Model Predictive Current Control Of Permanent Magnet Synchronous Traction Motor For Rail Transit

As the most widely used traction motor except induction motor,permanent magnet synchronous motor(PMSM)has the advantages of high efficiency and high power density.The research of its control technology is of great significance to the development of rail transit,and has been concerned for a long time.Vector control and direct torque control are the state-of-art technologies.Vector control is usually based on proportional-integral controller and pulse width modulator,which has poor dynamic response at low switching frequency and complex parameter setting and modulation.Direct torque control is usually based on hysteresis controller,which has fast dynamic response but serious current distortion.In order to overcome their shortcomings,there are a lot of improved methods to make the above two architectures more and more complex,but there are still many limitations.Finite set model predictive control(FS-MPC)has many advantages,such as simple and intuitive control structure,fast dynamic response.Besides,it’s easy to deal with multi variables and multi constraints.It is expected to develop into a new scheme of traction drive control system.However,the relevant theory is not mature,which needs further research and demonstration to be popularized in the field of rail transit.In this dissertation,FS-MPC is applied to PMSM traction system.Firstly,the framework and performance evaluation system of model predictive current control(MPCC)for PMSM traction system are established.Then,the topics of switching frequency control,common mode voltage suppression,dc voltage utilization improving and control parameter design are studied.One of the inherent characteristics of FS-MPC is that the switching frequency is not fixed.However,low switching frequency is a hard requirement for traction inverters.The MPCC with penalty(MPCC-P)method incorporates the characterization variable corresponding to each target as a penalty function into the cost function,and adjusts the relative importance of each target through the penalty coefficient.However,due to the different dimension and value range of each characterizing variable,the penalty coefficient is difficult to determine.In order to avoid the difficulty of dimensionless parameter design,this dissertation proposes a multi-objective control method called MPCC with bounds(MPCC-B),and compares it with the conventional MPCC-P method.Compared with MPCC-P,MPCC-B can transform the problem of switching frequency control into the limitation of current ripple,so that the control parameters have a clear physical meaning.What’s more,MPCC-B can keep the current ripple constant when the motor parameter error exists.The bearing erosion of traction motor will not only affect the safety of train operation,but also cause huge economic losses.In this dissertation,the mechanism of bearing electrical erosion is modeled and analyzed.It is pointed out that the common mode voltage of inverter is the main cause of bearing electrical erosion.In order to achieve common mode voltage suppression,the MPCC with multi bounds(MPCC-MB)method is proposed based on MPCC-B.On the one hand,it shows the scalability of MPCC-B,on the other hand,it further shows the advantages of MPCC-B compared with MPCC-P in control parameter design.In MPCC-MB,two key control parameters corresponding to switching frequency and common mode voltage have the same dimension and strong correlation with each other,so blindness in parameter design can be avoided.In order to improve the utilization of dc voltage at high speed,it is necessary to make the inverter work in square wave mode.The double current closed-loop mechanism is prone to voltage saturation at high modulation ratios,resulting in current tracking failure and difficulty entering square wave mode.This dissertation proposes a method of voltage vector clamping,which realizes the continuous control from over modulation area to square wave mode and improves the utilization ratio of dc voltage to the maximum extent.The essence of voltage vector clamping is to gradually reduce the action space of the closed-loop regulation controller until it enters the square wave mode.This method is simple and easy to use,neither need to add additional regulators,nor need to switch the control frame.In order to analyze the current trajectory under different modulation ratio,the idea of establishing x-y coordinate system based on ideal voltage vector is proposed creatively.Through the analysis of the current trajectory,it is concluded that the x component of the ripple current is significantly larger than the y component at high modulation ratio.Based on this,the control strategy of using rectangular boundary instead of circular boundary in x-y coordinate system is proposed,which greatly optimizes the current distortion and controls switching frequency in the over modulation region.When performing multi-objective collaborative control,setting the weight of each object is an important issue.Compared with MPCC-P,although MPCC-MB can preliminarily set the current boundary through analytical methods,so as not to fall into the dilemma of dimensionless parameter settings,but the optimal current boundary setting law still needs further study.The nonlinear mapping function of artificial neural network is very suitable for fitting the relationship between penalty coefficient or current limit boundary and corresponding performance index.Therefore,this dissertation proposes an MPCC parameter design process based on artificial neural network,and discusses the general rules of parameter design under different working conditions with examples.In a word,this dissertation establishes a full speed domain model predictive control framework which can be used in permanent magnet traction system,and meets the needs of high-power applications(low switching frequency and high dc voltage utilization),and has the ability to expand other auxiliary functions(common mode voltage suppression).The control parameter design method based on artificial neural network provides convenience for further application.A 4.4 k W PMSM experimental platform was established,and the theories and methods in this dissertation were verified.Figures: 84.Tables: 13.References: 126.