TLO-based Neural Discrete Predictive Approach for DTFC Induction Motor Drive

This paper represents a neural predictive (NP) approach optimized with the teaching learning-based optimization (NP-TLO) technique employed to a direct torque and flux controlled (DTFC) induction motor (IM) drive to extensively reduce the torque and flux ripple along with the enhanced dynamic as well as steady-state performance. In order to have a fair comparison and superior performance, a comparative evaluation is also illustrated here for NP-TLO with that of the conventional model predictive (MP) approach optimized with TLO (MP-TLO). NP approach has been proposed as a superior control technique as the MP approach has the disadvantage of the large computational burden imposed due to its long iteration process. Further, the TLO approach has been applied proficiently with the neural predictive approach to fix the solution for nonlinear optimization problems keeping in view the system constraints like output and states. The DTFC model incorporated with both predictive approaches is designed and developed with the MATLAB software and with a real-time validation as well with a 5 HP IM drive to show its potentiality in real-world industry applications. Moreover, the computational involvedness in NN-TLO is declined compared to MP-TLO and thus, the produced optimized control signal of NN-TLO contributes better performance and robust operation by retaining all characteristics of the DTFC drive.