中文摘要：

目前，利用数据挖掘方法进行电力系统暂态稳定评估的研究，对结果中不稳定样本被误判为稳定样本的情况重视不足，不符合电网运行对安全性的要求。针对该问题，文中提出了安全域概念下基于多支持向量机综合的电力系统暂态稳定评估方法。该方法首先利用网格法对支持向量机进行参数寻优，然后选取分类准确率高的若干组支持向量机参数，在这些参数下训练支持向量机，最后对训练得到的支持向量机进行综合，实现电力系统暂态稳定评估。对仿真系统的分析表明，文中提出的方法能够充分利用不同参数的支持向量机提供的有用信息，大量减少“误判稳定”样本的个数，可以对应用数据挖掘理论进行电力系统暂态稳定评估的实际应用提供有益的参考。

英文摘要：

The existing data mining methods for transient stability assessment（TSA） lack sufficient considerations for these situations that we mistake unstable samples for stable ones. In response to this deficiency, this paper proposed a multi-support vector machine（SVM） power system transient stability assessment method under the concept of security region. Firstly, with the proposed method we searched optimal parameters of SVM using a grid-search method, then selected several groups of SVM parameters that could get a high classification accuracy to train SVMs, and finally integrated these trained SVMs to assess the transient stability of power systems. The analysis of a simulation system shows that the proposed method can get a significant reduction in ＂misclassification＂ samples, which provides useful information for using data mining theory for TSA in real power systems.