Cascaded predictive flux control for a 3-L active NPC Fed IM drives without weighting factor

• Main Authors: Xiao, Dan, Akter, Md Parvez, Alam, Kazi, Dutta, Rukmi, Mekhilef, Saad, Rahman, Muhammed Fazlur
• Format: Article
• Published: IEEE-Inst Electrical Electronics Engineers Inc 2021
• Subjects: T Technology (General); TA Engineering (General). Civil engineering (General)
• Online Access: http://eprints.um.edu.my/28190/
• Institution: Universiti Malaya

Conventional model predictive control (MPC) for three-level active neutral point clamped converter (ANPC) utilizes weighting factors to achieve the optimal control of multiple control objectives in a single cost function. However, the selection of these weighting factors has an impact on the performance of the controlled objectives significantly. Tuning of weighting factors is usually tedious and lacks theoretical background. Moreover, the predictive and evaluation process for multiple objectives optimization have to be carried out by enumerating all admissible switching states within one loop. To simplify the complexity of this process and eliminate the effect of weighting factors, this paper proposes a cascaded predictive control scheme for a 3L-ANPC inverter fed induction machine (IM) drive. With the proposed approach, three separate cost functions for three control objectives, stator flux vector, neutral point voltage and device loss distribution are evaluated in a cascaded way, so that the optimization of these control objectives is performed independently. A 3L-ANPC inverter driven IM drive is studied in this paper to verify the effectiveness of the proposed algorithm. Experimental results in different operating conditions confirm that the proposed method achieves satisfactory steady-state and transient performances comparable to the conventional predictive control approach. The weighting factors used in the conventional method are eliminated and the computational effort is reduced by 68% compared to the conventional method without any sacrifice in the drive performance.