Research On Three-level Drive Control System Based On Model Predictive Control

With the development of power electronics technology,Model Predictive Control(MPC)has attracted wide attention in the field of AC drive because of its good control performance.Its control idea is to transform the continuous control variable optimization problem into the discrete basic voltage vector optimization process of the inverters.The control method is flexible,simple and easy to deal with the non-linear constraints.It has strong advantages for the control object with non-linear and strong coupling mathematical model,such as the asynchronous motor.Neutral-point-clamped(NPC)three-level inverters have been widely used for their advantages of lower voltage stress and wider range of vector selection.In this paper,MPC strategy is applied to three-level AC drive system,and the direct torque/current control strategy based on model prediction is studied.The main contents and contributions are as follows:1.An improved open-loop state observer based on stationary coordinate system and rotating coordinate system is proposed to solve the problems of unstable first-order discrete model,large discrete error and large computational load of asynchronous motor in high-speed domain.An improved closed-loop observer model with rotor flux and stator flux as state variables is constructed by adding the error between estimated and actual stator current to form a feedback system,and the feedback matrix is designed to allocate the poles of the system.The stable observation of flux and speed of induction motor in the full speed range is realized.2.Based on the traditional direct torque control idea,a three-level model predictive direct torque control scheme for induction motors is proposed.The relative deviation of torque and stator flux is analyzed and constructed as the value function.The optimal switching vector is obtained by real-time calculation.Compared with the traditional direct torque control scheme,the torque ripple is effectively reduced.Aiming at the problem of neutral-point voltage fluctuation in NPC three-level inverters,the neutral-point voltage balance control of NPC three-level inverters is realized by choosing appropriate small vectors.Further the low switching frequency model predicts that the direct torque control scheme can reduce the switching frequency of the inverters while maintaining the current and torque performance of the system.3.Current performance is an important performance index of AC motor drive system.On the basis of vector control scheme for induction motor,a model predictive direct current control scheme for induction motor is proposed,which uses model predictive controller instead of traditional current regulator in inner current loop.In order to suppress the fluctuation of neutral-point voltage,the objective of optimizing neutral-point voltage of three-level inverters is considered in the value function,and the neutral-point voltage is controlled within the expected range.On this basis,the influence of control delay on the control performance of the system is analyzed and studied,and the control delay is compensated to reduce the current ripple and torque ripple caused by the control delay.4.An improved two-vector model predictive torque control method for induction motor is proposed.According to the given values of the torque and stator flux,the expected voltage vector can be calculated.This scheme only needs to optimize and evaluate the limited basic voltage vectors in the sector where the expected voltage vector is located.In order to avoid the tedious design of weight coefficient,the control of electromagnetic torque and stator flux amplitude is equivalent to the control of basic voltage vector.At the same time,in order to reduce the motor torque and flux ripple,a model predictive torque control strategy based on dual-vector duty ratio control is established,and a non-zero vector + zero vector action scheme is adopted to improve the steady-state performance of the system.5.A set of three-level inverters experimental platform based on ds PIC30F6010 A MCU is designed,including the main circuit,voltage and current detection and corresponding protection circuit.It also includes the allocation of CPU resources,various control software design schemes and the realization of some control algorithms.The experimental verification of model predictive direct torque control and model predictive direct current control is completed on the experimental platform.The experimental results show that the proposed model predictive control strategy is correct and effective in three-level AC drive system.