A Study On Application Of Artifical Intellignce In Direct Torque Control

The basic principle of speed adjustment of motor is control of torque. Direct torque is based on stator flux field-oriented method, its structure is simple, and control torque directly. This dissertation analyzes direct torque control fully, and makes a conclusion that the key of realizing DTC is to choose stator voltage vectors reasonably according to the request of torque and flux. Also the functions of voltage vectors in each section are analyzed.In order to combine the intelligent control methods with the DTC theory, the design of controller for fuzzy control and neural network control and particular algorithm are summarized. Some intelligent strategies served to resolve the difficult problems are put forward.To improve the performance of torque control because the flux is not estimated precisely during the period of low speed, an on-line fuzzy observer for stator resistance is put forward.In order to overcome the weakness of Bang-Bang controller in DTC, a new kind of fuzzy direct control method of torque and flux is put forward; the design method and reasoning rules are presented. As for the design of membership functions depends on experiences of human heavily, a strategy of optimization of membership functions is presented by using fuzzy neural network. To resolve learning problem of FNN, an arithmetic operator of "soft maximum" is given.This paper describes a newly speed sensorless drive strategy based on neural networks. A back-propagation neural network is used to provide real-time identification of the motor speed. The object function is the sum of squares of the differences between object model and neural network model. Speed signal is regarded as a weight of neural network. The back-propagation algorithm is used to adjust the motor speed, so that the neural model output follows the real motor speed. An identification method of stator flux using the recurrent neural network is presented also.A neuron adaptive close-loop control tactic is given to improve the robustness, dynamic and static performance of intelligent DTC close-loop adjustable speed system. The design method and related algorithm of adaptive speed controller are discussed in detail. To remedy the defect that the neuron gain can't be adjusted online, the adaptive algorithm is adopted to correct the neuron gain.