Benchmarking of adjoint sensitivity-based optimization techniques using a vehicle ride case study

Abstract

Several studies on multibody dynamics optimization have been conducted. One important limitation of these studies is their computational efficiency, especially when optimizing a complex system’s performance. The co-authors developed a very efficient optimization technique based on an adjoint sensitivity analysis methodology. The scope of this article is to validate this technique by conducting a benchmark analysis against some of the most popular optimization methods, including gradient-based optimization using finite differences, design of experiment using optimal Latin hypercube, and design of experiment using full factorial design matrix. A vehicle system is used as a case study for optimizing its ride comfort.