Semi-active damper suspension road estimation and control based on neural networks

The use of semi-active damper systems in commercial automotive vehicle applications has become widespread in
recent years. In contrast with typical passive shock absorbers, semi-active dampers allow modifying the damping coefficient
online, introducing the requirement of appropriate control algorithms. In this article a novel approach for semi-active damper
manipulation in automotive vehicles is presented. The proposed scheme is based on a Neural Network (NN) road-profile
reconstruction together with a NN-based predictive controller. The main objective of the system is to improve passenger comfort;
however, the algorithm can also be tuned for road holding or a combination of both. The resulting suspension control system was
validated trough simulations using the CarSim software, which includes a comprehensive representation of the vehicle, and a
well-known non-linear damper model. The results show that significative improvements can be achieved with a relatively simple
NN topology, which is well posed for real-time implementation due to its low computational complexity.