One of the most important criteria on the automotive industry, in suspensions specifically, is the comfort of the vehicle. Neural Networks (NN) methods were analyzed for the sole purpose of designing a controller capable enough to predict the road profile and control the behavior of the damper of the vehicle. A neural network was trained to estimate a road profile to, subsequently, introduce this trained road profile on to another neural network for a predictive control , where this neural network will be trained with the objective of increasing the comfort of the vehicle. In an auxiliary way, a QoV model with a semi-active suspension was used to design the controller and then be co-simulated in Carsim to be validated in a real vehicle. As a result, the control system achieved good results and improvements were obtained in terms of the comfort criteria.
| Date of Award | 2021 |
|---|
| Original language | English |
|---|