摘要
双馈异步风机(DFIG)是目前风电市场的主流机型。由于风机需优化不同时间尺度的多个控制目标,传统线性控制难以满足其需求。模型预测控制(MPC)因其优秀的动态性能和多目标优化能力,成为控制风机的有效方法。然而模型预测控制通过单一代价函数实现多目标优化,这导致控制目标相互耦合,使控制优先级混乱、权系数设计复杂。为此,提出一种动态级联模型预测控制策略。所提方法采用多个代价函数级联的结构,通过计算每个控制目标所有代价函数的平均值来对控制目标排序,动态调整优先级。此外,所提方法使用控制阈值调整控制前级进入后级的候选开关矢量数量。所提方法无需使用权系数即可实现对多个目标的总体最优控制。硬件在环结果证明了所提方法的有效性。
The doubly-fed induction generator(DFIG)is the mainstream model in the wind power market.Since there are multiple time-scale control objectives in the wind turbine,it is difficult for conventional linear control methods to optimize the multi-objective.Model predictive control(MPC)is an effective method for wind turbine control due to the high dynamic response performance and multi-objective optimization capacity.However,model predictive control achieves multi-objective optimization through a single cost function,which leads to the coupling of the control objectives,making it difficult to design the weight factors and determine the control priority.Therefore,a dynamic sequential model predictive control(DSMPC)strategy was proposed,which a single cost function was replaced by an optimization structure composed of multiple cascade cost functions.The control objectives were ranked according to the average value of all cost functions of each control objective,and the priority was dynamically adjusted.In addition,the control threshold was used to adjust the number of candidate switch vectors from the control pre-stage to the control post-stage.This method realized the overall optimal control of multiple objectives without using weight factors.The effectiveness of the proposed method was proved by the hardware-in-the-loop results.
作者
田家彬
杨传江
李俊达
谷加腾
张祯滨
TIAN Jiabin;YANG Chuanjiang;LI Junda;GU Jiateng;ZHANG Zhenbin(Wind Power Equipment Research Institute,CRRC Shandong Wind Power Co.,Ltd.,Jinan 250022,Shandong,China;School of Electrical Engineering,Shandong University,Jinan 250061,Shandong,China)
出处
《电气传动》
2023年第12期85-92,共8页
Electric Drive
关键词
双馈异步风机
动态级联模型预测控制
动态优先级
控制阈值
doubly-fed induction generator(DFIG)
dynamic sequential model predictive control(DSMPC)
dynamic priority
control threshold
