Abstract: The common speed sensor-less motor control methods are introduced and classified according to parameter robustness and system anti-interference ability. Completely dependent on the physical parameters of the motor and the electromagnetic equation speed estimation algorithms have poor robustness and weak anti-noise ability. Partially dependent on the physical parameters of the motor and the electromagnetic equation speed estimation algorithms are divided into four methods. The model reference adaptation system (MRAS) is introduced, and the model reference adaptive system and sliding mode observer are simulated. Two high-frequency signal algorithms and some artificial intelligence algorithms independent of the physical parameters of the motor and the electromagnetic equation speed estimation algorithms are described. The advantages, disadvantages and applicability of each algorithm are summarized.