Damped Gauss-Newton method for direct stable inversion of continuous-time nonlinear systems

Existing numerical methods for stable inversion require the formulation and use of the system's inverse dynamics for implementation purposes. This paper introduces a new numerical method for stable inversion that is directly implementable using only the system's state and output equations. In addition to this practical benefit, the proposed direct stable inversion method is globally convergent and its local rate of convergence is quadratic. An example is used to compare the proposed direct method to existing indirect methods.