Using hybrid AI to optimise a sustainable bonding process
For ecological and economic reasons, companies are increasingly looking for ways to make products lighter and stronger. It is therefore important to find the right combination of materials for a particular application. This means that a lot of research is being done into the way in which these materials can be combined. Sustainability is also often a crucial aspect here.
In Industrie 4.0, robots and cobots are deployed in many bonding processes. They create added value by supporting the operators and improving the quality of glued joints in difficult or repetitive assembly tasks.
To achieve optimal gluing and bonding processes, a lot of time and attention is therefore paid to the correct settings of these cobots and robots. Regular fine-tuning of the settings of the robot and the take-off unit, for example, is very important to achieve a good glue bead quality. However, this is often time-consuming and based on trial and error, which has its impact on the stability and robustness of the production process.
In order to speed up this fine-tuning process, Flanders Make has developed a method to optimise a number of important parameters more quickly by using Hybrid AI, without disrupting the production process.
From two days to three hours
When looking for the optimal joining technique for specific industrial applications, we can optimise several variables. These are not only the variables of the robot itself, but also the (climatic) conditions of the bonding, the materials used and the dosage of the glue.
In our recent research, our Joining & Material Lab measured and determined among others the robot speed, the pre-pressure of the dosage unit, the quantity of glue liquid per time unit and the distance between the nozzle and the work object (height).
Afterwards, based on the measurements and using expert knowledge, the best variables of the robot and conditions were searched with the hybrid AI algorithms of Flanders Make and UGent. Savings were calculated by minimising the variation of the glue amount on the surface, the width of the glue bead and maximising the speed of the glue application.
By using a hybrid AI model on the two focal points in this research, namely optimisation of height and speed, it was possible to reduce the setting time of these parameters from two days to only three hours.
In addition, a saving in production costs was achieved, leading to a more stable and robust production process.