中文摘要：

针对铁磁谐振主动控制策略研究中，由于电网结构信息和参数信息不完整所导致的谐振电网状态空间模型无法实时确定的问题，设计了仅利用电网在线监测系统实时监测的电压信息辨识谐振系统模型。首先采用相空间重构技术，将监测所得的1维电压时间序列重构为多维向量；再根据最小支持向量机（SVM）原理设计了基于重构向量辨识谐振系统的最小支持向量机模型法；最后利用所建模型的预测功能进行谐振反馈控制器的设计。仿真结果表明，利用改进C—C算法确定嵌入维后，可采用最小延迟时间进行序列重构并建模来获得最小的电压预测误差；通过电压时间序列的稳定性分析确定控制器参数后，可将电压历史值和电压预测值代入所述的控制算法来确定控制量，从而实现铁磁谐振的可靠控制，将谐振过电压快速同步至无谐振时的目标轨迹。

英文摘要：

Because information of power network structure and system parameters in the research of the ferromagnetic resonance active control strategy is generally incomplete, the ferroresonance state space model cannot be confirmed in real time. Hence, we designed a model of voltage-information-identification resonance system to realize real-time monitoring through only online monitoring of power grid. Firstly, the one-dimensional voltage time series obtained by monitoring were reconstructed into multi-dimensional vectors. Then, using the support vector machine （SVM） theory, we designed a minimum SVM model based on the reconstructed vector-identification resonance system. At last, using the prediction function of the established model, we designed a resonance feedback controller. Further simulation showed that, after deciding the embedding dimension through the improved C-C algorithm, we could obtain the minimum prediction error of voltage when the minimum delay time was used to reconstruct series and build the model. In addition, after the controller parameters being determined by the stability analysis of voltage time series, we can calculate controlling quantities using historical values and predicted values of voltage with the described control algorithm, and control of ferromagnetic resonance can be realized reliably, synchronizing ferroresonance overvoltage to non-ferroresonance target trajectory rapidly.