We present a flight dynamics nonlinear model for a flybarless helicopter UAV, valid for a range of flight conditions, including the Vortex-Ring-State (VRS) and autorotation. To allow for computational efficiency, while maintaining a high-level of model fidelity, a greybox modeling framework has been adopted, in which model uncertainty such as parameter uncertainties, unmodeled higher-order dynamics, and unmodeled static nonlinearities have been replaced by empirical coefficients. The derivation of these coefficients has been based upon a novel identification approach, anchored in the combined paradigms of nonlinear optimal control and neural networks. Preliminary simulation results show that our model is in good agreement with an equivalent FLIGHTLAB model.