ANN based optimal flux determination for efficiency improvement in Direct Torque controlled induction motor drives

Squirrel-cage induction motors are replacing DC motors in variable speed applications due to the availability of advanced control techniques such as Direct Torque Control and Field Oriented Control. At part-load operation of an induction machine, the core losses and, to some extent, copper losses can be minimized, if the machine is operated at a reduced value of flux rather than at the rated value. There is no straightforward mathematical relationship that determines the reduction in flux that can be incorporated for a given load on the machine. In this paper, an Artificial Neural Network (ANN) based control scheme has been proposed for arriving at the most suitable flux value, given the speed and torque requirement at any operating point of the drive so that the losses are minimized and the efficiency of the drive is improved. The entire drive scheme along with flux determination block has been modeled in Simulink / Matlab environment and the losses and efficiency have been estimated for 2.2 kW and 22 kW capacity three-phase inductions motors. From the results obtained, it is evident that when the machine operates with the flux value determined by the ANN, it yields an improved efficiency especially under part-load conditions.