Research On Three-level Active Neutral Point Clamped Converter Based On Model Predictive Control Strategy

The three-level active-neutral-point-clamped converter(ANPC)has the advantages of low harmonic distortion of three-phase output voltage,low voltage withstand value of switching devices and high power density,which reduces the cost of converter and the volume of equipment.At present,it is widely used in medium and high voltage high power equipment.In this paper,the traditional model predictive control algorithm is applied to three-level ANPC,and the rolling optimization of each switch cycle needs to calculate all voltage vectors,which leads to the problem of large calculation of predictive model and value function.The main contents of this paper are as follows:1.The working principle of the three-level ANPC converter is briefly introduced.The mathematical model of the three-level ANPC converter and the space voltage vector is constructed,and the model is discretized.2.The multi-level ANPC converter experiment platform based on DSP + CPLD control unit of the project team is described.In this paper,two improved model predictive control strategies are verified on this experiment platform.3.The traditional model prediction algorithm of three-level ANPC converter computes 27 voltage vectors every time it scrolls to choose an optimal solution,so it needs a lot of resources of the controller and a long execution time of the program.To solve this problem,an improved model predictive algorithm control is proposed.In this paper,a virtual space vector algorithm is proposed to reduce the 27 voltage vectors in the finite control set to 21 voltage vectors,so as to reduce the number of vectors in each rolling optimization,and finally improve the calculation speed and reduce the consumption of controller resources and improve the efficiency.The algorithm has been verified by simulation and experiment.4.On the basis of the above,another faster model predictive control strategy is proposed.This strategy is divided into two steps of prediction.In the first step,six positive and small vectors are substituted according to the value function,and the one with the smallest value function is selected;in the second step,the positive and small vectors with the smallest value function are calculated according to the first step,and the value function of the quadrilateral vector centered on the positive and small vectors is calculated,and finally selected Select a minimum value as the output vector,which greatly reduces the amount of calculation and improves the efficiency of calculation.The algorithm has been verified by simulation and experiment.