Comparison of the speed estimation by an adaptive observer and by a dynamic neural network observer for an asynchronous machine

The article compares and characterises two nonlinear state observers for estimating the rotation speed of an asynchronous machine. The first is a deterministic adaptive observer based on dynamic modelling. The second is a dynamic neural net observer. A comparison is made of the dynamics, the convergence, the stability, the robustness and the ease of implementation of each observer. Finally, the simulations obtained with an asynchronous machine are presented in support of this comparative study.