Research On Direct Torque Intelligent Control Methods Of Asynchronous Motor

The Direct Torque Control(DTC) is a high performance Alternating Current(AC) speed regulating technology after vector control. It arouse wide concern because of its straight control thought, simple system structure, convenient for digitalization, as well as good static and dynamic qualities when it was proposed. Its basic principle is changing the angle between stator flux and rotor flux by adjusting the rotating speed of stator flux while the amplitude of stator flux is constant, then the electromagnetic torque and rotor speed are controlled. Compared with vector control, no need to coordinate transformation and control flux and torque straightly are DTC's main advantages. DTC is applied successfully in some fields. But as a new technology, it has some deficiency in term of its theory and practice application, needs to be advanced improved.Starting from the mathematical model of asynchronous motor, DTC's theory and system structure are analyzed deeply, found the key points to affect system main performances are the judgement of stator flux bias and torque bias and choice of voltage space vector. Therefore, some improved methods of switch state choice form are proposed in order to perfect stator flux map and restrain torque ripple.The first scheme is subdividing the flux bias and torque bias, increasing amount of combination of flux hysteretic comparer outputs and torque hysteretic comparer outputs and optimizing the choice of voltage space vector so that implement regulating stator flux and torque more accurately.The second scheme is to introduce fuzzy logic into DTC system. The new fuzzy controller replaces the traditional flux and torque two points Bang-Bang controller. Adopt member functions to estimate flux bias and torque bias and take them together into consideration, so as to achieve big error could be adjusted rapidly while small one should be adjusted accurately. The simulation results indicate that fuzzy logic voltage vector controller make system have rapid dynamic response, stator flux map is close to round very much and torque ripple is confined excellently.The third scheme is to implement the choice of voltage space vector with neural network. The neural network voltage vector controller employs BP network with forward structure and error backpropagation. It's trained off line and controlled on line. Because of neural network's strong tolerance, robustness is increased. Besides, neural network could process more rapidly as its parallel operation ability, and then the switch frequency could be further increased. Hence, the system has favorable dynamic response.Aimed at comparative analysis of the three improved methods, the simulation models of all kinds of control algorithms were established on Matlab. The simulation results verified that these control schemes are feasible and achieved anticipative effect.