CFNN-based Exact Linearization Control for a Nonlinear Suspension System

This paper addresses the challenge of using an exact feedback linearization approach in conjunction with a cascade-forward neural network. The proposed control methodology is outlined as follows: first, the state space equation of a nonlinear active suspension system is derived. Next, the exact linearization method is used to convert the nonlinear system into a linear controllable one through coordinate transformation and state space feedback. A cascade-forward neural network is then used to determine the desired dynamics using principles of linear control theory. An important advantage of this approach is that it can handle the non-linear relationship between input and output components whilst maintaining the linear relationship between them. By using a linear control law, this approach facilitates the realisation of a practical nonlinear control system, which greatly simplifies implementation. Finally, simulation results based on a nonlinear active suspension system validate the effectiveness of the proposed methodology.