逐步回归模型在互联电网主振模式辨识中的应用

Application of Multiple Stepwise Regression in Identifying Critical Inter-Area Oscillation Mode

针对联络线频发的有功潮流区间振荡,提出一种基于逐步回归模型的在线WAMS主振模式辨识方法。首先基于Prony算法抽取区域联络线WAMS有功轨迹或仿真轨迹中的振荡模式,接着以Prony算法所辨识出的模式参数构造解释变量全集并建立起多元线性化回归模型,通过逐步回归算法以循环迭代方式从解释变量全集中逐一增加和删除解释变量,求解满足回归方程和参数显著性的解释变量子集,最终以回归模型的参数显著性结果 sig值作为振荡模式重要性排序标准,自动确定联络线功率振荡主振模式,为进一步制定相关控制措施提供支持。最后,通过测试模型算例和四川电网实际系统振荡轨迹分析计算表明:所提方法能够自动辨识受扰系统联络线振荡轨迹主振模式,并且算法对噪声信号有较强鲁棒性,验证了所提模型方法的可行性和有效性。

As to inter-area oscillation which frequently appears in transmission lines, a novel online WAMS identification method for critical oscillation mode is presented, which makes use of stepwise regression method. Firstly, Prony algorithm is utilized to obtain oscillation modes of active power flow from WAMS or simulation data. Then parameters of oscillation modes are introduced to construct explanatory variable set and build multiple regression model. Secondly, stepwise regression method is raised to insert and delete explanatory variable in an iteration process and acquire new explanatory variable set that meet significance test of the model Sig-value of the specified procedures is used to evaluate importance of the obtained oscillation mode automatically. Finally, analysis results of a test model and Sichuan power grid show that the proposed stepwise regression method is able to identify critical oscillation mode and has strong robustness in noise signal, which verifies the feasibility and effectiveness of the proposed method.