AI will help Yokohama design molds
Yokohama Rubber announced that it has developed a mold design system using modeling technologies (finite element method, FEM) and artificial intelligence. The new development is expected to simplify the design process, including for less experienced employees, who will be able to obtain information, for example, on how changes in mold design affect tire characteristics, by conducting a large number of virtual experiments. The company expects that the new system will reduce the time required to create molds, while also reducing costs and the need for revisions, which are often necessary during the design process. In addition, a deeper understanding of the relationship between mold parameters and tire characteristics is expected to enable employees to develop products with improved properties.
“Mold design has a significant impact on key tire parameters, and a time-consuming and costly trial-and-error process was required to fully understand their relationship, involving the production of prototypes and subsequent testing,” notes Yokohama. “In addition, the process was largely dependent on the knowledge of highly qualified personnel, which is why the efficiency and speed of development varied depending on the experience and qualifications of individual employees.”
The new system is designed to solve existing complexities through automated modeling and AI prediction. At the first stage, a large number of FEM models with various mold parameters are generated, and tire characteristics are predicted, after which the calculation results are used as training data for building an AI model (surrogate model) that predicts the relationship between mold design aspects and tire properties. At the next stage, using such explainable AI technologies as SHAP (Shapley additive explanations) and PDP (partial dependence plots), developers can quantitatively visualize the impact of mold design parameters on tire quality and determine what changes are necessary to ensure the target characteristics of the final product.
Yokohama has been applying the HAICoLab AI-based system since 2020, which has been used since 2021 to predict key tire parameters, and since 2024 has been more widely used in the development of new products. The company says that thanks to the new technologies, it is easier for employees to understand what actions to take to achieve the optimal result, and in the future, it plans to continue working in this direction to further accelerate the creation of new innovative products.