A High-accuracy Torque Control Strategy of PMSM for Electric Vehicles

The actual torque of PMSM may discrepant from the expected torque due to the non-linear characteristics and inconsistency during mass production of PMSM. Meanwhile, achieving high-precision torque control is vital to the safety performance of electric vehicles. A proposed multi-closed-loop algorithm is introduced in this paper. In order to reduce cost and save space, no torque sensors are used. A look-up table model is established by analyzing the data of the calibration experiments, which realizes the high-precision observation of the torque. Under the control of the voltage and current double closed-loop, the proposed power closed-loop further improves the accuracy. The stability of the whole system is analyzed by small signal model, via which, design instructions of proper PI controller parameters are given out. The strategy is verified through the simulation and experimental results.

I. Introduction

PMSM has good performance in efficiency, output torque and power density. Especially, due to the ability to expand the speed through the weakening of the magnetic field, it is widely used in the drive system of electric vehicles. The current vector control method of PMSM is examined to expand the operating limits associated with inverter capacity in [1]. The method is optimum in the sense of deriving maximum output torque within the voltage and current constraints, which lays the foundation of the field weakening control theory. In [2], a convenient method is proposed to characterize the machine torque look up table, which is utilized in open-loop torque control algorithms. However, open-loop control is difficult to suppress external disturbances. The error of position and electrical signals will affect the actual control accuracy, even causes instability to the system. In [3], Zhang presents an analysis of inner current and outer voltage double closed loops control strategy and the influence of PI parameters on system stability through establishing small signal model. But it is difficult to avoid the effects of temperature changes as well as the problem of inconsistent parameters during the mass production process. It is necessary to observe the actual torque to achieve closed-loop control. In [4], a machine learning based method is proposed for rotor torque estimation. It consumes too much computing resources. In [5], a novel online estimation method is provided for the parameter estimation of the flux linkage and the q-axis inductance and estimate the torque by using the online estimated parameters to. In [6], Huang estimates the actual torque by calculating the mechanical power, which relies on the complex model of motor losses. The actual accuracy is difficult to be guaranteed. To solve these problems, this paper proposes a novel torque estimation method that can avoid complex calculations of loss model and effectively suppresses the influence of temperature and parameter perturbation on accuracy. Then a multi-closed-loop control strategy is proposed to realize the adjustment of the preset current vector to ensure the control accuracy of torque. This paper is organized as follows: section II introduces factors that affect the accuracy of torque estimation and proposes torque estimation method and multiple closed-loop control strategy. Section III presents the analysis of the system stability based on small signal model to design the PI parameters. At last, the effectiveness of the estimation method and control strategy is verified by simulation and experiments.