双馈风电机组次同步振荡自主学习控制方法设计

Design of Autonomous Learning Control Method for Sub Synchronous Oscillation of Doubly Fed Wind Turbines

我国风能资源丰富,但源荷逆向分布的空间矛盾突出,大规模远距离输电是消纳风电机组的必然方式。为提高风电机组输电能力,常采用串联补偿技术,但也会引入次同步振荡(SSO)的风险。由于同步振荡是一个典型的多因素耦合影响的结果,属于典型的少约束下最优解问题。提出基于多目标深度学习的双馈风电机组次同步振荡控制方法。深度分析双馈风电机组中产生次同步振荡的关联因素。分析双馈风电机的转差率、转子电压与定子电压之间的关系,推导出转差率作为不同频率范围下,次同步振荡控制的约束条件。设计深度学习模型,建立输入层、隐含层和输出层,输入次同步振荡特征并实施串联补偿,利用附加阻尼实现振荡控制。实验结果证明:该方法依托DSP完成的电路设计,降低了双馈风电机组次同步振荡幅值,次同步振荡控制时间最高为14ms,说明研究方法能快速并准确的实施次同步振荡抑制控制。

China has abundant wind energy resources,but the spatial contradiction of reverse distribution of source and load is prominent.Large scale long-distance transmission is an inevitable way to consume wind turbines.To improve the transmission capacity of wind turbines,series compensation technology is often used,but it also introduces t he risk of sub synchronous oscillation(SSO).Due to the fact that synchronous oscillation is a typical result of multi factor coupling,it belongs to the typical optimal solution problem under few constraints.Propose a sub synchronous oscillation control method for doubly fed wind turbines based on multi-objective deep learning.Deep analysis of the associated factors causing sub synchronous oscillations in doubly fed wind turbines.Analyze the relationship between the slip rate,rotor voltage,and stator voltage of a doubly fed wind motor,and derive the slip rate as a constraint condition for subsynchronous oscillation control in different frequency ranges.Design a deep learning model,es tablish input layer,hidden layer,and output layer,input sub synchronous oscillation characteristics,and implement series compensation,using additional damping to achieve oscillation control.The experimental results demonstrate that the circuit design based on DSP reduces the amplitude of subsynchronous oscillation in doubly fed wind turbines,with a maximum control time of 14ms.This indicates that the research method can quickly and accurately implement subsynchronous oscillation suppression control.