控制理论(社会学)
模型预测控制
MATLAB语言
电压
模块化设计
集合(抽象数据类型)
功率(物理)
理论(学习稳定性)
计算机科学
工程类
电子工程
控制(管理)
电气工程
物理
人工智能
量子力学
机器学习
程序设计语言
操作系统
作者
Chang Ming Jiang,Shaohua Zhang
标识
DOI:10.1016/j.conengprac.2023.105773
摘要
Multistep model predictive control (M-MPC) can improve control accuracy but leads to an increased computational burden. To reduce the computational burden and restore power quality, an improved M-MPC method with reduced finite set for unified power quality conditioner (UPQC) based on modular multilevel converter (MMC) is presented in this paper. First, a finite set of M-MPC is formed according to the inserted sub-modules (SMs) number calculated from cost functions. Second, we investigate the relationship between the inserted SMs number and the bridge arm output voltages and find that the inserted SMs number will increase as voltages rise. As such, by judging the change rate of the output voltages, switching states that cannot be realized within the finite set can be eliminated, thus the finite set can be reduced. In addition, circulating currents and capacitance voltage control methods are designed to ensure the internal stability of MMC. Finally, the MATLAB/Simulink simulation and the real time laboratory (RT-LAB) based hardware-in-the-loop (HIL) experimental results show effectiveness of the proposed M-MPC. The results show the proposed M-MPC exhibits better dynamic characteristics and fewer control options as compared to the single-step MPC.
科研通智能强力驱动
Strongly Powered by AbleSci AI