Search Results

Most of the existing control techniques applied to automobiles are based on a linear model of the relevant dynamic system. However, in spite of the increasing demand for more sophisticated vehicles, very little research is being directed toward understanding and predicting the nonlinear response caused by non-predictable operating conditions such as the variation of suspension characteristics due to wear and tear or the frictional coefficient of the road. In this research, we used a neural network based identification technique to predict the structural response of dynamic systems and validated the technique through comparison with experimental data obtained with a cantilever beam. This identification technique is further extended to forecast the frictional coefficient μ of roads. Numerical results indicate promising future applications.

This study is motivated from the investigation of vehicle suspension system with changeable damping and variant stiffness parameters. Such suspension system can be modeled as a dynamic polytope based on the mapping of affinely changed parameters. According to the polytopic dynamics decomposition, knowledge of linear time invariant system can be applied to each polytope vertex and the time variant system is solved by the polytope convex synthesis method. For time variant vehicle suspension system, the different model structures for control purposes are formulated. A quarter-car is taken as the example for demonstration in observer design and nonlinear control design.