Research On FNNS For Induction Machines Using Direct Torque Control

With the high speed development of power electronics and computer technology! AC electricity transmission systems are replacing DC ones step by step. The representative technology for AC motor control have VVVF, VSVF, VCandDTC.To obtain a quick response, the torque of the motor must be controlled effectively. But WVF can only control the air-gap flux and can not adjust the torque. Being based on the steady state equations of the motor for the design of VSVF, VSVF can control the torque to a certain extent, but it can not implement the real control in the dynamic process. Although it can dramatically improve the dynamic peculiarity in the theory, the vector control approach not only makes the complex reference frame changes, but also rely on the parameters of the motor to a large extent. At the same time, the completely uncoupling in the dynamic process is not ensured and the effectiveness of the torque control is reduced. But DTC is a kind of the more advantageous control technology that places stress on controlling the torque of the motor directly, which overcomes the disadvantages of VC and brings the second natural fly-by of the speed regulation control theory of AC motor.The theory and technology of DTC have a lot of advantages, however, as a kind of the new birth theory and technology, the unperfected aspects and some ones are difficult of being solved. Owing to the reasons above, DTC has become one of research not points. To the problems of compensating stator flux linkage, tuning stator resistance and the problems in the low speed region, it is proposed to solve above problems using Artificial Neural Network and Fuzzy Control.The dissertation consists of seven chapters. In the chapter 1 the presentation, the origin, the purpose and the significance are stated, the development and research status quo, the research direction and the main difficulties in DTC system domain are outlined, and the work and achievements in the dissertation are introduced. In the Chapter 2, the parameters variable and the control strategy of FNNS are presented based on the analysis of philosophy and fundamental equations and control strategy in DTC. In the chapter 3, based on the analysis and summary for the stator resistance-current relation, the fuzzy inference parameters of the stator resistance and the inference rules are determined, and the control parameters of the switching state selector are optimized. In the chapter 4, the particular network topology and the four training algorithms are designed for the switching state selector in DTC by the neural network control technology, and the learning speed, stability, and weight convergence of the algorithms are discussed and compared . In the chapter 5, the complete system simulation of the DTC is performed using MATLAB/SIMULfNK program. The simulation results of the DTC using the switching lookup table with and without neural networks method are compared. In the chapter 6> simulation results are tested by means of DSP and computer system. The chapter 7 is the conclusion and the further expectation of the related research work is put forward.