动态EEAC的自适应分段映射

Dynamic EEAC with Adaptive Subsection Mapping

在完整的扩展等面积准则(EEAC)算法流程中,动态EEAC(DEEAC)是在解析的静态EEAC(SEEAC)与精确的集成EEAC(IEEAC)之间的桥梁环节。为了更好地综合IEEAC的精确性与SEEAC的快速性,文中研究了DEEAC的分段数对完整EEAC算法性能的影响规律。并在此基础上,融合因果分析及机器学习方法,在DEEAC中采用自适应分段的泰勒级数展开及降维映射,以替代在故障中及故障后的2分段展开方式。根据DEEAC与SEEAC所得稳定裕度之差,定义算例的时变度指标,并据此自动调整DEEAC环节的分段数。对36套中国区域和省级电力系统数据的3 272个算例,包括对称及不对称故障在内的仿真结果表明:自适应分段的DEEAC比固定分段的DEEAC的计算量平均增加4%,但显著提升了强时变算例的分析精度,减小了稳定分类器的降效性误判率。

In the full process of extended equal area criterion（EEAC）algorithm,dynamic EEAC（DEEAC）is the bridge between analytical static EEAC（SEEAC）and accurate integrating EEAC（IEEAC）.For better coordinating the accuracy of IEEAC and the speedability of SEEAC,the influence of the subsection number of DEEAC on the performance of full EEAC is studied in detail.Based on theoretical analysis as well as the machine learning method,an adaptive segmental Taylor series expansion and dimensional reduction mapping are adopted to replace the 2-section expansions for post-fault in DEEAC.The difference between the stability margin given by the original DEEAC and that given by SEEAC is defined as the time-varying index.According to the index,the number of subsections in DEEAC can be adjusted automatically.Its robust performance is verified by 36 sets of data from 9 actual Chinese regional and provincial power systems,where both symmetric and asymmetric faults are considered,totally 3 272 cases.Compared with the average computational burden of the original DEEAC,the computational burden of the proposed adaptive DEEAC increases about 4%,however,the analysis precision for strong timevarying cases,as well as the effectiveness of the classifier for stable cases are significantly improved.