电力电子技术

传统的模型预测控制高度依赖于准确的数学模型,且在一个控制周期内只使用一个电压矢量。针对该问题,本文提出了一种LCL型并网逆变器多矢量鲁棒预测控制(MVRPC)策略。首先,为提高跟踪精度,在一个控制周期内作用3个电压矢量,这使得开关频谱集中,也便于LCL滤波器的设计;其次,为提高控制方法的参数鲁棒性,提出了一种适用于LCL型并网逆变器的鲁棒预测控制策略,该策略仅需采样信息,就可实现对系统参数的在线估计;最后,为解决随机采样噪声对参数辨识的影响,提出了一种可变的计算周期,参数辨识只发生在该计算周期中,计算周期中使系统控制误差最优的参数将被用于进行后续的控制,避免了随机误差对于系统控制性能的影响。通过实验验证了所提方法的有效性。

Traditional model predictive control highly relies on accurate mathematical models and only uses one voltage vector within a control cycle.In response to this issue,a multi vector robust predictive control(MVRPC) strategy for LCL type grid connected inverters is proposed.Firstly,to improve tracking accuracy,three voltage vectors are applied within one control cycle,which concentrates the switching frequency spectrum and facilitates the design of LCL filters.Secondly,to improve the parameter robustness of the control method,a robust predictive control strategy suitable for LCL type grid connected inverters is proposed.This strategy only requires sampling information to achieve online estimation of system parameters.Finally,to address the impact of random sampling noise on parameter identification,a variable calculation period is proposed.Parameter identification only occurs during this calculation period,and the parameters that optimize the system control error during the calculation period will be used for subsequent control,avoiding the impact of random errors on system control performance.The effectiveness of the proposed method is experimentally verified.