Abstract

Finite-State Predictive Torque Control (FS-PTC) is experimentally investigated based on different online-optimization methods by using a two-level voltage source inverter for an induction machine. The calculation time and the switching frequencies are important research points for FS-PTC industrial applications. Long-step FS-PTC methods are expected to improve the performance of the system. However, the calculation time will increase exponentially with the increase of the prediction horizon. A reduced switching frequency PTC (RSF-PTC) method by considering the reductions of the switching frequency and the calculation time is tested. Based on this algorithm, an extended prediction horizon is proposed and verified on a common test bench. A torque-band based PTC (TB-PTC) method is proposed and discussed in this paper. The TB-PTC method pre-calculates the torque error between the predicted torque and the torque reference. The optimization method focuses on the flux error and the switching frequency for switching states which constrain the torque error within the torque-band. The conventional FS-PTC method, the RSF-PTC method with one-step and two-step horizons and the TB-PTC method are developed and experimentally compared in this work. The results confirm that conventional FS-PTC, RSF-PTC and TB-PTC methods can work well in the full speed range. When the switching frequencies and the calculation effort are taken into consideration, the REF-PTC algorithm shows the better performance. However, the conventional FS-PTC method and the TB-PTC method have better current performance. (C) 2014 Elsevier Ltd. All rights reserved.