Two-step predictive current control of PM synchronous motor using parameter estimation

Abstract

A new hardware optimized predictive current controller is presented. As a driver for the PM synchronous motor, fully digitized controller is applied in the inverter nowadays. The fully digitized controller has lots of advantages such as easy to adopt a new or corrective control methods, high communicative, high immunity against noise. But it has disadvantages also. The major demerit of the controller is calculation delay. The system delay causes loss of system stability and different result from ideal one especially in transient state and in noisy environments. To avoid the sampling noise in analog to digital converter and safe loading the switching duty to the digital comparator, two-step delay in the controller is inevitable. In an application such as high-speed operation for multiple-pole PM synchronous motor, the sampling time of the controller is not much smaller than electrical fundamental frequency of the motor. The two-step delay in a current controller causes significant phase delay. For the high controllability of the output voltage, pulse width modulated inverter system is used in general. Inverter has to have dead-time to avoid shoot-through of the dc-link. The dead time causes phase voltage distortion of the inverter output voltage and also makes current distortion. This current distortion makes torque ripple, acoustic noise, and increasing power loss. In order to overcome the problem of the calculation delay, a two-step predictive current controller is proposed in this dissertation. The proposed control method has also demerits of the conventional predictive current control methods what are no immunity to the parameter uncertainties. To minimize the effects of the defects of the proposed method a parameter estimator is added. To apply the proposed current control method in VSI with PWM, output voltage distortion, position data retardation, and output voltage limitation should be considered. Because of the predictive characteristic of the proposed ...