Research On Torque Ripple Minimization For Speed-sensorless Direct Torque Control System

Direct torque control (DTC) has attracted much attention with its advantages such as new, simple control scheme and excellent performance. However, in one complete sample cycle of conventional DTC, the finite and discrete space voltage vectors applied dramatically increase or decrease torque and largely cause torque error exceed preset hysteresis bandwidth, resulting larger torque ripple. In this paper, the author makes a further research to the drawback about larger torque ripple in conventional DTC of induction motor drive at low speed region.Firstly, the reasons why classical DTC can cause larger torque ripple are analyzed in the view of physics and mathematics in this paper. Based on the above analysis, discrete space vector modulation-direct torque control (DSVM-DTC) scheme of induction motor is introduced in detail. It can be noted that a given voltage vector has a different effect on the drive behavior at different speed regions. According to the effects of different vectors, detailed switching tables in different speed regions and different sectors are built in DSVM-DTC scheme. Secondly, fuzzy logic control is applied to conventional DTC scheme. A fuzzy logic controller is adopted to replace the conventional hysteresis controllers of torque and flux. It solves the problems that the variation range of stator flux and torque can not be distinguished accurately and some small changes caused by indefinite factors can not be timely controlled in conventional direct torque control. At a given sector, some voltage vectors have dual functions (i.e. one half is positive, whereas the other half is negative). An improved fuzzy switching table is designed based on the influences of voltage vector to torque and flux, resulting smaller torque ripple. Then, the simulation analysis and comparisons on the effects with different defuzzification methods to the system performance are made in the paper. Simulation results show that COA method is worse than the other three methods that have similar effects.A speed estimator based Model Reference Adaptive Fuzzy Control (MRAFC) principle is presented in this paper. In classical MRAC, the adaptive law is a fixed gain P-I. The estimated speed lies larger overshoot in deceleration mode. So we propose a fuzzy adaptive law with self-adjusting P-I gains and build fuzzy rules for adaptive coefficients. In MRAFC estimator, the error between reference and estimated rotor flux continuously updates the P-I gains until the error vanishes to zeros. Finally, Popov's stability theory condition ensures the stability of the system. By simulation comparison, the proposed estimator is advantageous over the classical estimator in dynamic performances.In the end, to verify the effectiveness of the improved schemes, the simulation models of sensorless classical DTC, DSVM-DTC, FDTC of induction motor drive system are built by Matlab/Simulink. The improved direct torque control schemes are compared with conventional DTC method. The simulation results show that the improved DTC methods obtain good dynamic performances, dramatically reduce torque ripple and improve low speed performance of direct torque control.