Intelligence Algorithm And Optimizing Method Applied In Speed-sensorless DTC System

With the development of the computer, autocontrol technology and the electron and electric power technology, the performance of the frequency conversion of AC drive has been improved greatly, and some new technique such as Vector Control and Direct Torque Control have been applied widely. At the same time, the technique of the frequency conversion of AC drive has been more and more consummate. Recently artificial intelligence has showed its power in the field of the AC drive control. It is proved by some researcher that the AC system can get high performance in self-adapt, self-diagnose, self-protect and dynamic capability. In this paper, some artificial intelligence, such as Fuzzy Control, Genetic Algorithm and Neural Network, are applied in Direct Torque Control (DTC) system, and lots of emulations are done under MATLAB/SIMULINK.As a excellent control method, Direct Torque Control takes the parameter of torque as its control object, and it based on stator coordinate. As a result, DTC system is affected by fewer parameter of the motor and has less complex structure and fewer calculations. Since it was brought forward, Direct Torque Control has been took serious attention by domestic and foreign scholars because of its novelty idea, simple structure and wonderful property. This paper adopts the Fuzzy controller to select the voltage vectors after expound the basic theory of DTC, and this method is improved to be effective by the result of the emulation. However, the design of Fuzzy controller is always subjective and blindness, a method used Genetic Algorithm to optimize the Fuzzy controller is brought forward.In order to obtain a high quality in the low-speed area of DIC system, a precision model of motor in the low-speed area is necessary of which a important parameter is rotate speed. Because the installation and the maintenance of the speed sensor have brought about many troubles to people who use it, furthermore, it has less accuracy. The exploitation of the speed-sensorless is able to not only reduce the cost of the whole system but also enhance the reliability, so it is under hot discussion nowadays. In this paper a method of speed identification applying the Neural Network technology is presented, and the model of the Neural Network is built up. Since the BP algorithm has the low speed in studying and can'tensure reach the whole minimum. In this paper, a fuzzy-inference-based neural network (FNN) is presented, which is applied to speed identification in a Direct Torque Control (DTC) system. The proposed scheme uses a fuzzy inference system to verify the learning coefficient and the momentum term in neural network. To test this approach, simulation results are shown in the end of the paper.