双馈风电机组振荡模态与影响因子的关联分析

Correlation Analysis Between Oscillation Mode and Impact Factor of Doubly-fed Induction Generator

目前对风电振荡模态的研究建立在对风机系统建模的基础上,但建模时所考虑的外界因素有限,分析风电机组实测数据,对功率实测数据中的相应振荡模态与风速和并网电压等因素之间的关联关系进行研究。在对现场测得的风机机组输出功率数据分段之后,采用Prony方法分析功率中含有的振荡模态,并依据小信号分析结果对相关振荡模态进行提取;同时用K-Means算法将实测序列数据按风速和电压进行聚类;最后利用Apriori算法讨论风速/电压聚簇与振荡模态之间的关联规则,并对振荡模态进行预测。关联分析结果表明,风速/电压波动聚簇数对关联规则影响较大,在最佳聚簇数的条件下,部分风速/电压聚簇可分析出输出功率含有特定振荡分量,预测结果与实际情况吻合。

The research on wind power oscillation modes is usually based on system modeling of wind power system with limited external factors considered. The more accurate model means more complicated solution of the model. This paper would analyze the measured data of the wind generation system to mine the correlation between oscillation modes and factors(including wind speed and voltage fluctuation). The output power of wind generating machine sets was segmented first. Each segment was decomposed by the Prony algorithm and the important oscillation modes were extracted according to the results of small signal analysis. Meanwhile, the K-Means algorithm was adopted to cluster the data according to wind and voltage factor. The Apriori algorithm was finally used to discuss the association rules between wind speed/voltage clusters and oscillation modes and to predict the oscillation modes. The results of association analysis show that the wind speed/voltage clustering number has a great influence on the association rules. Under the optimal value, some clusters can analyze some specific oscillation components of the power and the prediction results are consistent with the actual situation.