Weighting Factorless Model Predictive Thrust Control for Linear Induction Machine

Compared with various commonly used control algorithms for linear induction machine, the model predictive thrust control can achieve multiple optimum control objectives, such as high dynamic performance, low power loss, and low flux and thrust ripples, but it is difficult to find suitable weighting factors to balance them. The most commonly used method to tune the weighting factors is by enumerating numerous cases, and then evaluating and comparing each case until the best set of weighting factors is achieved, which can be a very long and tedious procedure. This paper proposes two different methods to solve this problem for minimizing the flux and thrust ripples. One method is to use the variable weighting factor related to flux ripple so that the flux ripple can be seen as a hard constraint without complex tuning process. The other is to replace the flux control term by a variable defined with the same unit as the thrust so that the weighting factor can be equal to one without tuning. These two proposed methods have been successfully applied to a test platform consisting of two 3-kW arc induction motors. The experimental results have shown smaller flux and thrust ripples, as well as tracking errors in comparison with two existing common methods, i.e., the fuzzy decision based method and model predictive flux control.