基于统计学习的模糊暂态稳定事故筛选与排序

暂态稳定事故筛选与排序的目的是针对一组预想事故集合挑选出严重事故或滤除掉无害事故,以减少待分析的预想事故数目,满足在线动态安全分析的需要.文中提出一种基于统计学习的模糊暂态稳定事故筛选与排序方法,该方法采用了反映事故严重程度的10个性能指标,并通过样本学习,在每个指标集合上具体定义了系统稳定性的模糊隶属度函数;然后综合运用这10个性能指标的稳定性模糊隶属度得到的平均稳定性模糊隶属度,对预想事故集合按严重性进行了排序;接着根据排序结果,结合所设定的稳定阈值,将无害事故过滤掉.最后,用新英格兰10机39节点网络验证了该方法的有效性.

Fuzzy Contingency Filtering and Ranking of Power System Transient Stability Based on Statistical Learning

The aim of contingency filtering and ranking is to reduce the number of contingencies needing detailed analysis and meet the requirement of online dynamic security assessment with the selection of severe contingencies or the screening of harmless contingencies. This paper proposes a method of fuzzy contingency screening and ranking of power system transient stability based on statistical learning. The method adopts ten performance indices which reflect the servererity of contingencies, and define fuzzy member functions of system stability for each performance indices through sample learning. Then it calculates the average value of fuzzy contingency filtering and ranking of power system transient stability of the ten performance indices, and ranks the predefined contingencies. With the ranking result and predefined threshold of stability, harmless contingencies are filtered out. Finally, the results of New England 10-generator 39-bus system prove the effectiveness of the proposed method.