Set-Based Design Method for Rigid Axle Suspension using Bayesian Active Learning
-
- Shiraishi Hideki
- トヨタ自動車(株)
-
- Shintani Kohei
- トヨタ自動車(株)
-
- Iwata Motofumi
- トヨタ自動車(株)
-
- Takada Yasuaki
- トヨタ自動車(株)
Bibliographic Information
- Other Title
-
- Bayesian Active Learning を用いたリジッドアクスルサスペンションのセットベース設計法
Abstract
It is important to discover feasible region that satisfy multiple performances in the early stage of vehicle development. In this paper, we propose a set-based design method of rigid type suspension by introducing machine learning. In the proposed design method, surrogate model of the characteristics of rigid axle suspension are trained by using Gaussian process (GP). By using the posterior distribution of GP, adaptive sampling strategy to find feasible region is introduced. To show effectiveness of the proposed design method, a numerical example is demonstrated. In the numerical example, feasible region of suspension characteristics that satisfy multiple performances was identified.
Journal
-
- Transactions of Society of Automotive Engineers of Japan
-
Transactions of Society of Automotive Engineers of Japan 54 (2), 259-264, 2023
Society of Automotive Engineers of Japan
- Tweet
Details 詳細情報について
-
- CRID
- 1390857977632358784
-
- ISSN
- 18830811
- 02878321
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed