NTN Enhances Automotive Hub Bearing Design Efficiency through AI Integration

NTN Corporation (hereafter, NTN) has introduced machine learning*¹ technology based on AI (Artificial Intelligence) into its automated calculation system, which has been used in the design of 3rd-generation hub bearings that support the rotation of automobile tires. This represents the first use of this method in the bearing industry*². By doing so, NTN has significantly accelerated the time required for performance evaluation analysis to less than one-tenth of the conventional time, while also enabling the automatic design of dimensions that meet requirements. This helps reduce design workload and shorten customers’ development periods.

In automotive parts design, the introduction of model-based development (MBD), which simulates performance on computers to create higher-quality products more quickly, is progressing. In 2022, NTN introduced the automated calculation system “ABICS,” which performs a series of design processes for 3rd-generation hub bearings, reducing design person-hours by approximately 80% compared to conventional methods and contributing to shorter development time for our customers.

When using FEM analysis*³ to verify whether the design meets customer requirements, if the requirements were not satisfied, a redesign followed by another FEM analysis was required. However, 3rd-generation hub bearings required advanced calculations through FEM analysis due to their complex shape, which integrates the bearing and peripheral components, such as bolts.

By introducing AI technology into “ABICS,” we have achieved high-speed prediction of certain FEM analysis tasks in less than one-tenth of the time. Furthermore, if the results do not meet the required specifications, the system automatically suggests appropriate design dimensions.

By combining a simulation model using Lasso Regression*^4, which selects only the necessary data from a large dataset to predict analysis results based on input design dimensions, with Bayesian Optimization*^5, which is an algorithm for efficiently obtaining optimal solutions, we achieve high-precision predictions and dimensional proposals across a wide range. NTN is the first company in the bearing industry to introduce machine learning technology using Lasso Regression with Bayesian Optimization into its design process.

NTN aims to leverage the newly introduced AI technology to enable automatic prediction of all FEM analyses implemented in ABICS and provide optimal design proposals by FY2029. Once all these functions are implemented, design person-hours are expected to be reduced by more than 90% compared to before ABICS.

NTN will also continue to utilize digital technologies such as CAE and AI to improve the efficiency and sophistication of research and development operations, enabling us to promptly offer high-performance, high-quality products to our customers. In addition, we will actively develop digital talent capable of utilizing these technologies.

Comparison of FEM analysis workload with AI implementation

Image of the design process using AI

Notes:

*¹ Machine learning: A method that learns patterns from data to perform prediction and classification
*² According to our internal study of publicly available sources, including academic publications, as of December 2025
*³ FEM Analysis: A computational method for numerically analyzing physical phenomena such as stress concentration and deformation on a computer by dividing complex shapes or structures into fine elements
*⁴ Lasso Regression: A type of regression method (a general approach for predicting numerical outcomes from given inputs) that selects important variables and builds a predictive model
*⁵ Bayesian Optimization: An algorithm for efficiently finding the optimal solution with as few trials as possible in a short time by selecting the next trial point based on the results of previous trials