Discrete Sinusoidal Current Control For PMSM With Step-motion Control

Step-motion control of permanent-magnet synchronous motor (PMSM) can gain betterpositioning performance. It is different from the continuous torque control of Vector Control andthe point-to-point torque control of Direct Torque Control. The torque of Step-motion control canbe regulated discretely by choosing the discrete torque vector. To make the step-motion controlmethod work, the sinusoidal current imposed on PMSM must be changed into discrete-sinusoidalcurrent. The control effect of the discrete current control will affect the performance of thestep-motion control of PMSM. So it is the key to reduce torque fluctuation and improve theprecision of positioning for PMSM.Started from the theory of step-motion control of PMSM, this paper analyzed thecharacteristics of the discrete-sinusoidal current used in step-motion control of PMSM.Discrete-sinusoidal current presents a severe nonlinearity. It keeps a certain value during onestep, but has to jump at the moment when the motor steps. It is hard to achieve such adiscrete-sinusoidal current using common current control methods.Considering the apparent current error and severe current spike when using common currentcontrol methods, the fuzzy control and neural-networks control were used to improve theperformance of incremental PI control method, and the direct current vector control wasintroduced to control the current vector as a whole.When the fuzzy control was used into incremental PI, we got fuzzy-incremental PI currentcontrol. Thus the parameters of incremental PI can be regulated when the current error changed.By doing this, the apparent current error and severe current spike could be reduced. When theneural network control was used into incremental PI, we got neural network control. Theparameters of incremental PI could be optimized. Put the advantages of fuzzy control and neuralnetworks control together, we got the fuzzy-neural-networks-incremental PI control method. Thecurrent track precision during one step could be improved and the current spike at the moment ofjump could be reduced.When using common current control, there will be some unnecessary switching and thezero voltage vectors can not be gained. These problems can be solved by using direct currentvector control. By controlling the current vector directly and regulating the time which the activevoltage vector imposed, the switching frequency and current error can be reduced. Thus we canget a better performance in high-voltage-low-current control.The methods above were verified using a custom-designed experimental setup. The control,protection, monitoring and power transition were implemented using the controller made by ourteam. The experimental results helped me to analyze the characteristics of those current controlmethods and verified the effect of those methods.