占空比
控制理论(社会学)
模型预测控制
电感
整流器(神经网络)
脉冲宽度调制
功率(物理)
三相
计算机科学
电压
工程类
电子工程
控制(管理)
物理
电气工程
循环神经网络
人工智能
机器学习
随机神经网络
量子力学
人工神经网络
作者
Junpeng Ma,Wensheng Song,Shunliang Wang,Xiaoyun Feng
标识
DOI:10.1109/tpel.2017.2681938
摘要
This paper presents a new model predictive direct power control (MP-DPC) to overcome the drawbacks of model predictive control (MPC) for single phase three-level rectifiers in the railway traction drive system, including huge online calculation, poor power control precision at the low switching frequency and variable switching frequency. To do so, an exact analytical solution of instantaneous power estimation is adopted to predict active and reactive powers in next duty cycle updating interval for achieving the deadbeat control and reducing the predictive error at the low switching frequency (below 1 kHz). The optimal d -axis and q -axis components of input voltage within next duty cycle updating interval of the adopted rectifier in rotating coordinate system are directly calculated by minimizing the cost function. And the optimal drive pulses are generated by pulse width modulation stage in the proposed MP-DPC, other than evaluating cost function for each voltage vector in traditional MP-DPC. Finally, the influence of inductance mismatch on control system is analyzed, and an inductance estimation method is shown to improve the control precision. An experimental comparison with other five different DPC schemes has verified the effectiveness of the proposed MP-DPC scheme.
科研通智能强力驱动
Strongly Powered by AbleSci AI