In permanent-magnet synchronous machine (PMSM) applications, traditional deadbeat predictive current control (DPCC) utilizes the PMSM model to evaluate the expected voltage vector and applies it to the inverter through space vector pulse width modulation (SVPWM). Once the expected voltage vector is inaccurate, the torque ripple and speed fluctuation are amplified. There are two main factors that cause the inaccurate voltage vector, namely model parameter mismatch, and current measurement error. To enhance the robustness of DPCC, first, this paper proposes an accurate PMSM voltage model with nonperiodic and periodic disturbance models. Second, this paper proposes a novel current and disturbance observer (NCDO) which is able to predict future stator currents and disturbances caused by model parameter mismatch and current measurement error simultaneously. Finally, the scheme of the proposed DPCC with NCDO is presented to enhance the robustness. This paper presents a comparative study of two types of algorithms, namely traditional DPCC and the proposed DPCC with NCDO. The theoretical verification, simulation results, and experimental results are demonstrated to verify the effectiveness of the proposed DPCC with NCDO.

Introduction

Recently, permanent-magnet synchronous machines (PMSMs) have been widely used in the modern applications because they have a range of benefits such as high efficiency, high torque density, and excellent control precision. To achieve high steady-state and dynamic performance, some control strategies have been applied in the drive system of PMSMs, such as classical proportional–integral (PI) control [1], hysteresis control, and predictive control. Hysteresis control [2] has good robustness, fast current responses, and simple computation, but there are large current ripples in the control system. Compared with hysteresis control, PI control has some benefits such as small current ripples and fixed switching frequency, which is popular in practical applications. However, the PI parameters need to be tuned, which is time-consuming. Recently, predictive control is applied in PMSM drives within ten years. Predictive control has some advantages, such as excellent performance in the transient state [3]. It can be categorized into two types of predictive control normally, namely finite control set model predictive control (FCS-MPC) [4] and deadbeat predictive control (DPCC). FCS-MPC applies finite voltage vectors based on characteristic of inverters to predict next instant motor stator currents by minimizing cost functions.