Potential of statistical model verification, validation and uncertainty quantification in automotive vehicle dynamics simulations: a review

Benedikt Danquah, Stefan Riedmaier, Markus Lienkamp

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

The modelling, simulation and analysis methods in automotive development are in the process of transformation. Increasing system complexity, variant diversity and efforts to improve efficiency lead to more complex simulations and depend on virtual vehicle development, testing and approval across a large application area. Consequently, the new key requirements of modern validation involve more precise reliability quantification of large application areas, achieved with reasonable effort of cost and time. This paper identifies that the neglection of uncertainties, low information in validation results, low extrapolation capability and the resulting small application area are preventing the state-of-the-art validation meeting those new requirements. In an extensive analysis examining more than twenty frameworks in detail, this paper shows that statistical methods exhibit a high potential to remedy these four key insufficiencies. The paper justifies comprehensively that consistent statistical validation is necessary, important and crucial for precise reliability quantification, which enables accurate model selection, knowledge building and decision making in modern automotive vehicle-dynamics simulations. An example is given explaining the basic principle and benefit of consistent statistical validation. Since automotive statistical methods are still at the beginning, the aim is to enable further investigation by showing their potential and providing deeper knowledge about this topic.

Original languageEnglish
Pages (from-to)1292-1321
Number of pages30
JournalVehicle System Dynamics
Volume60
Issue number4
DOIs
StatePublished - 2022

Keywords

  • Modelling and simulation
  • automotive vehicle
  • statistics
  • uncertainty quantification
  • validation
  • verification

Fingerprint

Dive into the research topics of 'Potential of statistical model verification, validation and uncertainty quantification in automotive vehicle dynamics simulations: a review'. Together they form a unique fingerprint.

Cite this