Accurate Shock Absorber Load Modeling in an All Terrain Vehicle using Black Box Neural Network Techniques

This paper presents the results of a study of using a neural network black box model of a shock absorber of an ATV (All Terrain Vehicle, four wheel drive, off road, single person vehicle) for accurate load modeling. This study is part of a larger investigation into the dynamic behavior and associated fatigue of an ATV vehicle, which is conducted under the auspices of the Fatigue Design and Evaluation Committee of SAE of North America (www.fatigue.org).

The general objectives are to develop new correlated methodologies that will allow engineers to predict the durability of components of proposed vehicles by means of a “digital prototype” simulation. Current state of the art multi body dynamics predictions use linear frequency response functions or non-linear polynomial approximations to describe the behavior of non-linear suspension components such as shock absorbers or bushings.

The proposed method yields more accurate predictions due to the fact that both the non-linear and hysteretic behavior of the shock absorber are modeled. This paper demonstrates how neural network black box technologies, particularly in the form of the Empirical Dynamics Models, can be used for accurate prediction of shock absorber loads encountered by a vehicle and the potential improvement in fatigue life predictions under this approach.