Speed sensors of ventilation motor in mineral well are susceptible to various problems, such as water seepage and blockage as well as lack of feedback signals for entilators. As soon as they happen, the ventilation motor will not be controlled. These questions are mainly caused by the damp and dirty working environment. To solve the problems, control System of Speed Vector on ventilation Motor based on ARM (Advanced RISC Machine) is proposed, with IPM (Intelligent Power Module) PS21865 as inverter. The hardware design optimized system structure and expanded the controller memory and compatibility. Based on the optimized system structure, an estimation way of speed vector for the ventilation motor was given with main current control, auxiliaryvoltage control and speed-sensorless, The simulation results show that the adopted methods are correct and effective. With the rapid development of coal industry, a ventilator is widely used in safety production of coal mine because of its structure is solid and durable structure to prevent itself from moisture, explosion and low noise. And it has high efficiency and energy saving, too. But the environment in mineral well is wet and muddy. This environment often leads to water and blockage of speed sensorwithout feedback signal to ventilator. Once a ventilator has no feedback signal to be out of its control, causing poor ventilation, the gas concentration to increase and the temperature to rise in the well head. These conditions will result in gas poison or explosion. To address the problems, the estimation control of speed vector for the ventilation motor was proposed through a variety of transformations and based on the rotor field orientation with the equivalent principle of magnetic field. It used the actual speed of fan motor as the speed feedback signal by indirect calculation method, without speed detecting device. The core problem of the control strategy is accurate observation of the rotor flux and accurate esti- mation of motor speed. The current of motor stator coil can be directly detected, but magnetic flux and electromagnetic torque can not be measured, and we can use the decoupling method. The stator cur- rent vector can be divided into two components and they are mutually vertical and independent. Then a component is used to produce excitation flux, the other is used to generate the electromagnetic torque, This can approximately realize the indirect control of flux and torque. However, at high speed the speed estimation method on a ventilator can ignore the error influence. At low speed, the influence of magnetic