(English follows Japanese)
Japanese Title: Utilization of CAE and Statistical Methods in the Automotive Industry (NVH Robust Design and Multiobjective Optimization Study)
Altair Engineering Co., Ltd. Marketing Department
Technical marketing manager Ikuro Shibata
CAE has become an indispensable tool for pursuing automobile NVH (vibration and noise) characteristics and collision safety, but in general CAE is deducted to uniquely quantify input and output relation to a model It is considered as a deterministic and deterministic tool, and it is not possible to perform probabilistic evaluation considering various uncertainties possessed by actual car body.However, probabilistic evaluation is important to promote the movement of trial production (aggressive utilization of virtual prototyping by CAE) being advanced in the automobile industry. In the NVH robust design method actually carried out as a collaborative research with an automobile manufacturer, cases using JMP in DOE, multiple regression analysis, probabilistic prediction model construction using simulation on a mathematical model, and cases with higher order nonlinearity We introduce examples of utilizing JMP in DOE, cluster analysis, neural network etc in pattern recognition and machine learning targeting deformation mode control of unstable automobile body structure.
The Application of Computer-Aided Engineering and Statistical Methods in the Automotive Industry
(An Investigation of NVH Robust Design and Multi-Objective Optimization)
Ichiro Shibata, Technical Marketing Manager, Marketing Division, Altair Engineering
Computer-aided engineering (CAE) has become an indispensable tool for the pursuit of automotive NVH (noise, vibration, and harshness), collision safety, and so on. However, generally speaking, CAE is believed to be a deductive and deterministic tool that uniquely quantifies input-output relationships in models and can not be used for stochastic evaluation, considering the various uncertainties of actual automobiles. That being said, in order to promote prototype-less methods, which are gaining traction in the automotive industry, stochastic evaluation is important. Through joint research with automotive manufacturers, I use NVH robust design methods to conduct DOE and multiple regression analysis and build probabilistic projection models using simulations in numerical models with JMP. I will introduce a case example of DOE, cluster analysis, neural networks using JMP for pattern recognition and machine learning for the deformation mode control of high-order, nonlinear and indeterminate automotive structures.
This post originally written in Japanese and has been translated for your convenience. When you reply, it will also be translated back to Japanese.