Development of All-In-One Virtual Application Process for Efficient R&H Performance Development

Jinhee Lee (Hyundai Motor)

In this paper, a new virtual application process called R&H All-In-One Application Process is proposed. This process is developed to achieve more efficient R&H performance development. This process consists of 3 sub-process. First one is the CarMaker model generating process based on CAT data that provides reliable vehicle model from SPMD and inertia database rapidly. Second process is the CarMaker design of experiment automation analysis process and model based benchmarking process. This process provides the environment for analyzing the effect of R&H performance on changes in the combination of the vehicle parameter sets and test conditions but also, contribution analysis through systematic separation method. The last one is the Handling/Steering feel test database with integrated analysis platform. Through proposed process, it is expected that engineers related R&H performance development can conduct efficient virtual activity above a certain level.

二慣性共振系の共振周波数より高い周波数領域でのばね定数とねじれ摩擦の同時同定

矢田　哲也・宮崎　敏昌・Padron Juan（長岡技術科学大学）・北条　善久（東洋電機製造）

多慣性共振系において，適切なモデル化や制振制御を行うためには，正確なパラメータの取得が重要となる．本論文では，二慣性共振系の共振周波数より高い周波数領域でのばね定数とねじれ摩擦を同時に同定する手法を提案する．提案する手法の有効性をシミュレーション及び実機実験により確認する．

機械学習を用いたドライバビリティ官能評価の開発と知見の定量化

浅野　貴文・土屋　康祐・柏倉　賢二・川口　拓真・志村　春樹・新　僚介・金子　隆・金堀　凌也（SUBARU）

ドライバビリティ性能の向上には，複雑な運転操作と車両挙動の組合せを，熟練者が経験と勘で網羅的に評価する必要がある．本研究では，膨大な試験工数を低減し知見を共有すべく，機械学習を用いた評価手法を開発した．更にエンジンベンチ試験で活用し，ECUデータから加速度を予測する事で実車に頼らない評価を試みた．

振動エネルギーを用いた加振条件作成への取り組み

下村　智也・松田　文一・畠山　俊克（日産自動車）・中村　勝彦・奥永　樹（IMV）

並進運動と回転運動が可能な6自由度加振試験は，実環境の振動を再現する試験に適しているが，実施できる試験機が限られる．
より汎用的な単軸加振機での試験実施に向け，S-N曲線をベースにした疲労損傷度を用いる手法に加え，振動エネルギーを用いた手法による振動条件の作成に取り組んだ．

AI-Powered Vehicle Concept Development
-Disruptive Method to Reduce Time-to-Market and Select Right Vehicle Architecture & Technology-

Mario Oswald･Joerg Schlager (AVL List)･Kisu Lee･Sungho An (Hyundai Motor)･Stefan Kellner･Nathan De Kerpel (AVL List)

This publication covers a comprehensive methodology based on AI (artificial intelligence) presenting a paradigm shift in the selection of the right technology for all vehicle systems. The overall goal is to not only meet but also balance vehicle-level targets, setting the stage for a more streamlined and efficient development process. The center of this innovative approach are multi-layered and cascaded neural networks, which act as powerful tools enabling precise and simultaneous prediction of vehicle attributes like energy efficiency, range, charging, performance, vehicle dynamics and driveability. Furthermore, these networks even provide insights for the optimization of geometric vehicle architecture and crash.