Rotational speed sensor is one of the most important components used in wind energy conversion system control strategies. The latter is very expensive and may be faulty due to the harsh environment working, causing by that a low efficiency operation of the system. The use of a speed estimator or an observer may be a good alternative solution for the use either in sensorless control or detecting the sensor failure or degradation (FDI and FTC), provided that the observer is robust and ensures high rotational speed estimation accuracy. This paper presents a comprehensive study to solve the optimization problem of a model reference adaptive speed (MRAS) observer in a typical wind energy generation system based on permanent magnet synchronous generator using five adaptation mechanisms: The first one is the classical MRAS observer; it is based on proportional-integral (PI) controller. The second one uses a fuzzy logic controller (FLC) with two inputs. The third one uses the single-input fuzzy logic controller which represents the simplification of the conventional FLC. The fourth one uses the sliding mode controller, and the last one uses the super-twisting algorithm. A detailed comparison between the five adaptation mechanisms is carried out. The obtained results show a good estimation stability as well as a fast speed estimation at dynamic regime for the improved adaptation mechanisms compared with the conventional PI controller.