Neuro-Fuzzy Dynamic Inversion Control for a Hypersonic Cruise Vehicle

A method of "nonlinear inversion" is developed in this paper. This nonlinear approach captures all of the complex cross-coupling dynamics of the vehicle control effectors for the entire flight envelope. Important nonlinear effects such as those due to flex-to-rigid control ratios, non-symmetric c.g. location and control saturation due to local aerodynamic separation can be easily captured using this inversion method. Additionally, traditional linear analysis theory can still be applied to this inversion method for analysis of aeroservoelastic effects, robustness and digital sampling effects. Most linear inversion methods require simplifications where nonlinear effects are often omitted and complex switching logic is needed to handle effector saturation conditions. Other optimization-based inversion methods may result in non-repeatable conditions that can be a concern for verification and validation testing and for certification demonstration. However, the nonlinear inversion algorithm proposed in this paper can impose a large computational burden due to its large multi-dimensional database and can become a concern for operational flight software implementation. A fuzzy logic algorithm is therefore developed in this paper to significantly reduce the computational requirement for real-time implementation and at the same time accurately capture the complex inversion database, which is crucial for the problem of augmentation and control of unstable vehicle.