Improved Model Predictive Torque Control For Permanent Magnet Synchronous Motor With Three-level Converte

Neutral Point Clamped(NPC)type three-level converters have the advantages of high output voltage quality and low voltage stress on the switching tubes.PMSM systems are widely used in aerospace,rail transportation and mining.The model predictive torque control algorithm is simple and particularly suitable for controlling complex systems with multiple objectives and variables,but it has the disadvantage of tedious weight coefficient adjustment process.In this paper,the traditional model predictive torque control algorithm is improved to address the problem of weight coefficient adjustment in the three-level converter-permanent magnet synchronous motor system.In order to eliminate the weight coefficients of the motor torque term and the magnetic chain term,the improved model predictive control algorithm converts the two into the stator voltage term by using the no-difference beat formula,in which the value function includes the stator voltage term and the converter neutral-point potential term.For the stator voltage term,the optimization strategy is equated to the distance relationship between the reference vector and the basic vector,and the basic vector corresponding to the shortest distance is the optimal vector.For the neutral-point potential term,the offset threshold is set to control the neutral-point potential partitioning.When the neutral-point potential offset amplitude is less than the threshold,the value function does not consider the neutral-point potential term,and only optimizes the motor magnetic chain and torque.When the neutral-point potential offset amplitude is larger than the threshold,the amplitude and phase of the medium and small vectors change in each control cycle,and the sector dynamic division mechanism is introduced to recalculate the medium and small vector amplitude and phase,and only the neutralpoint potential suppression of the converter is considered in the value function.At the same time,the alternative voltage vectors in the finite control set are reasonably selected according to the sector where the reference voltage vector is located to further simplify the algorithm execution time.In this paper,an improved multi-step predictive torque control algorithm is proposed based on the traditional finite set multi-step predictive control.In this paper,a three-level converter-permanent magnet synchronous motor experimental system is built using a d SPACE rapid prototyping simulator as the controller,and the conventional model predictive torque control algorithm and the improved model predictive torque control algorithm are experimentally verified.The experimental results show that the improved model predictive control algorithm achieves fast balancing of neutral-point potential fluctuations when neutral-point potential offset amplitude is large,completely eliminates the rectification of the weight coefficients and outperforms the conventional control algorithm in terms of steady-state and dynamic performance.