The Research On Induction Motor Direct Torque Control On Fuzzy Neural Self-adaptive Methods

Direct torque control (DTC) of induction machine is a new control strategy in the field of variable speed AC motor drive applications developed in the mid eighty of last century. It directly controls the electromagnetic torque and stator flux linkage. It has simple structure, directly control means and high stable and transient performance. Conventional DTC employs two hysteresis comparators to regulate the stator flux and torque respectively, which results in approximately decoupling between the flux and the torque control. And also it brings torque ripple problem. The paper's emphases are two aspects: One is reducing the torque ripple using advanced control methods and the other is the research of the speed sensor-less schemes.Firstly, after introducing the induction mathematic model, the conventional DTC is discussed and the simulation model of it is build up by using of the SIMULINK toolbox of MATLAB software. Simulation results testify the DTC has good static and dynamic performance.Secondly, conventional control strategies are based on induction mathematic model. The performance would be affected by the model parameters or random noise. But the intelligence controllers does not need the precise mathematical model and have great robustness. In this paper the intelligent control theory, including fuzzy control and artificial neural network control are introduced. On the basis of it, two controllers are designed respectively, which are fuzzy logic controller and artificial neural network voltage space vector controller by use of BP network. Simulation work is done respectively for the two controllers above, and the simulation results prove that the two controllers both can depress the toque ripple effectively. The two type of controllers can both replace the conventional hysteresis comparator to make the structure more simple and improve the dynamic and static performance further.Finally, the speed sensor-less algorithm is a hot research field in recent years. When familiar several schemes on the speed sensor-less filed are listed recently, two different speed sensor-less observers are proposed by using model reference self adaptive system and self adaptive line neural network respectively. They enlarge the approaches of speed sensor-less technology and they are helpful to put in practice in the future. The simulation results prove that the schemes are correctness and effective.Through the research on this paper, it indicates that the advanced technology can improve the performance of induction motor direct toque control. And it will be a...