Artificial neural network-based DTC of an induction machine with on FPGA

Direct Torque Control (DTC) of Induction Machine (IM) has received increasing attention due to its high performance and low dependence on machine parameters. Recently, intelligent approaches have been proposed to improve the DTC performance, in particular the reduction in the torque and the flux ripples. In this paper, an approach for designing DTC for IMs that is based on Neural Network Control (NNC) is proposed. In this method, an artificial neural network, which can better manage the state of switches, is used instead of the switching table and two hysteresis controllers. The improvement is achieved by reducing the stator flux ripples and the torque ripples in the IM drive. The suggested architecture, which is developed using the VHDL, is designed using the modular architecture and the principles of the parallel architecture. The Direct Torque Neural Control (DTNC) simulation and experimental results are compared with those of the conventional DTC. Results of the comparison demonstrate how the DTNC reduces torque ripples and stator flux ripples. Results from simulation and experiment are used to verify how effectively the suggested control works. The ZC702 SOC, which is Xilinx Board based on Zynq FPGA, has been selected for experimental implementation.