide consequences for optimum storage and distribution on schedule.
“One of the great challenges for manufacturers in food production is defining possible areas for optimisation,” says process expert Braunreuther. “Which focus do you set? Which data do you need? Which are the critical points in time during the production process, those which influence shelf life? How can data be recorded and brought into a suitable format? Which algorithms and models are best suited to predict shelf life?
The list of questions is far from closed. Which process parameters have the biggest influence on shelf life? How can conditions along the supply chain be monitored in real time? Which sensory analysis or microbiological tests are needed? What can automated test systems look like?
“Technological progress with the help of AI is no topic for a weekend workshop,” Braunreuther adds. “The companies which took part in REIF needed several years to work out concrete ideas for their areas of business.”
AI offers huge chances and poses extremely high challenges to those who want to raise the standards to the next level. Implementing such solutions requires comprehensive interdisciplinary cooperation between data scientists, food technologists, engineers and other experts. The future will bring new solutions and more sustainability to the sector. One thing is clear: the excitement is not over.
About the SAVE FOOD Initiative
The SAVE FOOD Initiative supports innovative projects which enable a sustainable way of treating food the world over. The SAVE FOOD Initiative was founded in 2011 by Messe Düsseldorf and interpack and has since cooperated with, for example, the Food and Agriculture Organisation of the United Nations (FAO) and the UN Environment programme. The initiative stands for supporting the UN goals for sustainable development (SDGs), especially the goal to reduce the worldwide waste of food (SDG 12.3). The initiative aims for the use of innovative processing and packaging technologies to secure a sustainable food supply while at the same time using resources efficiently.
Sign up for our monthly newsletter to get newest information about the packaging sector:
- latest news
- trends
- expert interviews
- infographics
