Research On Parallel Predictive Control Of Three-level Induction Motor With Midpoint Potential Dynamic Compensation

With the wide application of high-power asynchronous motor-driven technology in industrial manufacturing,ship propulsion,mining,locomotive traction,special military industry and other important fields,three-level inverter has attracted more and more attention.Compared with the traditional two-level inverter,its single power switching device needs to bear a lower voltage and is more suitable for high-power asynchronous motor-driven technology.However,there is an inherent problem-"the drift of midpoint potential will lead to the distortion of output voltage waveform and affect the control effect of the system".As a new method of high-power asynchronous motor-driven control,three-level asynchronous motor predictive control can effectively suppress current distortion and reduce switching frequency without affecting the dynamic characteristics of the drive system.It is very suitable for high-power and low switching frequency drive applications.However,due to the large number of voltage vectors generated by three-level inverter,the amount of on-line operation is more than three times higher than that of two-level and the dynamic characteristics are seriously affected when it combines with predictive control.In view of the above difficulties,this paper studies the parallel predictive control method of three-level asynchronous motor integrating midpoint potential dynamic compensation.The main work is as follows:(1)Aiming at the problems of multiple output voltage vectors of three-level inverter,the significant increase of on-line computation of predictive control and the impact on the real-time performance of the drive system,a circular boundary limited form predictive control method for three-level asynchronous motor based on parallel computing is proposed in this paper.Firstly,a three-level current gradient prediction model with high disturbance rejection is established based on the predictive control of circular boundary limited form;Then,combined with parallel computing,the fast polling of voltage vector of three-level predictive control is realized;Finally,a simulation system with double sampling rate is constructed to analyze and verify the controllability and algorithm efficiency of the parallel predictive control method.(2)Aiming at the problem of midpoint potential drift of three-level inverter,based on the in-depth analysis of the internal mechanism of midpoint potential drift,this paper focuses on the impact of midpoint drift on the key factors of predictive control,establishes the dynamic compensation mathematical model of three-level voltage vector,and proposes a high current prediction accuracy predictive control method integrating midpoint potential dynamic compensation.At last,the improvement of current prediction accuracy is verified by simulation.(3)The operating characteristics of low switching frequency about predictive control are studied.Firstly,starting from the influence of motor inductance on current gradient,the mechanism of low switching frequency operation of motor drive system is analyzed and verified by simulation;Then,starting from the topology of three-level inverter and two-level inverter,the internal mechanism of three-level inverter with lower switching frequency characteristics is analyzed theoretically;Finally,an equivalent research method of medium voltage asynchronous motor based on motor parameter unitary processing is explored,which provides a new idea for the research of low switching frequency drive control of medium voltage asynchronous motor.(4)Aiming at the problem of short circuit of bridge arm in the same direction caused by power device switching delay in the practical application of three-level inverter,a method of switching constraint on switching state combined with predictive control algorithm is proposed in this paper,which is realized by direct triggering of hardware logic,and the actual effect of the constraint method is verified by experiments.(5)A set of three-level asynchronous motor experimental platform combining real-time processor and programmable logic device is developed,and the design of relevant hardware circuit and software is optimized.At the same time,the programmable logic device design with a method of switching constraint on switching state and the digital implementation of parallel predictive control algorithm based on master-slave processor cooperative operation are completed.(6)Finally,based on the self-developed experimental platform,this paper focuses on the comprehensive experimental research on the new predictive control method proposed in this paper,such as speed closed-loop control,current distortion suppression,low switching frequency characteristics,and comparison with the control effect of traditional methods.