High performance computation for flux and speed estimation of vector controlled adjustable-speed drives

Abstract

Induction motor has been widely used because of its salient advantages. Its potentially excellent dynamic performances can indeed be achieved with suitable control strategy. However, it is very difficult to control precisely, because it is a coupled, nonlinear and high-order system. An additional complexity is that the parameters of induction motor are dependent on the operating conditions. Speed estimation is very attractive for motor control. This work focuses on decoupled, robust high performance motor control, especially on flux and speed estimation for rotor field orientation vector controlled induction motor drives. These estimation algorithms are critical techniques for high performance vector controlled adjustable-speed drives. A novel observer-based the flux estimation scheme for speed sensorless operation is developed in this work. This observer does not require rotor resistance value and rotor speed information. Consequently, the accuracy is not influenced by rotor resistance deviation and speed measurement errors. The speed estimation can be obtained using the observed flux value to implement the sensorless operation. A simplified indirect rotor field orientation control (IRFOC) implementation method is presented in this work. This method does not need to use current and voltage sensors. It also removes two current feedback loops and their associated controllers, resulting in overall design simplicity and cost reduction. A novel stator synchronous frequency estimation algorithm is proposed. The vector control performances are completely immune from motor resistance parameters variation or mismatching. Rotor current is derived in rotor field-oriented reference frame, and both stator and rotor resistances are eliminated in the final algorithm. Furthermore, the mechanical speed signal is not required, thus speed measurement error does not affect the estimation, and hence the correct field orientation can be maintained. An adaptive flux position observer is developed based on adaptive theory. This observer does not need any prior knowledge of motor parameter, the flux position is estimated by adaptive law when the error transfer function is strictly real positive. Thus the parameter sensitivity problem and the online identification algorithms, which are quite common in most observers, are not required in this method. In order to guarantee stability and convergence of the algorithm, strictly positive real error transfer function is obtained through the introduction of filters and the rearrangement of the input signals. Normally, when induction motors are fed by SPWM inverters, the "sufficiently rich" requirement for estimation convergence can be satisfied.