Model predictive current control method for PMSM drives based on an improved prediction model
- PDF / 3,327,351 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 74 Downloads / 205 Views
ORIGINAL ARTICLE
Model predictive current control method for PMSM drives based on an improved prediction model Shengwen Fan1,2,3 · Chaonan Tong1 Received: 11 March 2020 / Revised: 13 July 2020 / Accepted: 23 July 2020 © The Korean Institute of Power Electronics 2020
Abstract In motor drive systems based on model predictive control, a mathematic model of the motor is used to predict the future behavior of the system. However, the parameters in the motor model may not match their real values since these parameters may vary under different operation conditions. All parameter variations result in inaccurate predictions, and influence the steady-state control performance of the whole control system. In this paper, an improved model predictive control method is presented. Firstly, when parameter mismatches exist, the sources of the current prediction error are analyzed. It is revealed that current prediction error is directly affected by a prediction model with parameter mismatches and inaccurate one-step delay compensation. In particular, the influence form one-step delay compensation is discussed in this paper. Then a reachinglaw-based sliding mode discrete observer is introduced to implement accurate one-step delay compensation and to observe all parameter variations. Finally, a predictive control method combined with sliding-mode discrete observation is presented to reduce parameter sensitivity. Simulation and experimental results show that the proposed method can increase the robustness of model predictive control systems. Keywords Parameter variation · Model predictive current control · PMSM
1 Introduction To increase the control performance of permanent magnet synchronous motors (PMSMs), a number of methods including backstepping control, robust control, optimal control, and model predictive control (MPC) have been presented in the literature [1–5]. In these methods, MPC is widely used due to its simple structure and fast dynamic response [6–8]. In addition, a lot of research results have been presented. To reduce the current and torque ripple in MPC, duty-cycle control in direct torque control has been introduced [9–12]. This means that one control period includes two different parts, where one part is used to apply for an active voltage
* Shengwen Fan [email protected] 1
University of Science and Technology Beijing, Beijing, China
2
North China University of Technology, Beijing, China
3
Collaborative Innovation Center of Key Power Energy-Saving Technologies, Beijing, China
vector and the other part is allocated for a zero vector. Recently, a generalized double-vector MPC method was proposed in [13], which selects an arbitrary vector as the second vector and adds a back EMF compensation for further current ripple reduction. In addition, as the candidate voltage vectors increase, MPC needs more calculation time to implement control algorithms, such as the application of multi-step prediction and multilevel converters. Thus, the sphere decoding algorithm [14, 15] and the binary search tree m
Data Loading...
