Tokyo — Yokohama Rubber Co., Ltd has begun using a new in-house artificial intelligence system to predict the values of key tire characteristics, such as grip, wear and handling.
Developed under Yokohama Rubber's new AI utilization concept, HAICoLab ('humans and AI collaborate'), the system uses AI to make the predictions based on data input by tire designers, announced the tire maker Dec. 2.
The data, Yokohama Rubber said, could include specifications-related information for design parameters such as structure and shape, the physical properties of compounds, materials and evaluation conditions.
The Japanese manufacturer said it expected the system's ability to conduct "a large number of virtual experiments" to help it accelerate new tire development, reduce development costs, and develop better performing tires.
The system also will make it easier for less-experienced engineers to develop new tire designs.
According to Yokohama, the new system reduces the "deterioration of AI prediction accuracy" that tends to occur during tire design.
As the number and types of possible design parameters differ depending on the tire's internal structure, it is necessary to create separate databases used for AI learning according to the tire's structural features.
The use of such narrowly composed learning data can sometimes reduce AI prediction accuracy.
However, Yokohama said it improved the new system's prediction accuracy by transferring AI prediction ability learned in other related areas (transfer learning).
The new development comes a year after Yokohama started using an AI system to predict the physical properties of rubber compounds used in its tires.
The combination of the two, Yokohama said, will enhance the development of a wide variety of new tires.
